Kismet Wireless

Kismet Forums

 

Posted by:dragorn
Subject:Multiple kismet_server instances hard locks machine
Date:01:01:48 29/03/2009

> My setup:
> Ubuntu 8.04
> Madwifi-ng 0.9.4 (patched to fix eventual channel change failure)
> Two AR5211 PCI cards
>
> We need to tag data depending on what interface it arrives in on, so currently we're running one instance of Kismet per interface on the box. Unfortunately, this usually incites a complete freeze of the machine within 18 hours or so. Is this a known limitation of Kismet or Madwifi? Is there anything we can do to mitigate it? From a layperson's POV, it looks like yet another Madwifi race condition.

This is madwifi, sorry - anything that locks the system up is kernel layer. User-space behavior may exacerbate it, but the failure is in the driver :(

>
> I'm aware the setup is suboptimal, especially considering newcore's support for multiple interfaces, but I don't know if it reports this data yet.

So newcore tracks networks by what interface saw it, and it knows at the packet layer what interface it came from, but the PPI header doesn't have a capture card field.

I think the AVS header does - if you felt like writing some patches, check out dumpfile_pcap.cc and expand the pcapdumpformat line to understand 'avs'. You can pull a pointer to the source that saw it via the packet record and PACK_COMP_KISCAPSRC, which you could use to make an AVS header w/ capturing source data.

That'd actually be a really useful patch if you felt like looking at it,

-m


Reply to this message