Matthieu CASTET wrote:
> I manage to associate to an AP, but when I ping the packet are sent tree 
> times :
> - when I ping from the client to the AP, the AP see 3 ping request on sent 
> 3 ping reply.
> - when I ping from the AP to the client, the client see 1 ping request.
> So it seem the packet are duplicated when sending them on the air.

I have also noticed this. The driver only transmits them once (as you 
can see in your client tcpdump), but the hardware seems to duplicate 
them. I suspect it is possibly timing out after no acknowledgment 
received and retransmitting -- maybe we aren't writing the correct 
timing values into the registers or something.

Now, some duplication is perfectly reasonable on networks (in this case 
we are just getting a *lot* of it) and it is up to the protocol to 
handle this. ICMP doesn't have any error checking or correction so it's 
not out-of-this-world unreasonable that you are seeing the duplication. 
Standard protocols like TCP will handle duplicates just fine.

Another interesting thing is that you aren't seeing duplicate ping 
replies received on the client - I see that a lot (or maybe tcpdump is 
just hiding duplicates?). It's not ultimately clear whether the driver 
should be weeding out duplicates (I think probably not) so this is again 
somewhat normal.

Anyway, does the driver work OK for normal usage (accessing the network, 
browsing websites, etc..)? If so, I will add the device ID.

> PS : the monitor mode return garbage. Is that normal ?

I think this is a bug in ethereal 0.99.0. The older versions work OK for 
me. Waiting for Ulrich's take on this as he wrote that code, but I think 
it may be ethereal-specific: the out-of-tree intel wireless drivers now 
use radiotap and we see the same bug there.

What is really really really frustrating me is that I can't get monitor 
mode working reliably.

My intel card works for about 20 seconds then goes partially blind and 
only sees about 50% of the traffic. After I finally figure out that 
ethereal is the culprit, ZD1211 monitor mode also seems to be missing 
quite a few frames.

I have a new laptop with an ipw3945 which I'll try next. It's just 
getting really awkward as the ipw3945 driver requires an out-of-tree 
ieee80211 so I'll have to set up another kernel, etc...

If you have a working monitor mode setup it would be interesting to see 
if you see the same kind of duplication from the ZD1211 vendor driver.


Zd1211-devs mailing list -

Reply via email to