Ahem. I just had to write that all-caps headline. Broadcom’s wireless chips – used by just about everybody in this industry – have been a major pain in the bum for Linux users, because the company did not release open source drivers. Workarounds had to be created, lots of pain was had in the process, but now, Broadcom has finally seen the light: they have open sourced their wireless Linux drivers.
I still can’t believe it. Let me copy the official announcement verbatim.
Broadcom would like to announce the initial release of a fully-open
Linux driver for it’s latest generation of 11n chipsets. The driver, while still a work in progress, is released as full source and uses the native mac80211 stack. It supports multiple current chips (BCM4313, BCM43224, BCM43225) as well as providing a framework for supporting additional chips in the future, including mac80211-aware embedded chips.The README and TODO files included with the sources provide more details about the current feature set, known issues, and plans for improving the driver.
I’m at a loss for words. Except, uhm, Broadcom – couldn’t you guys’ve been a little faster with this? It would’ve saved us a lot of effort. Kthnxbye.
That pretty much leaves only nVidia as the only idio^h^h^h^h company holding out on us. Granted there are other graphic card options that make an nVidia announcement less important than this one. But it will be nice when the day finally comes that closed hardware is no longer an issue and we can concentrate on competing with closed solutions higher up the stack.
Thanks Broadcom
Credit where due and I must give Nvidia credit for at least providing a usable closed binary driver. It installs easily if the distro does not already have it packaged in a non-free repository. As far as companies who’ve chosen to go it alone and only provide a closed driver; Nvidia has done well. Many other company’s have delivered piss poor token drivers if they’ve delivered drivers at all.
I sure wouldn’t complain about them releasing source so that all effort can be focused on the existing code base instead of rewriting it under the Nouveau project. Drivers have no business not being open source and anything that makes general purpose hardware more general purpose software compatible is a good thing. I am currently waiting for some polish for Debian Squeeze packaged nvidia-glx for my GTX260 and polish for the Debian Squeeze packages xserver-xorg-video-intel for the Intel gpu in the Lenovo X201. Until those two items work I’m stuck with Nouveau’s lack of 3D support for Nvidia and the Xorg Vesa driver which is sadly lacking compared to what my notebook GPU should be doing. I’m patient since Squeeze is still “Debian Stable 6 RC” with emphasis on Release Candidate at present but still, there is really no valid justification for the amount of hardware grief bad vendors cause.
You can easily get by with Nouveau already.
So far it’s been easy to avoid broadcom. As an example, a Dell laptop I got offered the option to choose either intel or broadcom wlan chip.
Netbooks seem to ship with Atheros cards that work great now.
I imagine that some OEMs have told Broadcom that they will do their hw sourcing elsewhere because of the driver situation – all big companies have Linux savvy people around at places where their voice gets heard.
Wifi chip is a cheap commodity, and I can’t think of anything special that Broadcom could offer that would justify choosing them over anything else.
How does closed hardware prevent this?
It’s easy as long as the PC is Intel based. AMD based stuff get stuck with Broadcom networking chips more often then not.
Maybe now AMD PCs won’t give me heartburn when I think about buying one.
What the open source world needs to provide is a Mesa stack with a good interface so that at least the driver itself can remain closed source (for the short term) and the OpenGL part of the equation is then pushed off to the open source world thus to have a standardised single library as with the case of MacOS X rather than the current situation of Linux where you have a mixture of OpenGL implementations each introducing their own peculiarities and bugs.
With that being said, there is always ATI but even then the open source drivers are always behind the times when it comes to performance and support – the man power just isn’t there and open source isn’t the silver bullet that’ll solve problems as some try to make it out to be.
Well BroadCom have been providing a driver for quite some time:
http://www.broadcom.com/support/802.11/linux_sta.php
Sure it includes a binary blob but all what is required is provided in an open source way which doesn’t prohibit developers from shipping it with their Linux distributions. I guess the big step forward is having an entirely open source driver rather than a hybrid which is a step in the right direction.
