Re: [Madwifi-devel] Re: HAL stomps on driver/wlan TX seq?

[Sam L. <sa...@er...>]

On Wednesday 02 June 2004 10:27 pm, Anton Rager wrote:
> Anton Rager <a_rager <at> yahoo.com> writes:
>
> Guess I should supply a bit more info....
>
> 1 - I'm using madwifi from a CVS pull around 1/25/2004 - perhaps this is a
> known issue and newer sources/HAL have resolved.
> 2 - I'm using a 2.4.21 kernel and an a/b/g Linksys cardbus card (5212).
> 3 - I've hacked up the monitor mode checks and a couple other areas to
> allow TX during monitor mode. This works and allows pings to be injected
> while driver is emulating monitor mode (with an IP assigned to ath0). My
> next step is some user injection utilities to replace the hackish way I'm
> testing right now.
> 4 - I'm monitoring my TX traffic via another system with a prism2 card and
> ethereal.
> 5 - static 802.11 sequence numbers in my pings get overwritten at
> transmission (observed from anoter system), but the value of wh->i_seq is
> unchanged (my static value) during encapsulation/transmission. Frag and
> duration values are transmitted with my static settings.
>
> One of my goals is to create an airjack-like capability (monitor +
> injection) but provide more of a nemesis/hping3 like interface for 802.11
> frame crafting and injection of all possible 802.11 fields. Atheros/Madwifi
> appeals to me due to the lack of intrusive firmware (unlike
> Intersil/prism2), but this current issue makes me wonder what HAL/Chipset
> is checking/correcting.
>
> Another goal is arbitrary frame/response creation on 2.4/5ghz radios -- ie
> non- 802.11 MAC spec traffic on these frequencies. I'm wondering how much
> freedom I have for crafting custom protocols with Atheros/madwifi
> drivers....

I don't believe the h/w touches the 802.11 header.  The hal certainly is not 
involved.  If you provide packet traces things might be more clear.

	Sam