Subject: Re: ath seems still buggy
To: David Young <email@example.com>
From: Vincent <firstname.lastname@example.org>
Date: 10/20/2005 18:10:45
David Young a écrit :
> Incidentally, me and some of my clients see a periodic "ath0: hardware
> error, resetting" on a bunch of Soekris net4526s running in ad hoc mode.
> Coincidentally (?), but the period is about 62 seconds, i.e., two
> re-calibration intervals.
> Sometimes we see a continuous stream of "ath0: hardware error" until
> the watchdog timer expires.
Well, I could say that with the ath_calinterval in ath.c set to 300,
behaviour is almost nominal. I have worked all the afternoon without
experiencing a single flaw, whereas with the original value of 30, the
link would drop usually half to two minutes after being brought up.
I was wondering if all the mess wasn't caused by a "collision" between
an incoming TCP packet and a recalibration operation, something which is
much more likely to occur if the calibration period is small and/or the