You don’t support open source by enabling closed source solutions. You support open source by pursuing and encouraging open source solutions.
You’re right, open source isn’t a silver bullet. But open source can be a goal in and of itself. Open source solutions have certain traits that are desirable beyond the latest and greatest performance metrics. There is real liberation gained when you’re not tied to a single vendor who controls the source. As one example, support for old video cards can be maintained long after the original business that sold them has dissolved or just lost interest in the product.
IMO people need to stop being so obsessed with performance or the latest bells and whistles to appreciate the real value of open source in freedom and risk mitigation.
That doesn’t help address some of the important qualities that attracts business and hackers to open source in the first place. You don’t have to be a radical freetard to appreciate and value open source. And once you appreciate its positive qualities, the nVidia or Broadcom closed-source offerings for Linux aren’t interesting or praise worthy.
That being said, sometimes a company like Broadcom can not make the internal shift needed toward open source in a single leap and must make a gradual transition as we saw in this case. Though not ideal, it’s understandable and we can forgive them their late arrival to the party now that they’re really ready to join in.
It will be really nice if someday we can say the same about nVidia.
I disagree; you get to open source by making sure that people conform to the open standards and don’t use trickery behind the scenes to artificially boost the performance of their drivers. In the case of Mac OS X the OpenGL implementation is provided by Apple, it conforms to the OpenGL specification tests and more importantly it does not tweak or optimise for one video card over another. The downside is that it probably doesn’t use nVidia trickery to get higher scores but the upside is a consistent experience across the board. I for one would sooner the focus be moving towards getting uniformity in open standards implementations before we even more onto discussing such issues as open source.
But at the same time if you’ve spent a couple of thousand on a laptop or desktop you want to be sure that you are using it to its maximum potential rather than having it hamstrung by inefficient drivers or poor implementations of an open standard such as OpenGL and OpenCL. With the latest drivers from Apple the performance of OpenGL might not be as great as say on Windows but I’m willing to give up a small amount of performance for the sake of stability, conformance to open standards and so forth but there shouldn’t be a massive gap or otherwise the end user will simply argue that they’re not going to tolerate such a compromise.
I appreciate open source from the point of view that I’ve seen businesses refuse to support hardware because it 2 years old – printer companies forcing upon end users obsolescence simply to pump up profits. What I would like to see is a law that forces companies that after they’ve stopped supporting the hardware they must provide the full open specifications so that third parties can implement hardware support – and gradually work to develop a set of standards so that a standard driver can be provided that supports any hardware conforming to it – and any product sold must conform to those standards.
I hope you do realize that the Mesa stack is a pivotal element in the current FOSS driver development for graphics cards.
Get NVidia to open source their driver and the trickery they play with their blob isn’t possible anymore. In which case it doesn’t make sense to keep providing a separate OGL implementation and Mesa is the logical choice to fill the gap.
FOSS, once adopted, tends to promote the use of common infrastructure and reuse of code.
This sounds good in theory, but doesn’t always work in practice. For example the current issues with Intel i8xx graphics chips.
If you don’t know about that, the gist is that those chips don’t work well with the new architecture (KMS/GEM/DRI2). And there’s simply not enough manpower to get those chips working, as they’re very tricky beasts. The solution Intel came up with is introducing ShadowFB support into KMS. That way you at least get modesetting on those chips, but shadowfb means no 2D or 3D acceleration whatsoever.
Bottom line, even in open source, hardware will lose support after a while.
That pretty much leaves only nVidia as the only idio^h^h^h^h company holding out on us.
Hahaha. What a clown. Nvidia doesn’t give a damn about you. They have business customers who rely on Linux. That’s why there is a driver. That’s why the driver doesn’t suck. Go back in your cave. And stop open sourcing your brain. It hurts.
