Have you ever wondered why it is hard to get wireless networking right on alternative operating systems such as Linux and OpenBSD? It appears the blame for this situation lies with chip makers including the likes of Broadcom, Intel, Marvell and scores of other companies. These are the findings of Jem Matzan, who writes The Jem Report.
Matzan’s investigation shows that open source software coders are getting stymed by the chip makers. A major reason for the problems is that most companies are using proprietary firmware which cannot be redistributed, or has substantial restrictions.
Unrestricted redistribution of firmware files is satisfactory for some open source operating system projects like OpenBSD, FreeBSD, and many varieties of GNU/Linux, but others like Fedora Core and Debian demand an entirely free software environment, so redistribution of the firmware without the ability to modify and distribute the source code is prohibited.
Beyond that, the chip makers don’t provide documentation for their firmware, and as result it is really hard for programmers to write drivers for these wireless networking chips. The problem which essentially plagues wireless networking chips is now spreading to the wired variety as well, Matzan notes.
I am just offering a summary of what is an extremely thorough report including conversations with some chip company executives. Predictably, many of the majors dodged Matzan’s questions.
A lot of these proprietary features in chips may be coming because of Digital Rights Management requirements imposed by Microsoft for Vista. In one of the two links I’ve included in the post, it mentions a requirement that prohibits open source drivers… and some of that will be hard on Linux.
I suspect that it will mostly impact the desktop, not servers.
Check it out here: http://www.channelregister.co.uk/2006/12/27/windowsdrmmonstered/
And the original paper here:
http://www.cs.auckland.ac.nz/~pgut001/pubs/vista_cost.txt
I discovered these two gems via Slash.
One point often missed about Open Source control of computers versus Open Source control of radios is that poor programming of a radio can cause interference to other communications systems (that said, I’m NOT casting aspersions on the relative quality of open source versus closed source programming – Linux rocks!)
Many Wi-Fi chipsets are in essence “Software Defined Radios” – SO many of the RF characteristics are changeable under software control. Some regulatory agencies might say TOO many…
That’s why the real reason why the Wi-Fi chipset vendors are reluctant to encourage open source code to control their chipsets is that such open source code can make the chipsets perform in ways that are illegal in a given country. For example, in the US, the 2.4 GHz band is 2.4000 – 2.4835 GHz; a quick Google for the Japanese 2.4 GHz band says it’s 2.471 – 2.497 GHz.
If there is a “stink” raised about a particular Wi-Fi chipset being particularly easily “reprogrammed” to be able to operate illegally, a country’s spectrum regulatory authority could probably restrict the sales of that particular Wi-Fi chipset, including the products that it’s embedded in.
Hi Om,
I like the story and read through the original, but the analysis and conclusions are actually old hat. For a far more in-depth analysis of the problem (one that predates this one by over two years and is on exactly the same topic), I’d recommend a read-through of “Hidden Interfaces to Ownerless Networks” which I co-authored back in 2004 with Christian Sandvig and David Young — available at: http://www.saschameinrath.com/misc/Hidden_Interfaces.pdf
We presented the findings at the annual Telecommunication Policy and Research Conference in Washington, DC and the results were picked up by The Economist in January, 2005:
http://www.economist.com/science/displayStory.cfm?story_id=3535732
What Matzan’s piece does uncover is that absolutely nothing has changed in the 2.5 years since we conducted our original analysis. Contrary to what Matzan states, clearly the PR tactic is working for these companies — almost no one is talking and/or reporting about the monopolistic power held by chip makers, their refusal to allow open source development, and the subsequent retardation of innovation that this bad business practice causes. I’m happy to discuss more (including some of the more disturbing stuff that we couldn’t discuss publicly) — just drop a line to sascha@cuwin.net
In solidarity,
–Sascha
Answer here is simple. Most of the time the chip guys license this stuff themselves from third parties and don’t own it. No big conspiracy. Just economics.
Om – See Theo’s (of OpenBSD) presentation on Open Documentation for hardware here:
http://www.openbsd.org/papers/opencon06-docs/