I have seen many messages regarding supplicant, feel I can offer some insights here. Simple to the question, yes and no. Yes, you can port any open source supplicant to wimax device, while wpa_supplicant is the most popular one. No, the work is not just porting, supplicant is much more than that. The purpose of supplicant is to communicate with wimax backbone AAA server to derive the MSK/eMSK key, which it can hand over to 802.16e mac key manangement module for the key managing stuff, i.e. kek, tek, etc. 802.16e device and AP mac software don't participate in this process, and they don't understand this either. So 802.16e spec use pkm message to wrap around whatever the supplicant protocol, mostly it's radius, and just tunnel thru the message to backend AAA server. In addition, when the key expires, key management module need to work with supplicant again to rederive the key. In theory, you can move everything the host/device side (which is this linux wimax project) into 802.16e mac software, assume the mac has enough memory. You can build all the NAP/NSP parameters into the firmware download process so mac can run it alone to achieve network entry. But supplicant is the only exception, because it needs to deal with certificate and use ssl package. To port over wpa_supplicant, you need to: . strip out wpa_supplicant outer layer, i.e., eapol, replace it with properitary protocol state machine that host and mac agree on, so the device mac layer will tunnel thru supplicant protocol in pkm message. . in addition, you need to work out all the reauthentication and expiration issues. In tradditional supplicant, once the MSK key has been derived, supplicant is done. In wimax device, supplicant still need to work with mac software for 3 way tek handshaking, and provide API for outside world for its security parameters. In essential, you need wpa_supplicant for its eap state machine, which in turn links to openssl for certificate processing, and its data structure. There can be a whole document describing 802.16e key management and supplicant. I just throw out a simple outline. In practice, I have developed wimax supplicant for a Motorola wimax device which is based on wpa_supplicant, using the steps that I have described above. I have ported over Beceem's supplicant to an embedded device, you guess it, they use wpa_supplicant too. Charles -----Original Message----- From: wimax-bounces at linuxwimax.org [mailto:wimax-bounces at linuxwimax.org] On Behalf Of Hoi-Ho Chan Sent: Wednesday, September 30, 2009 3:14 PM To: Greg Lee Cc: Richard Farina; Inaky Perez-Gonzalez; wimax at linuxwimax.org Subject: Re: 64 bit Hi Greg, Is it possible to port the open source wpa_supplicant and make it work with Intel WiMAX service? Though I am not sure if it works on 64-bit systems. On Wed, Sep 30, 2009 at 2:15 PM, Greg Lee <glee-list at swspec.com> wrote: Hi Inaky, do you (or any others from Intel) have an estimate for when the open source supplicant will be available? I want to propose the Intel WiMAX Link 5350 for a product that my company is making, but we run Linux on an Intel XScale (ARM) processor so we will need to compile all of the source code ourselves including the supplicant. Please give us an estimated time frame on when the open source supplicant will be available so that we can factor this into our product plans. Thanks, Greg On Sat, Aug 29, 2009 at 2:49 PM, Richard Farina <sidhayn at gmail.com> wrote: Inaky Perez-Gonzalez wrote: On Fri, 2009-08-28 at 14:25 -0600, Richard Farina wrote: Inaky Perez-Gonzalez wrote: On Fri, 2009-08-28 at 11:56 -0400, Richard Farina wrote: ... First of all, thanks for your quick response Inaky, it is appreciated. While the prospect of an open source supplicant is obviously preferred, is there any way to take the old binary supplicant and compile it for 64 bit? Or make the 64 bit wimax network service able to use the 32 bit binary blob of a supplicant? I love the idea of an open source supplicant, but at the moment I'd take free as in beer while waiting for my free as in freedom. At this point, that'd be probably more work than the final push to the open source one. I understand the frustration, but we are severely tight with resources here, so we need to place them where most makes sense. In the long term, the open source solution is the best one, so the little we can divert resources from anyone, we try to put them there. Understandable to be sure. I figured if it was as simple as just compiling the binary in 64 bit that it could be done easily but if work is required I would definately rather see the resources go into the open source version. I help fun a small livecd for which I've added wimax support to the 32 bit version. If you need anything related to the 64 bit work you are doing tested, I have a small user base that is equipped with appropriate hardware and I'm sure they would be very happy to 1) Help test and 2) Have a working device :-) Thanks again, Rick _______________________________________________ wimax mailing list wimax at linuxwimax.org http://lists.linuxwimax.org/listinfo/wimax _______________________________________________ wimax mailing list wimax at linuxwimax.org http://lists.linuxwimax.org/listinfo/wimax -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.linuxwimax.org/pipermail/wimax/attachments/20091001/e884e0b8/attachment.html>