Edited 2010-09-10 04:43 UTC
Excellent point, very thoughtful.
You think so, too? Excellent. Have you anything else to say besides “making everything open source will stop the wars and make the world a better place”? I doubt it. You’re a pitiful fundamentalist without a real opinion.
Would have been nice to find a more positive note before calling it a day here ;o) My first thought was to criticize you for substituting venom in place of rational argument. But maybe you’re just having a really bad day and need to vent. Hope that after a good night sleep you can see that not everyone who supports and appreciates open source is a fundamentalist. Cheers.
I can see that already. But people like you, whose criterion for quality, above all else, is whether software is Open Source or not, qualify as fundamentalists in my book. So sorry, Hans.
Being separate from the kernel means that the Nvidia driver breaks with every kernel update.
Being separate from Xorg means that the Nvidia driver is incompatible with recent developments in the X stack, such as kernel modesetting (KMS).
KMS is a pre-requisite for the coming upgrade to X that will allow “rootless” X … as long as the graphics driver is part of the kernel and supports KMS, soon it will be possible to run an X session with non-root privileges. This is a great advance from a security and stability point of view. It will mean also that it might become possible for X to crash and be re-started without having to close applications or login sessions.
http://www.phoronix.com/scan.php?page=news_item&px=NzM2MA
None of these advances will be available to people who run the binary Nvidia graphics driver.
Very soon, if not already, Nvidia will be very much the non-preferred graphics option for use with Linux machines. It may still be possible to run Linux on a Nvidia graphics machine using the nouveau open source driver, but that project is regrettably falling further and further behind in capabilities.
Edited 2010-09-10 05:36 UTC
No, that would be the fault of the non-existence of a stable API/ABI, which Greg “Clown” Kroah-Hartman claims isn’t necessary.
Linux would be even more useless without an Nvidia driver.
Edited 2010-09-10 07:27 UTC
The Nvidia driver is pretty bad because it is a binary driver. You don’t have a working system with it installed. Power management doesn’t work for instance. Intel and ATI have the best drivers for GNU/Linux and stuff actually works. Nvidia just has the best gaming cards on GNU/Linux at the moment because the free ATI GNU/Linux drivers aren’t finished. The free ATI GNU/Linux drivers are coming along though and you’ll find any day now the better buy will be ATI on GNU/Linux. So your claim it would be “more useless without an Nvidia driver” is baseless. I doubt there is a game you couldn’t play if Nvidia ceased to exist on GNU/Linux. Check out http://www.thinkpenguin.com/ for free hardware. Then you won’t have these issues. These broadcom cards still probably won’t be problem free and I’d stick with an Atheros chipset. ThinkPenguin is the only company selling for instance an 802.11n USB wifi card with a free chipset. It’s the only card that’ll work with GNU/Linux if you plug it in that doesn’t require a high degree of hacking to get it to work. Unfortunately companies don’t get that you can’t require users to download non-free firmware, drivers, and other non-free software. GNU/Linux is based on freedom and you need to comply with these norms if your hardware is going to work on GNU/Linux.
Edited 2010-09-10 08:59 UTC
Ah! Now comes the time where I can finally use the all-purpose Linux evangelists defense spell:
WorksForMe!(TM)
I have a working system with the Nvidia driver. I have a working system with the Nvidia driver and power management works. Your hypothesis has been falsified. Try again, Hans.
Maybe you can come up with something more substantial. Like proving that a major number of Linux users using the Nvidia driver has severe problems. Should be easily possible because the Nvidia driver is: bad, bad, bad! But then there would be still a question left to be answered: is it the fault of the Nvidia driver or the fault of Linux?
The mystery remains.
Look, Hans, the X server is a certified piece of crap. Unfortunately nothing better exists. Thus we have to make do with this abomination.
Oh! Already? How marvelous. And here I thought, Linux was behind the times.
Please! Give me a break! Nvidia works. Now. Case closed.
