Re: CISCO Aironet 350

From: Jim Thompson (
Date: 2002-07-04 20:02:39 UTC

	I'll give Cisco a shout and see what they have for technical information.
	We've successfully written Fully-functional Access Points for Prism 2, 2.5
	and Lucent, with support from both Intersil and Agere, and would like to add
	Cisco to our arsenal.  We don't use Cisco, but many have requested it.

Good luck, man. I was about as far inside Cisco as you can get. I was one of the two people (the other was someone who shall remain nameless at MSFT) that convinced them to buy Aironet rather than Symbol, I know **the** guy who wrote all the AP4x00 and card firmware, and had a really important business relationship with both Cisco and Aironet (back when they were Aironet) since I was the CTO @ Wayport, and while I have the driver manual for the 4500/4800/350 cards (the Aironet MAC), and I know all the biz dev guys there, and both Larry Birnbaum (the VP of the BU that has the Aironet division in it) and Charlie Giancarlo (his boss, who reports to Chambers), I've never been able to get my mitts on the firmware.

Please do let us all know if you're successful.

I do have the Agere code (and firmware). Since I can't put it in Linux due to GPL restrictions, I haven't bothered to whack the driver into accepting the ternairy f/w download, press the button on the MAC, and deal with the frames that come up from the card in BSS mode. If there was a 'hostap' mode for the Heremes or Ruby chipset, then I'd have finished a driver for that by now.

        If you are interested in some of our products we handle, feel free to check our site.

Your business model is completely broken, I'm afraid. Your choice of FreeBSD has locked you to x86 platforms, and btw, where is the source?

> They only need to give the source code to whatever GPLd binaries they are
> distributing to those that they have distributed to, not everyone. I don't
> know if they are doing this or not but I certainly hope so. Of course, any
> modules that they've created themselves they do NOT have to provide source
> for.

Not quite. Read on.

David Sifry writes:
> Of course, Tony doesn't need to license the code he wrote (AP drivers, etc)
> under the GPL. He's the author, and he can create non-GPL'd binary modules
> if he chooses, amongst other options.

This isn't true. I've been engaging with Eben Modgen, rms and Linus over exactly this issue since last November on a wireless driver. While I can't discuss the details, I can say one thing for sure: You are (currently) quite wrong, though this is a commonly-held (but still incorrect) belief..

If Tony's kernel module code uses no Linux code, then he might have a case, but its doubtfull. In general, if your code runs in user-space, only links against LGPLed libraries, and contains no GPLed code, then you don't have to distribute the source. You should, of course, but that is a different subject.

What is absolutely true is that Tony distributes FreeBSD now, which isn't GPLed. However, if any of his code is based on, or contains, or links to code covered by the GPL, then he has to make the source available. Period.

By his own admission, Tony allows that there is probably GPL code in his product:

  Our product has Linux roots, which, along with the GPL is on our manual, but   our Wireless drivers are not based on any GPL code, and are developed   in-house by yours truly.

I've covered the (entirely unclear) issue of binary loadable modules above.

Acording to The tests for a GPL license violation are roughtly:

This all said, Linus has stated that he "probably won't sue" unless a GPL violation against Linux is just too wicked to let stand. Since Linux and a whole bunch of others hold 'the copyright' to Linux, there is little chance of fixing the issue.

Still, Eben Mogen is looking for a test case to prove the GPL is enforcable. I'm not sure that Tony wants to be it.

Again, good luck.


p.s. sorry this as gone so far off-topic, everyone.

This archive was generated by hypermail 2.1.4.