Free Linux Kernel Driver Development FAQ

schwaang writes "The recent announcement by Linux Kernel Developer Greg Kroah-Hartman that 'the Linux kernel community is offering all companies free Linux driver development' seems to have stirred up some interest as well as some questions — see the Slashdot discussion about the announcement. Greg K-H addresses some of the questions raised here, and raises a few more, in a new Free Linux Driver Development FAQ on his blog. An excerpt: 'Q: Are companies really going to do this? A: Yes, already we have received a number of serious queries from companies about producing Linux drivers for their devices. More information will be available later when details are firmed up."

Comments

Sweet

[Stormx2]

This makes an awful lot of sense. Currently, a lot of hardware gets drivers but without the assistance of the companies. This way, consumers benefit (if you can call us basement-bound linuxers consumers) and the companies benefit by reaching more people.

Re:Sweet

[Jekler]

Linux users are definitely consumers. Believing that having the basic functionality of operating your computer for free isn't about exiting the market as a consumer. It's along the lines of how you should be able to operate your television for free. You don't need cable, satellite, or any product separate from the equipment itself for your TV to function. You can buy a TV, watch DVDs, tapes, home videos, set it up as a security monitor, or use it just as speakers for an audio tape player, whatever you want to do with it.

Up until Linux, if you bought computer hardware, it was a dead box unless you also paid for an operating system, of which the only real choice was Windows. You couldn't use the hardware you had bought. You had a computer that couldn't compute anything. No matter how small your needs were, even if the only thing you ever needed to do was create plain text files, you had to pay $90+. With Windows XP Professional, it should burn any purchasing manager to think about paying $300 for a pinball game and another half dozen games, movie maker, media player, and various libraries and functions to support those applications, when the reason the computer is needed may not ever involve any of those. Whether or not those applications are needed you must pay for them. Enter Linux.

It's a misnomer that Linux users are cheapskates. Linux users will pay just as much cold, hard cash as the next guy for applications and products that fill his needs, they're just not willing to pay for peripheral garbage that has no value to them. As a Linux user, I've personally paid $4500 for an IDE/toolkit.

There are many Linux users who want to be consumers, and would gladly pay for things like 3D Studio Max, Photoshop CS2, etc. but those things aren't being offered to Linux users. Linux users are consumers lacking producers. We might have money but, for some reason, a lot of companies don't think a Linux user's money spends like the other kind, which is a shame because companies like Google have a lot of Linux money to throw around.

Basement-bound Linux users are no different, eventually they become purchasing managers, company owners, or hobbyists who build up some cash and want to spend it on something. If someone makes a decent offer, they'll fork over the cash just like anyone else.

Can users sign up for this?

[Excelcia]

Can users sign up for this? I've got some hardware I'd love Linux to support. :p

Re:Somebody please explain what this all is about

[Excelcia]

You and I both should have read the FAQ before posting. ;)

Q: This is a lame publicity stunt, Linux development has always been done this way.
A: Well, the NDA program that we have set up with The Linux Foundation is new. But yes, other than that, this is exactly how Linux kernel development has been done. But it is good to point out exactly how it all works for those who are not familiar with how it works. (emphasis added)

Sure you can

[codepunk]

Point your hardware vendor to the web site and ask them to participate. If you mean will someone reverse engineer your hardware, well they answer that question on the site FAQ and the answer is no.

Re:Somebody pleasexplain what this all is about

[BigBuckHunter]

"We'll write free drivers for your hardware if only you release the specification" - isn't this the same that was always done in Linux kernel? Or is the issue only about signing the NDA to get the specs?

From the FAQ

Q: This is a lame publicity stunt, Linux development has always been done this way.
A: Well, the NDA program that we have set up with The Linux Foundation is new. But yes, other than that, this is exactly how Linux kernel development has been done. But it is good to point out exactly how it all works for those who are not familiar with how it works.

BBH

Could've solved broadcom

[g4sy]

Yes, this is not a lot different than the way device drivers have always been incorporated into the kernel. But having a willingness to work with NDAs of various companies MIGHT have solved the whole fiasco with Broadcom wireless chipsets (if you didn't think it was a fiasco, you didn't buy a iBook G4 the day they were released, May 2004, only to find out that you would be unable to use wireless on it for the next 2 years at least).
I don't know but I think that maybe such a system might have made the suits and lawyers with Broadcom comfortable enough to allow co-operation on a linux device driver... *sigh* would have been nice.

Cirrus Logic doesn't appreciate help

[ClaesMogren]

Unfortunately a lot of companies seem totally unable to see the benefits of external contributors. They don't even see the point in getting drivers into the mainline kernel. Just take a look at this response from Cirrus Logic regarding their ep93xx boards:

http://www.freelists.org/archives/linux-cirrus/02- 2007/msg00026.html/

Looks like the in house coding team was bummed that Lennert Buytenhek did a better job on the port then their whole team. Ridiculous response! /C.M

Re:What we really need

[oneandoneis2]

Most of the online guides for the few bits of hardware that do exactly what you suggest tell you to ignore the drivers the hardware comes with, as a newer kernel will almost certainly be out by now with an updated driver. The last thing we should do is try to get them to supply the drivers with the hardware.

What we want - and what this process does - is to get them to release enough information that a driver can be written and incorporated upstream into the kernel so that Linux supports their hardware out-of-the-box. This bypasses all the "critical mass" problems because they don't have to pay for the development costs, and negates the need to supply drivers with the hardware. How can they lose?