Why should I give you a break when you simply dismiss, with irrational venom and no actual point, the sound reasons why a closed binary driver is not preferable?
Why should I give you a break when you simply dismiss, with irrational venom and no actual point, the sound reasons why a closed binary driver is not preferable?
Me thinks DerGenosse is just a certified troll. I can’t imagine any legitimate Linux user liking the situation, where you need to get by on a sub-par 2D driver until you can install the NVidia binary blob or the situaton where you have to wait months before NVidia gets around to make their blob compatible again with the next iteration of the X server.
An open source driver, maintained by NVidia and shipped out of the box, would be leaps and bounds better for usability then the current mess with hooking their closed blob in the mix after installation.
The NVidia driver being closed source does nothing inherently to enhance the quality. closed or open, the code remains the same. Being closed does however reduce the ease of use / end-user experience considerably in comparison to FOSS drivers shipped out of the box.
This is a falsehood too often repeated, so much so that some probably think it’s true. X has problems, but so does any large block of code.
Do you actually have a basis for this fallacy or are you just parroting what other people say because “I can do $a in Windows but I don’t know how to do it in X! I’ll bet that $b feature of X that I don’t use is the cause, therefore I’m sure it’s not worth having! QED, X sucks.” If this sounds like you then please stop talking about X.
I agree, which is why for years I have bought nVidia cards despite them being closed source. I am not Stallman; works trumps doesn’t work. In the future I won’t do this, because when I have works+closed and works+open I’ll take the latter, even if it has a few more bugs.
The nvidia blob doesn’t require X running as root and hasn’t for quite some time. All it requires is access to the /dev/nvidia and /dev/nvidiactl device nodes, which is achieved by putting the user in the video group, where they need to be for open source drivers too.
There’s other stuff requiring X to run as root for now, I don’t know details, but I think it regards input device handling, and there’s also X writing it’s log to /var/log where a user doesn’t have write access. Once that stuff is fixed, both open source KMS drivers and the nvidia blob will run on a “rootless” X.
Not according to this article:
http://www.phoronix.com/scan.php?page=news_item&px=NzM2MA
It reads as if KMS is a requirement, and that given that then “it’s rather easy to get the X Server running without any special rights”.
It also infers that Nvidia users will have to be running Nouveau and not the Nvidia proprietary binary driver.
Phoronix is basically one person. He can and does make mistakes. Here you have a confirmation of what I said, straight from a Nvidia employee: http://www.nvnews.net/vbulletin/showthread.php?p=1946515#post194651…
And there’s other such posts from Nvidia employees at nvnews.net too, this was just the first one I found after a very quick search.
On the exact same page that you linked:
The monolithic huge blob nature of the Nvidia driver, derived as it is from Windows, in conjunction with the closed-source licensing, means that it must be run as one chunk as a loadable kernel module via an open source wrapper layer.
This means that although in theory the Nvidia driver could be run without root privileges, in practice on Linux it cannot be, primarily due to the licensing issues.
Ergo, no “rootless X” on Linux for users of the proprietary Nvidia driver. Its KMS, using an open source driver, or nothing, if you want “rootless X” on Linux.
At the very least, the KMS parts (which go in the kernel) have to be open source, but perhaps Nvidia might get away with providing the rest of the driver as a binary blob. That wouldn’t be ideal, and it might be a lot of work for Nvidia to split their driver, but it is the only way for Nvidia’s driver to remain mostly closed and still compatible with the future directions of X on Linux.
Edited 2010-09-11 13:54 UTC
Sorry, I don’t follow your logic. All the nvidia driver needs is access to /dev/nvidia* and that’s it. I don’t see how licensing affects that.
Licensing issues mean that the Nvidia driver cannot be part of the kernel itself, and it cannot be delivered with Linux distributions. Because they are distributing GPL code which they did not write, Linux distributions must make the source code avialable for that which they distribute. Therefore, video modes cannot be set by the kernel directly on boot (unless the nv driver or the Nouveau driver is used) … so one cannot have kernel modesetting.
