Subject: Re: ath0 timeout clues?
To: Steven M. Bellovin <smb@cs.columbia.edu>
From: Tobias Nygren <tnn@NetBSD.org>
List: current-users
Date: 08/18/2007 10:27:47
On Fri, 17 Aug 2007 21:32:48 -0400
"Steven M. Bellovin" <smb@cs.columbia.edu> wrote:

> For the last several weeks, I've had trouble at home with my ath0
> interface.  (My house network uses WEP because until very recently, one
> of the laptops ran 3.1, which did not have WPA.)  Specifically, I saw
> constant
> 
> 	ath0: device timeout (txq 1)
> 
> messages and constant hangs of the interface -- I had to ifconfig it
> down/up to restore functionality.  I especially had trouble with longer
> tranmissions, though I don't know if it was the number of packets or
> the packet size that did it.
> 
> Tonight, though, while going through the same problems, I fired up mtr,
> just to give me something that would easily monitor the first home.
> 
> The problems vanished.  I'm not seeing any timeouts, I'm not seeing any
> disconnects, and I can run ttcp at >16M bits/sec to other machines in
> the house.  (wlanctl indicates a line speed of 24M bps just now, so
> that's reasonable.)
> 
> When I killed mtr, the problems -- the timeouts and the interface wedge
> -- returned.

I am experiencing the same problem. I use -current, but with dyoung@'s
per-queue tx timeout change backed out. I still have the IFF_OACTIVE
change and that doesn't seem to cause and problems. While I still do get
occasional timeouts, they go away completely if I keep a ping process
running on the server, constantly attempting to ping the laptop.

-T