Re:how do these developers get paid?

[TheRaven64]

Many of them are paid by distributions. Novell and RedHat, for example, make money buy selling support for Linux. The more hardware their distribution supports, the easier it is to sell. Getting good drivers in the upstream tree is the easiest way of doing this, so they pay kernel developers. If you are able to write kernel code and interested in doing it for a living, I would suggest you send your CV off to the big distros (and possibly IBM) and ask if they have any openings.

You may also be able to get some short-term work from companies wanting to switch existing infrastructure to Linux and needing drivers for existing hardware, although this is likely to be contingent on your acquiring the device specs first.

Not entirely true either

[WindBourne]

For starters, the norm to starting this process is that the hardware is out and somebody has an itch. Then they have to go to the hassle of getting the manufacturers specs, etc.

Now, some manufacturer will be approaching the kernel team and offering the specs. The kernel team will probably pick an active developer who wishes to do it. Interestingly, manufacturers will be more likely to bring in alpha (or beta) hardware to have the drivers built BEFORE going to market. Once they figure out the sales potential from Linux, then they will be more likely to develop the drivers in-house.

Yes it's always done like that mostly, but....

[vdboor]

isn't this the same that was always done in Linux kernel?

Yes, but... Who told the hardware vendors about that? :-p

Someone finally did, explained the benefits, and got an amazing number of responses :-)

We take much for granted. When you meet a Linux newbie you'll notice how much "hidden knowledge" we have. Who the community is, that the FSF / GPL is, how the OS is layered in tools and front ends, what "compiling" does, how communication is done, how to find answers for problems. Linux newbies are not aware of this. The same can be said about hardware vendors.

Even if a vendor jumped in a random channel, the average response is "Open Source it". We understand the meaning and advantages of that approach. They only think "help, I must give away my code". It was about time someone stepped up to shed some light on these matters.

Re:liability?

[Frequently_Asked_Ans]

11. because the program is licensed free of charge, there is no warranty for the program, to the extent permitted by applicable law. except when otherwise stated in writing the copyright holders and/or other parties provide the program "as is" without warranty of any kind, either expressed or implied, including, but not limited to, the implied warranties of merchantability and fitness for a particular purpose. the entire risk as to the quality and performance of the program is with you. should the program prove defective, you assume the cost of all necessary servicing, repair or correction.

heard of the GPL? http://www.gnu.org/copyleft/gpl.html

Re:What we really need

[0x0000]

Linux explicitly doesn't allow that. With Linux it is all or nothing, and many hardware manufacturers choose the predictable 'nothing' instead of the less predictable 'all'.

You're kidding yourself - I have used, and continue to use - a number of closed source binary Linux device drivers acquired from 3rd party manufacturers. The real reasons you don't see Linux device drivers shipping with hardware are:

1. The manufacturers typically lack the software skills to create a linux device driver in-house, and don't feel that sales to Linux users will comprise sufficient dollar amount sales to justify paying a developer to create one.
2. Plain Old Ingnorance (POI) due to not having anyone in a position of influence at the company with sufficient knowledge of things like GPL licensing and non-Windows systems to know just how full of shit remarks like the one you make above really are.
3. All the usual hate, religion, and bribery that are brought into play when anyone in the industry so much as thinks about doing something that is not Windows.

It's worth noting that in many cases Microsoft produces or buys drivers for hardware. thereby obviating the [percieved] need for the manufacturer to spend much effort on any OS drivers, let alone one as arcane as a *nix driver with some hippie "licensing" scheme...

Also, if a device is designed to an existing h/w spec utilitized by M$, again - no driver needs to be produced by the h/w manufacturer.

It's all about margins, market share, ignorance, and prejudice. The relative openness of the Linux systems has nothing to do with it, nor does your imagined inability of Systems other than "Solaris and Windows" to dynamically link a loaded binary module. Futhermore, I am unaware of any consumer-grade h/w device which has Solaris OS driver support, which does not also have support under Linux.

The fact that you and/or your shop have never so much as looked at the Linux OS to a degree sufficient to producing a device driver for it is obvious from your posting, so please: Sit down, and STFU until such time as you have poked around a bit and actually know someting about what you're talking about - you've forced me to waste an unacceptable amount of bitwidth trying to clear the smoke you've blowing in front of the mirrors...

Re:Zero

[chill]

It'd be funnier if an unmeasurable number of new Linux drivers were created. "An infinite number of Linux drivers were created this week."

If only we had an infinite number of monkeys...

(Gotta love preview. I just noticed the original article is on the Linux Kernel Monkey Log. Maybe we DO have more monkeys than I realize.)

Re:liability?

[chill]

I think the real point in those situations is "if it goes wrong I can put the blame on a big company the PHB has heard of, otherwise it will be my fault".

Guess what. Odds are, it'll still be your fault. Your fault for not getting the specs right. Your fault for not working with the major vendor to make it work. After all, they're a big company and have hundreds if not thousands of other installations working right. Thus, if they all work and yours doesn't -- it is your fault.

The ONLY way you might get away with that is if some executive MANDATED you use a specific product, overriding your objections or advise to the contrary, and he is a known asshole in the company. Even then, it is still iffy.

And finally, even if it isn't your fault and you can successfully blame someone else, you'll still get a bit of a reputation of "that guy who couldn't get it done".

Good luck!