AFAIK, kernel modesetting is a requirement within the design of X itself in order to have “rootless X” on Linux.
These are not restrictions brought about by the characteristics of the nVidia driver itself, rather they are restrictions that Linux distributions must follow due to the way Linux distributions are licensed. They are restrictions of the licensing of the Linux kernel and X itself.
These restrictions could be worked around if Nvidia were prepared to either use parts of their nv driver or the nouveau driver for kernel modesetting only, and then have their binary blob driver handle the rest of driver funtionality, but as yet nvidia haven’t been prepared to do that.
Edited 2010-09-13 03:57 UTC
It’s a requirement with open drivers. But this has nothing to do with the nvidia blob (which BTW, does modesetting in the kernel too and always has, just not via the KMS interface). Again, with nvidia you only need access to /dev/nvidia*. All the other stuff about “not being able to ship the driver” you write, while true, is completely irrelevant regarding this.
awesome news
The wifi mini-pcie card I’m looking on my desk is now no longer worthless. I pulled it out of my Dell laptop, which now has a Intel card in it.
Edited 2010-09-09 18:13 UTC
That means, before long, I’ll finally be able switch to FreeBSD on my laptop!
You can already do that IIRC, FreeBSd has the bwi driver which is ported from OpenBSD the downside I don’t think it supports WPA Maybe someone can create an abstraction layer so that it is easier to port Linux wireless drivers to other platforms.
Cool. I’ll check it out.
Last I messed with it, the only way was to use NDIS wrapper, which for some reason or another, simply does not work on my laptop.
It is the one thing that holds be back from OpenSolaris – I just hope that moving OpenSolaris out to a real community orientated development that we’ll see faster progression of drivers being ported over so that one doesn’t need to result to ugly hacks such as Windows drivers on Solaris/*NIX/*BSD.
Btw, interesting side note, just had a look at the ONNV notification over at OpenSolaris and it appears that Oracle/Sun has stopped merging their code into the tree – have Oracle/OpenSolaris stopped providing changes to the public?
http://www.osnews.com/story/23683/Oracle_Kills_OpenSolaris_Moves_De…
Cool, thank you for the link – I just hope that the consequence of such a decision rolls of in the form of Solaris 11 which really addresses the short comings of Solaris/OpenSolaris. Hopefully the focus will not just be on adding more features that niche customers use but also adding features that will address demands by potential customers who have put off going with Oracle because of the lack of these features.
On a good side with the release of the BroadCom source code there is already a move in the illumos to port the driver to it – which hopefully will roll off to being able to get installed on Solaris 11 once it is released
You can already do that IIRC, FreeBSd has the bwi driver which is ported from OpenBSD the downside I don’t think it supports WPA Maybe someone can create an abstraction layer so that it is easier to port Linux wireless drivers to other platforms. The good side as someone mentioned is the fact that it is MIT/BSD licence which will hopefully mean that it’ll be ported soon assuming it is well documented and readable code.
This means the end is coming right? Wow… I would have thought Broadcom would have gone out of business before getting a clue, but here we are! What changed? Did the CEO who was recto-cranially inverted retire or something?
–bornagainpenguin
Hi,
For years people have been saying that Broadcom will open source its drivers when Duke Nukem Forever is released… ;-D
-Brendan
Just browsing the copyright headers of some of the code files:
http://git.kernel.org/?p=linux/kernel/git/gregkh/staging-next-2.6.g…
It even appears to be a MIT/BSD license
umccullough noted…
Could someone from the Haiku project comment on the usefulness of this release towards getting WiFi working well on Haiku?
–bornagainpenguin
I suspect it will only have any major advantage to Haiku when it has been merged into FreeBSD where we have already “lifted” most of the WLAN stack and drivers from already.
