Re: Q: 802.11 vap monitor mode

On Thu, Jul 03, 2008 at 08:06:31AM -0700, Sam Leffler wrote:
> Weongyo Jeong wrote:
> >During testing a new driver, I got a problem to test a monitor mode at
> >802.11 vap I'm using.
> >
> ># uname -a
> >FreeBSD kkk 8.0-CURRENT FreeBSD 8.0-CURRENT #6: Wed May 14 11:27:37 KST
> >2008     root_at_kkk:/usr/obj/usr/src/sys/GENERIC  i386
> ># ifconfig wlan0 create wlandev upgt0 wlanmode monitor
> ># ifconfig wlan0 up
> ># ifconfig
> >[...]
> >upgt0:
> >flags=108943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST,NEEDSGIANT>
> >metric 0 mtu 2290
> >        ether 00:0f:b5:dd:6f:36
> >        media: IEEE 802.11 Wireless Ethernet autoselect mode 11g
> ><monitor>
> >        status: running
> >wlan0:
> >flags=28943<UP,BROADCAST,RUNNING,PROMISC,SIMPLEX,MULTICAST,PPROMISC>
> >metric 0 mtu 1500
> >        ether 00:0f:b5:dd:6f:36
> >        inet6 fe80::20f:b5ff:fedd:6f36%wlan0 prefixlen 64 scopeid 0xd 
> >        media: IEEE 802.11 Wireless Ethernet autoselect mode 11g
> ><monitor>
> >        status: running
> >        ssid "" channel 1 (2412 Mhz 11g) bssid 00:0f:b5:dd:6f:36
> >        country US authmode OPEN privacy OFF txpower 0 scanvalid 60
> >        protmode CTS
> >
> >As seeing, it looks work but the problem is happend when I ran stumbler
> >(src/tools/tools/net80211/stumbler) program.
> >
> >The error message is "ioctl(BIOCSDLT): Invalid argument" that it's
> >emitted to screen when I execute the following command:
> ># ./stumbler wlan0
> >
> >Are there something I missed?
> >  
> What DLT is stumbler using?  If it's requesting IEEE802_11_RADIO then 

DLT was DLT_IEEE802_11_RADIO so it looks the stumbler code is broken as
Paul B. Mahol mentioned.

> this will fail as that is not available for the vap, only for the 
> underlying device.  This is an open issue that I'm looking for help 
> with--previously an 802.11 device supported 3 DLT's: 802.3, 802.11, and 
> radiotap but now it only supports the first 2.  The net80211 layer does 
> not have all the data required to dispatch radiotap headers (only the 
> driver does) so we need to push the information up to net80211 for each 
> rx'd packet (possibly expensive) or provide some net80211-level support 
> for dispatching radiotap data through vap's.

regards,
Weongyo Jeong