Broadcom Switches to the Light Side: The Start of a New Era?
As anyone familiar with the Linux wireless scene before 2006 knows, Broadcom, which manufacturers the wireless chipsets found in many laptops, was for a long time synonymous with everything evil about closed-source software. That’s changing. Here’s how.
Although Broadcom had very good in-house Linux drivers for its wireless cards, which it sold to manufacturers of wireless routers that ran on Linux, it refused to release the drivers to the community, even in binary-only form, or to provide documentation that would assist Linux users in writing their own drivers. Consequently, ndiswrapper was for many years the only way to get Broadcom cards running on Ubuntu and other distributions.
Eventually, the bcm43xx project (which has now become the b43 project and merged into the larger Linux wireless stack) managed, through tedious clean-room reverse engineering, to write native Linux drivers supporting most Broadcom chipsets. Taken for granted by many Ubuntu users today, bcm43xx represented a huge milestone when it first appeared in functional form a couple of years ago.
Last summer, Broadcom suddenly and quietly reversed its anti-Linux policies by releasing drivers for certain Broadcom wireless cards. Most of the source code is closed, and at this point only a handful of chipsets (mostly newer ones) are supported. But this development, which has remained more or less under the radar, nonetheless has important implications that extend well beyond the wireless scene.
Most significant is the fact that, as far as the (admittedly scattered and minimal) evidence indicates, Broadcom’s change of heart towards Linux was a direct result of the newfound leverage that Canonical is able to wield on behalf of the free-software community. An Ubuntu developer writes:
Dell and Canonical are both driving efforts to ensure that Broadcom is opening their drivers. With the quantity of Ubuntu systems being shipped outside the US to places like China, Dell does have the ability to speak with the dollar and will be doing so.
This ability for the Linux community (or its representatives) to “speak with the dollar” is enormously remarkable. Linux distributions have managed to secure limited deals with uncooperative third-parties in the past, especially in situations where it helped a particular company compete in the server market (think of Novell’s dealings with Microsoft, for example). But Broadcom’s sudden embrace of the Linux community seems different, due not to some self-interested scheme by Canonical, but to Broadcom’s realization that Linux—even on the desktop—matters.
Many individuals, especially die-hard disciples of Richard Stallman, will argue that Broadcom still has a ways to go if it wants to prove that it’s serious about Linux support. The source of its drivers remains essentially closed, with their future status unclear. As the Ubuntu developer writes, “It’s hard to go on a company’s word that they will open a driver up and play nice, especially Broadcom.”
All the same, these drivers are a concrete example of the legitimacy that Ubuntu has gained, thanks in large part to the leverage that Canonical enjoys. This isn’t to say that 2009 will become the elusive Year of the Linux Desktop, but it’s certainly a step in the right direction.
N.B.: if you want to test Broadcom’s closed-source drivers on Hardy without the headache of compiling them, check out this thread. They are supposed to be included by default in Intrepid, although I haven’t verified that assertion.
WorksWithU Contributing Blogger Christopher Tozzi is a PhD student at a major U.S. university. Tozzi has extensive hands-on experience with Ubuntu Server Edition and Ubuntu Desktop Edition. WorksWithU is updated multiple times per week. Don’t miss a single post. Sign up for our RSS and Twitter feeds (available now) and newsletter (launching January 2009).
That is great news. The Broadcom wireless chips were a very early drawback to my conversion. I was using Mandriva at the time and it was a pain in the rear to say the least. About this time Ubuntu started the “Restricted Drivers Management” that made things much easier and led to my switching to Ubuntu completely.
Wesley: Mandriva has included the bcm43xx driver for a long time. It has also always been possible to set up ndiswrapper through the Mandriva networking configuration tools. bcm43xx requires an external firmware which we cannot legally ship with the distribution, but the Mandriva network configuration tools make it as easy as possible to install this and enable the driver.
The process is documented here:
http://wiki.mandriva.com/en/2008.1_Notes#Required_firmware_for_Broadcom_wireless_adapters
and more here:
http://linuxwireless.org/en/users/Drivers/b43#devicefirmware
I feel sure that if you had simply asked for help on the Mandriva forums, or mailing list, or IRC, you would have found it quite easy to set up your card.
I’m posting this via a Broadcom wireless chip on Mandriva 2009, as it happens.
I think the most important thing about the closed Broadcom driver is it may actually provide a channel for legal redistribution of the firmware required by the bcm43xx / b43 driver. But the driver developers don’t seem interested in pursuing this idea, though I posted it to the mailing list. They continue to have the b43 driver work only with firmware extracted from the Windows drivers, which cannot be legally redistributed.
It’s a step in the right direction, but I’d prefer fully open drivers because then the network ‘just works’ in any linux distro. We’re not asking broadcom to write Linux drivers, we’d just like them to stop keeping everything secret and tells us how to talk to the chip so we can write our own drivers.
