tcpdump mailing list archives

Re: Request for DECT LINKTYPE


From: Matthias Wenzel <tcpdump () mazzoo de>
Date: Tue, 23 Dec 2008 11:16:38 +0100

Guy Harris wrote:

At some point I'll be adding to libpcap APIs for getting lists of
supported radio channels and setting the current radio channel, for the
benefit of 802.11.  Those APIs could probably be made usable for other
network types.

setting a RFPI (like a MAC address),

So what effect does setting the RFPI have?  Does it act as, in effect, a
capture filter, so that only traffic to or from that RFPI is captured?

No, it's not like a capture filter, its synchronizing the HW to a TDMA
timeslot, and just follow this device's conversation. Partly this is
done very low in the firmware, some parts in the kernel driver.

switching
between async scan mode and synchronized packet capure?

Is that scanning by changing the channel?  If so, would that be done by
the hardware/firmware on your device, or could it be done by the
software changing the channel and looking for a signal?

There are two all different modes of operation:

Asynchrounously cou can scan for basestations and handsets. You walk
through the channels to get all of them, but you don't gain much
information other than the existance of a device, and its ID (RFPI). In
this mode you likely also loose packets (HW related).

Synchonous operation is the "normal" mode, where you lock to a certain
basestation (RFPI), and can follow all conversations it does. In this
mode we follow timeslots and channels by reading the basestation's
control channels. This is handled within our driver and firmware.

For an API layer I guess we can abstract it to
- set channel (in async mode)
- set RFPI (in sync mode), maybe that can correlate to a "set ESSID" in
802.11

this should suffice for a basic sniffer, but there's more I can think
of, e.g. "get a list of RSSI values, all timeslots, all channels", as in
DECT the handset decides which is the least noisy channel, and chooses
it for calls.

matthias
-
This is the tcpdump-workers list.
Visit https://cod.sandelman.ca/ to unsubscribe.


Current thread: