kmem_map too small panics with Soekris/Atheros access point

[Boris Kochergin]

Stef Walter wrote:
> Boris Kochergin wrote:
>   
>>> _______________________________________________
>>> freebsd-net at freebsd.org mailing list
>>> http://lists.freebsd.org/mailman/listinfo/freebsd-net
>>> To unsubscribe, send any mail to "freebsd-net-unsubscribe at freebsd.org"
>>>       
>> More information: I upgraded it to a 7.2-STABLE with August 20th sources
>> and increased kern.ipc.nmbclusters to 65536 in hopes of getting the
>> panic less often. I managed to catch it in a state where there were a
>> bunch of "ath0: ath_rx_proc: no mbuf!" messages in the kernel buffer.
>> One line of "vmstat -m" stood out as the leak:
>>
>> 80211node 12677 101401K       -   120901  16,512
>>
>> "ifconfig ath0 down" freed the memory. Is there any other useful
>> information I can provide if I catch it again? Someone suggested the
>> output of "vmstat -z" off list, and I will have that next time.
>>     
>
> This might be a long shot:
>
> Back when I deployed some similar devices on FreeBSD 6.0 there was a
> leak when using oddball ath_rate driver. I believe it went away when I
> used ath_rate_sample. But according to 'man 4 ath' nothing else besides
> ath_rate_sample exists anymore.
>
> Cheers,
>
> Stef
>
> _______________________________________________
> freebsd-net at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-net
> To unsubscribe, send any mail to "freebsd-net-unsubscribe at freebsd.org"
>   
I, too, recall the days when you had multiple rate-control algorithms to 
choose from, but that doesn't appear to be the case anymore. As a 
workaround, I wrote a little script that checks if that part of the 
driver is using more than 200 KiB of memory and run it every minute via 
cron. It seems to be doing its job so far:

#!/bin/sh

memory=`vmstat -m | grep "80211node " | tr -s ' ' | cut -f4 -d' ' | sed 
's/[^0-9]//g'`

if [ $memory -gt 200 ]; then
  ifconfig ath0 down && ifconfig ath0 up
fi

-Boris