I have two well supported cards here, and I made a video that shows how easy wireless can be when you do have a well-supported chipset.
http://nz.youtube.com/watch?v=Hp3y96ayzY4
zcat,
One other issue to consider, which I didn’t have space for in the post, is that although Broadcom’s drivers are closed-source, Ubuntu (and everyone else) has permission to redistribute them freely. In contrast, the bcm43xx/b43 drivers rely on firmware cut out of proprietary binary files that Ubuntu doesn’t have permission to redistribute. This creates a very gray area from a legal standpoint, because we’re not redistributing the whole files, but only parts of them. So on the one hand, the licensing of the closed drivers makes them easier to distribute without legal uncertainties; on the other hand, their source is closed, which as you say makes it harder for them to be customized to work on any distro. Just another thing to consider.
Thanks for commenting, and nice video.
There’s no way I’ll use Broadcom’s closed drivers. I’ll stick with my Ralink wireless, thanks. It “just works” out of the box without this hassle. For my one work laptop that has the Broadcom crap, I’ll stick with bcm43xx.
Until Broadcom starts playing nice like Ralink does, screw ’em.
–SYG
Ahah! Broadcom. I guess that is who Mark Shuttleworth hinted amidst the Firefox EULA debate:
“For example, at the moment, we’re in detailed negotiations with a
company that makes a lot of popular hardware to release their drivers as
free software – they are currently proprietary. It would not be possible
to hold those negotiations if every step of the way turned into a public
discussion. And yet, engaging with that company both to make sure Ubuntu
works with its hardware and also to move them towards open source
drivers would seem to be precisely in keeping with our community values.”
Actually hmm… that just might not be Broadcom, if they’re only opening up proprietary ones.
also all telecomm devices have mandated back doors. If the software drivers aren’t right the back doors might not work. obviously the spy agencies would mandate to provide the official interfacing code to deliver a working system. beware closed systems.
Go Intel, Go ATI, Go HP, Go Broadcom (from now). NVIDIA, what are you people waiting for?
Vadim,
Very interesting. I hadn’t read that, but Shuttleworth may well have been hinting at the Broadcom deal–Broadcom’s drivers now are closed and proprietary, but Canonical is supposed to be pushing to get them more open. Unfortunately, this whole Broadcom thing seems to be happening behind closed doors–when I asked some Ubuntu developers for comments on it, no one gave me any more than very cursory remarks, citing “contractual obligations” not to talk about it. So it’s hard to know at this point what the future of these drivers is.
2008 WAS the year of Linux desktop(or netbook if you prefer it so). DELL , HP , ASUS, ACER amp; MSI sell Linux netbooks. And market share doubled to 2%.What else do we need to declare 2008 as THE YEAR OF THE LINUX DESKTOP???
Getting Broadcom’s drivers open-sourced is not the real issue here. Rather, getting Broadcom to allow distribution of their firmwares is. Since Broadcom will now permit redistribution of their entire proprietary driver, there is no good reason not to premit redistribution of the firmwares. Remember, when you redistribute the whole driver, you *must* redistribute the firmwares if you want the wireless NIC to actually work.
We *already have* a fine driver which is GPL’d, and that’s the bcm43xx/b43 driver. All we need now is simply permission to redistribute the firmware. *THAT’S ALL.*
If Canonical is indeed going to put Broadcom’s proprietary driver into the Ubuntu repositories, then that’s a disincentive for Broadcom to grant this permission. For Free Software’s sake, I hope Canonical doesn’t make that mistake.
–SYG
Sorry, but to me Broadcom has a long way to go to earn my trust. I’ve had to fight constantly with their products because they refused to release the specs to their chipsets. I’ve recently ripped out the broadcom from my Aspire One and replaced it with an Intel 3945ABG. The Intel does “Just Work” under Linux.
Sorry Broadcom, but until you release ALL the specifications on ALL your chips, I won’t be buying anything with them in it.
Thanks Donovan, your comment helped me to choose a wireless option for a laptop I am planning to buy soon.
N’est pas d’accord [url=http://runfr.com/acheter-cialis]cialis 20[/url] http://runfr.com acheter 1 comprimй viagra cidlis achat cialis
A gink begins cutting his wisdom teeth the senior chance he bites off more than he can chew.
Kid labor and meagreness are inevitably likely together and if you continue to manipulate the labor of children as the treatment on the group complaint of indigence, you pleasure have both destitution and descendant labor to the supersede of time.
Child labor and meagreness are inevitably likely together and if you endure to exercise the labor of children as the treatment on the group malady of pauperism, you inclination suffer with both poverty and youth labor to the d‚nouement of time.
Thank you for your site! [url=http://resingtuge956.hotmail.ru/doc28.html ]Antivirus[/url] and [url=http://resingtuge956.hotmail.ru/doc19.html ]skype[/url]