As for “working well” – that isn’t driver-related anyway. The missing support for full WPA/WPA2 is something that needs to be written/ported – and the necessary userland utilities to manage the network and encryption selection need to be created.
How long ago did ATI / AMD release their graphics drivers as open source? Aren’t they still bad compared to NVidia’s?
No, they are WAAAY better than nVidias provided open source drivers….
Oh, that not what you meant?
As someone who has used radeonhd, and now radeon opensource drivers since they because available. Its been about a year since they have been usable.
Just over three years since they announced their intent to release documentation. They are not to the point that they are so stable and automatically configured that you just forget they exist… but they are getting close.
Meanwhile, as nouveau with kms is enabled by default on many distros, using the nvidia drivers has become more painful to set up. Hopefully nvidia has a plan to use kms. Their drivers have long rendered the other ttys on your system unusable unless you disable any kind of framebuffer.
But back on topic… This is fantastic news!
One can only hope Broadcomms embedded GPU drivers follow suit!
> Their drivers have long rendered the other ttys on your system unusable unless you disable any kind of framebuffer.
Erm, no, I’ve never had an issue with nvidia drivers and vesafb. Granted I’ve never used the new vesa fb, and the old one is a little.. fixed.
Do you mean “It’s been about a year since the time when they were usable,” meaning they were once usable but are no longer, or “It’s been about a year since the time that they became usable,” meaning they have been in a usable state for a while?
From the rest of your post I think you mean the latter, but it was confusing as written.
In my opinion, they are now pretty much both of same quality.
On Windows, you download the same amount of crapware, and you have a fairly low failure rate (except for new technologies like Optimus that always have a fairly high failure rate).
On Linux, ATI drivers now work much better (haven’t seen them failing in a while, except for window rendering mistakes, though they used to be nightmarish), while NVidia drivers sometimes won’t work on hardware marked as supported (especially laptop chipsets) for no reason.
A better level of quality can now only be achieved through better standardization (in hardware interface and in Linux API), test-driven development, massive open sourcing, and more user feedback, in my opinion.
Edited 2010-09-09 22:02 UTC
ATI / AMD have never released their graphics drivers as open source. Rather, AMD released the programming specifications for their graphics chips, so that open source programmers could write a new open source driver.
Here they are:
http://www.x.org/docs/AMD/
As you can see, for example, the specifications to program 3D functionality for ATI R600/R700 GPUs were released on 07-May-2009. This means that the open source programmers have had just over one year to write their driver for 3D functionality.
This has nothing at all to do with the closed-source proprietary binary driver for Linux from ATI, called fglrx. This driver has improved considerably in recent times, and is now on par with the nVidia closed-source proprietary binary driver for Linux.
The new open source driver written for Linux for ATI chips by open source coders is called xf86-video-ati. It basically works, but it is still under heavy development. Being only a few years old, much of it much younger even than that, performance-wise, stability-wise and feature-wise, it is not on par as yet with either of the closed-source binary drivers for Linux mentioned above.
Edited 2010-09-10 02:27 UTC
Duke Nukem Forever, Apple allowing Flash developed apps in the App Store, Open Source Broadcom drivers, surely the end of days approaches.
From the glass is half empty department….
http://www.phoronix.com/scan.php?page=news_item&px=ODU3NA
It is snowing in Louisiana, in July.
(yes, a reference to places freezing over.)
I am glad to see this a big thanks to the company.
Surely Broadcom wouldn’t make an announcement with a glaring punctuation error and refer to themselves in the third person.
Not sure if this is a hoax or not–though there are actual source files so if it is it’s an elaborate one–but it’s not that uncommon for large corporations to refer to themselves in the third person in their press releases. I don’t really understand why myself and it has always seemed rather pointless and pompous to me, but this is hardly the first time I’ve seen it done.
