Promiscuous strangeness with Prism II card + orinoco-0.11 driver

[Jim Carter]

On Fri, 26 Apr 2002, Miklos Szeredi wrote:
> When the card is put into promiscuous mode (e.g. by starting tcpdump)
> the receiving throughput drops from 4.5Mb/s to about 300kb/s.
>
> Using a wireless sniffer I've managed to uncover the cause of this:
> the card in not sending ACK's in reply to data frames, even though it
> does receive them.  This will cause the AP to retransmit the data
> frames many times, thus decreasing the throughput.
>
> Has anyone experienced similar problems?

Oh, shit!

On a Linksys WPC11 (Intersil firmware 1.03), I routinely set spy mode, and
I noticed a message in syslog that it was entering promiscuous mode.  I
have had unending trouble with low data rates, even lower than yours, in an
Ad-Hoc network where the partner is a Dell Truemobile 1150 (Lucent-Agere
firmware 6.16).  The Dell card complains of Tx excessive retries on every
packet.  Also, the Linksys card can't talk at all (under Linux) to various
Cisco-Aironet AP's at work.  Both cards are fine under Windoze.

With spy mode set on the Dell card, it does not say anything about
promiscuous mode, and its partner (Linksys) does not complain about Tx
retries.  It can talk to the AP's.  As I understand it, the Lucent-Agere
firmware can collect statistics (signal strength) on a limited number of
MAC addresses without being promiscuous.

I think I've been totally faking myself out.

On your question, a packet received in promiscuous mode should not be
acked, because presumably it was intended for a different partner, not for
the sniffer, and the true partner should ack it.  However, when the card is
in promiscuous mode and a packet is addressed to its own MAC address, the
card should behave normally, i.e. ack the packet.  It looks like the
Lucent-Agere firmware does that, and the Intersil firmware does not.

BUG, BUG, BUG!  Now who do we report this to, who will listen?

Confirmation of all this will have to wait until I go home, since I don't
have the Linksys card with me at work, but be assured that I'll let you all
know what happens when I get rid of the spy mode.

James F. Carter          Voice 310 825 2897    FAX 310 206 6673
UCLA-Mathnet;  6115 MSA; 405 Hilgard Ave.; Los Angeles, CA, USA  90095-1555
Email: jimc at math.ucla.edu    http://www.math.ucla.edu/~jimc (q.v. for PGP key)