On 06/03/07, Larry Finger <[EMAIL PROTECTED]> wrote:

> > Before I start bashing netgear who sells crappy wpn824's, I would like
> > to rule out possibility that it's a bcm43xx fault. Anyway something
> > went wrong in the module.
> >
>[...]
> If it is limited to wireless,

No, it isn't. However it wasn't problematic for tg3 driver to cope
with a flawed dhcp server.

> do you have another wireless system that could run ethereal or kismet
> and capture the traffic?
>
I've got ar5212-based pccard, so I was able to check link layer - seemed ok

What is appalling is the way the netgear router handles dhcp. No
matter what happens it always send 802.11 frames aligned to 566B to
clients. I also found that whether bcm43xx survive or not, depends on
the difference between declared length in an ip header and the real
one. E.g.  the difference of 220B doesn't lead to crash. The same
cannot be said about 262B . This might suggest a buffer overflow.
-- 
at
_______________________________________________
Bcm43xx-dev mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/bcm43xx-dev

Reply via email to