[CUWiN-Dev] ath0 messages
Bill Comisky
bcomisky at pobox.com
Fri Aug 19 19:27:33 CDT 2005
On Thu, 18 Aug 2005, David Young wrote:
> On Thu, Aug 18, 2005 at 01:39:53PM -0500, Paul Smith wrote:
>> Bill will have a more detailed explanation of how we encountered this,
>> but I wanted to get it posted quickly.
>>
>> This is from a build against the 0.5.8 branch. Bill will have the
>> specific rev.
>>
>> These messages were only seen on the console and in dmesg(8), not when
>> ssh'd in.
>>
>> Below was during startup and then similar messages subsequently appeared
>> once per minute or so. Later, they appeared constantly, every few
>> seconds. We can't tell if they are debug messages or error messages but
>> they seemed to prevent the router from starting up normally.
>
> Those are debug messages. I expect them to appear at startup, but very
> rarely thereafter. I patch /etc/sysctl.conf to turn those messages
> on. See boot-image/patches/patch-sysctl. You can turn them off with
> sysctl -w hw.ath0.debug=0x00000000, or else by changing the patch file,
> patch-sysctl.
>
> It is too bad that we see "ath0: hardware error; resetting" at all.
> I have been looking for the underlying problem, and not finding it.
Sometimes I see the behaviour you describe, a couple of the dumps at
startup and then one every so often. However, I've seen several nodes
dump those messages continuously from startup. These nodes don't join the
mesh, and there's too much console noise to investigate from the serial
console. I've seen at least one case where a reboot switched it
from the continuous dump variety to the occasional dump variety.
I haven't tried switching off the debugging in sysctl.conf yet, to see if
that makes the node responsive after all reboots. I had a few nodes up
for almost two weeks prior to switching to your snapshot BSD sources for
0.5.8.. and I *think* those nodes had the hw.ath0.debug setting included.
I hadn't seen the debug messages, but then again I hadn't been logging in
via the serial console, just ssh.
I'll try and revert to old builds to see if I can find one where I can't
reproduce this problem, but just wanted to give you a heads up that
something whack is happening.
Bill
--
Bill Comisky
bcomisky at pobox.com
More information about the CU-Wireless-Dev
mailing list