Re: driver similarities..


From: Jouni Malinen (jkmaline_at_cc.hut.fi)
Date: 2002-09-02 17:17:53 UTC



On Thu, Aug 29, 2002 at 11:22:06PM -0700, Jerritt Collord wrote:

> Would perhaps the architecture of the hostap driver lend itself as a
> reference for developing the Aetheros drivers?

I do not know the details of Atheros hardware, but I would assume that it is quite a bit different from Prism2 as far as MAC--host interface is concerned. I would assume that it would use a much more efficient method for transfering frames between the card and host system (well, at least it better use or it won't see maximum 802.11a performance). Thus, at least the low-level Prism2-specific part should probably not be used as a reference.

However, I would assume that AP management stuff, i.e., hostap_ap.c part of the kernel driver and hostapd daemon for user space, could well be quite useful with other wlan cards. I'm planning to modify the hostap_ap.c interface with the rest of the driver so that there would not be Prism2-specific data structures (i.e., just sent 802.11 headers and if something more low-level is required, use something that could be easily used also with other hardware platforms). After this, it would be useful to separate hostap_ap.c to a separate module that could be used by multiple drivers.

User space daemon, hostapd, is already more or less hardware model independent. It gets messages with 802.11 headers and I think it could be used more or less directly with other kernel drivers--assuming the other cards would support something like Host AP mode in Prism2 firmware. In other words, the management part of 802.11 AP (authentication, association, and now also the new IEEE 802.11X Authenticator code) could use the same code for different hardware models. Handling of ProbeReq/ProbeResp etc. might be added, if those are needed for other firmware implementation.

-- 
Jouni Malinen                                            PGP id EFC895FA


This archive was generated by hypermail 2.1.4.