Because those press releases are sent to major news outlets to be copied and published verbatim… it wouldn’t sound so nice if major news outlets like BusinessWire, etc. published these releases with statements like “We are announcing…”
Standard PR practice.
Edit: As for the punctuation error – marketing drones aren’t always the most educated people – some are merely interns!
Edited 2010-09-09 21:15 UTC
Seems to me that’s what quotation marks are for. But then, that’s probably why the PR field never interested me.
You mean someone might be pulling a Phoronix-style announcement here?
Edited 2010-09-09 21:11 UTC
Great newz
Never has an article been more deserving of an all caps title!
I just left a thankyou message on the Broadcom website. Good news.
This doesn’t just leave nvidia, there’s still all the tv tuner companies that don’t have open drivers or drivers in general. Look at Philips among other manufacturers. Sure now you can buy any laptop and get support for most of the hardware but if it has TV tuner support in most cases you are still screwed. For video card support Linux needs the open source stack to implement things like opengl 4.1. Why would nvidia open their driver so their competitors can start offering opengl 4.1 and cuda? The community should already have implemented this stuff in the open source drivers.
You’re right. But Broadcom and nVidia were two very large and important holdouts relative to the other players.
It’ll happen. But it would happen a lot faster if there wasn’t so much wasted energy going into creating open source drivers for video cards. A huge amount of time has been wasted by very talented programmers on reverse engineering ATI and nVidia devices.
For the same reasons that AMD opened their drivers and technical manuals.
Often the fear of openness is much worse than the reality. There is still lots of room for product differentiation at the hardware level.
Hiding the source will only work until a critical mass of people appreciate and demand open source solutions. With companies like AMD contributing such solutions, the taste for them will surely grow over time.
Nitpick: AMD did not open their drivers, only their technical programming manuals. AMD did, however, release some SDK-like demo code as open source. This code was not a driver.
Edited 2010-09-10 02:24 UTC
Hello,
Now they hopefully can release the source of the rest of their drivers for servers. Their network drivers are a world of suck on any platform or media type.
Good on them for finally doing this. My company pulled all the Broadcom wireless cards from the wireless PCs on the floor anyway. I’ve never seen wireless cards that just drop connections and act up like theirs. We replaced them all with Intel cards and never looked back. The drivers are a ton better too.
…both in Windows and Linux. In Windows, sure, it works… at first. But for some reason, the drivers decide to completely stop working at random and need some fiddling or a reinstall; either way, it’s plenty of wasted time, and I’ve seen some weird things happen in Windows with wireless drivers (ie. complete inability to remember what password I just entered two seconds ago for a wireless connection). Linux? Good luck. Either it doesn’t work or… it doesn’t work. If you’re lucky enough to have a wire going to the machine, some distros have the ability to easily download the drivers… but that defeats the point of wireless in the first place, doesn’t it? Having to use a wired connection to download wireless drivers?
This is a major win here. I’ve tried buying wireless chips that are Linux-friendly, but when I do some snooping around, I always find out that the particular revision/version of the chip is the dreaded Broadcom. Just one question… does anyone know if this includes devices that use firmware that must be uploaded to the chip? Is that firmware part of the “drivers” or is it still locked down by copyright?
Too bad this didn’t happen sooner, like before Debian froze Squeeze…
Edited 2010-09-10 03:37 UTC
Debian has apparently re-instated support for backports.
http://www.linuxcompatible.org/news/story/debian_backports_service_…
Yeah, I heard about that. But doesn’t that mean that, still, the backports repository must be added manually and anything in it needs to be installed manually as well?
What I meant is, it would be nice if this would have happened sooner so Debian could have put it directly in the distribution. In other words, no need to have an ethernet cable connected to the machine to download bc43-fwcutter or whatever it is, it would just work after installing the OS. AFAIK, backports doesn’t work that way. It’s 100% manual.
AFAIK:
Once the driver is accepted into the mainline kernel, you will automatically get the driver when you update the kernel.
Debian versions such as squeeze do not update the kernel version (only the minor version). The only means to get a new driver is to have it available as a kernel loadable module via a backport to the earlier kernel version. Even then you have to suppress (blacklist) whatever module is being loaded now for wireless, and explicitly load the new backported driver module. Happily this can all be achieved through editing configuration files in /etc
Edited 2010-09-10 05:46 UTC
That’s cool if true, but unfortunately, it still means you’ll need to haul the machine to another room and connect it to another monitor, mouse and keyboard just to download the updated kernel. And then move it back to where it belongs and set it up yet again.
Unless a future version (Debian 6.0.1?) contains this updated kernel on disc, if what you’re saying is true… if so, awesome, that means the next point-release of “Squeeze” will be fully Broadcom-aware.
But I don’t know… I saw how Debian treated a completely broken package (Pidgin) when Yahoo! changed their protocol; they effectively said “it’s not a security issue, so we’re not fixing it.” Well, yeah, I guess if you can’t even connect to the damn service, then it must not be a security issue, eh? Never mind that the fact that the Yahoo! Instant Messenger service completely refused to work rendered Pidgin completely broken for anyone who used the service. That’s why I’d be surprised if Debian supports this new driver now, after the freeze.
Edit: Nevermind, I completely misread the second paragraph quoted. Sounds like it’s just the way I expected. [I’m drinking, so admittedly I’m not paying complete attention… heh.]
Edited 2010-09-10 06:59 UTC
Or you could just download the backported driver package on another machine and copy it from there to a USB stick … this method is called sneakernet.
http://en.wikipedia.org/wiki/Sneakernet
This is a high-latency but nevertheless relatively high-bandwidth tried and trusted data transport protocol that goes way back to the days of floppy disks.
Edited 2010-09-10 11:01 UTC
Did Broadcom also release hardware specs and documentation, enough to write a driver from scratch? The BSD developers I read tend to favor that approach more so they can write their own drivers.
Broadcom’s opening their drivers is a good first step, and I applaud them for it. Hopefully they will start doing it for some of their other hardware (like that video codec coprocessor they have).
As for nVidia, I bought an ATI card for my last desktop I built, since they had recently released their hardware specs, in order to support them. The open-source drivers didn’t have full 3D functionality yet, so I used the fglrx ones — and had terrible luck with them. System crashes, display corruption, various things. Went through several updates and finally got so frustrated I bought an nVidia card. Binary blob or no, the thing just worked — no muss, no fuss. I hope the ATI drivers have caught up, as some have commented, but unfortunately I cannibalized the card’s fan, and I now can’t find the screws to the heatsink. 3rd-party sinks don’t work due to the card vendor taking liberties with component layout
Anyone check before they posted this?
-Hack
At this rate, it may end up nearing absolute zero down there in 40-50 years.
Edited 2010-09-11 19:27 UTC
Wow, that was *fast*. I’ve already sold my bwi0-equipped notebook just because it didn’t want to play nice with my OSS OSs.
Ahhh, whatever. I’m not planning to buy anything from this sort of company.
Sounds like you already did…?
Either way, problem solved. The drivers are now open. I don’t see what the problem is, other than your bad timing…
OMG, I have been waiting for this to happen for what seems like an eternity–the better part of the last decade in fact. After this, the only thing more shocking I could imagine coming to the OSnews frontpage would be:
a) nVidia getting something decent open source out the door, and
b) News that a supreme diety, aka “the God” (whatever that is), was going to release a new mobile phone OS to better compete with other religions in an advertising “holy war”. If the UNIX wars were a form of holy war, which they were for some folks, then surely we need to have divinely inspired OSes, right? 😉
Note: Sarcasm and extreme smirking went on while I wrote that. I hope it tickles someone else. If it bothers you, then “move along, these aren’t the droids you are looking for.”
Edited 2010-09-13 04:19 UTC