Linux Kernel Devs Offer Free Driver Development

schwaang writes "Linux Kernel hacker Greg Kroah-Hartman, author of Linux Kernel in a Nutshell has posted an epic announcement on his blog. This could portend increased device compatibility for Linux users, higher-quality drivers, and fewer non-free binary blobs." From the announcement: "[T]he Linux kernel community is offering all companies free Linux driver development... All that is needed is some kind of specification that describes how your device works, or the email address of an engineer that is willing to answer questions every once in a while. If your company is worried about NDA issues surrounding your device's specifications, we have arranged a program... in order to properly assure that all needed NDA requirements are fulfilled. Now your developers will have more time to work on drivers for all of the other operating systems out there, and you can add 'supported on Linux' to your product's marketing material."

How many

I wonder how many companies will be imprudent/progressive enough to take up this offer.

Re:How many

I would disagree. Linux drivers are not made since they do not generate profit, largely due to the small user base verses the cost of developing the driver. If there is but a modest cost of a dev answering a few questions, it may be worth their while if it means shipping another 200 widgets.

Re:How many

The key driver (pardon the pun) will not be how many extra widgets they sell but the the strategic importance to most companies of reducing reliance upon Microsoft's hegemony. If you are a widget developer, you do not want to be in the position of most, dancing to the unrestricted tune of Microsoft. You need some collective force to help push back on Microsoft when necessary, or to demonstrate the worth of new ideas which Microsoft may not have picked up on. Having a competitor to Microsoft (even quite a small one) is a massively powerful force in this.

Re:How many

On the other hand, working with a competitor to Microsoft may not be a wise strategic decision.

"Hi, good to see you again. I heard about how you were working with those Linux guys on giving away free drivers for your new card. That's a great move. But that has nothing to do with what I wanted to tell you, which was that there was an accident in the Vista certification lab and we lost the drivers you sent us. Until we can get a fix pushed out for this that means that everybody who buys your product from now on will be told that it has absolutely no support and that even if they download something directly from you it will be flagged as foreign code and won't run. The guys in the lab are really broken up about it and can't figure out how that kind of mistake could happen. Don't worry though, we'll get everything straightened out in the next big service pack. Honest."

Extortion

> That's called extortion, and it's illegal.

Like that ever stopped Microsoft before.

> I would hope the DoJ steps in at that point.

Like that ever stopped Microsoft before.

Re:How many

That misses the point. A very large number of Linux users are in the position where they are consulted by others about what devices to buy. The availability of devices which will work with Linux increases a company's exposure to all kinds of user, not just Linux users.

Companies worried about IP issues should ask themselves if they are in the hardware business or the software business. If their objective is to sell more gizzmos, then opening the API to developers is an excellent way to sell more product.

If a company is concerned about the number of questions they'll be asked by the developers, then (a) they don't know the software business, and (b) they should take a long, hard look at the quality of their documentation.

The biggest problem is that many companies are already making so much from selling their gizzmos to Windows users not to need to sell them to Mac or Linux users as well, even though it takes no significant effort to do so. The extra profit, even at virtually 100% per unit) simply isn't attractive.

Users need software freedom for all their software

People who fall into the trap of installing and using proprietary video card drivers then later discover that their video card (which still works fine) is no longer "supported" by the latest driver update would disagree with you that "Video cards are already well-supported by their manufacturers.".

I believe this kind of thing happens [digitalcitizen.info] more than others know, particularly as GNU/Linux distributors that distribute proprietary software make it easier for users to acquire proprietary software (as I understand Ubuntu is working on). Users shouldn't be left without their software freedom, nor should they have to choose between updating their system kernel and using their video card.

Making users helpless and keeping them separate is no way to live. Users need software freedom now.

Re:How many

Nope. I would still rather have those cycles for computing
things I want done rather than supporting some lame hardware
vendor's attempt to save 5 cents on some bit of hardware.

Intellegence in peripherals should be expanding outwards
rather than shrinking. The former aids parallelism and the
latter sabotages it.

Re:How many

Except the sorts of things that rely on the host PC for all/most of their processing (Winprinters/Winmodems), aren't limited by the host PC's CPU. If you've got a Winmodem, doubling the host PCs CPU speed doesn't double performance. If you've got a Winprinter that does 8 PPM, getting a faster host PC CPU doesn't mean that you'll start getting 10 PPM as printer performance is limited by how fast you can physically feed paper or how fast you can get the print head to traverse the page.

If a peripheral is taxing the host CPU enough that upgrading the CPU will increase the performance on that peripheral, it's already taking up too much of the host CPUs time.

Re:With few notable exceptions...

"Am I missing something, I don't see your point?"

Apparently not. What PP said was that there is the potential for profit there, but the hardware industry may be underestimating the buying power of the Linux desktop market.

For example, I, and many other Linux users buy my WiFi based on what works in Linux. I am not a Linux-only human; I have a Mac, a Windell at work, and a Ubuntu laptop.

But therein lay the point: The commercial OS developers already support your product; you wrote drivers for them. Now, at no extra cost, you can have the edge up on your competition for that 1.6% of the computer-using market that has Linux in one form or another. As of 2004, that's 1.6% of 61.8% of all people in the US, or about two-and-a-half million people. At median income of $30k, and assuming that 5% of their purchases are technology-oriented, that's a chunk of $3.75B you're fighting for.

Not a small number, and the slice of it you get could mean the difference between shelling out a new product in a month or shelling it out in a year. Meanwhile, if you can expand that slice at low (rather than building a monolithic driver, build drivers as modules that can be 'plugged in' to existing kernel scaffolding for Linux, Apple/Mach and NT) or no (having the Linux Devs build it) cost, there's no reason you shouldn't.

Of course, the bigger the company, the less this matters to them; large companies have the opportunity to 'rest on their laurels', as it were, when it comes to new accessibility.

Linux a Viable Alternative?

if Linux is going to become a viable alternative for the girlfriend

Dude, don't get me wrong, I think Linux is sexy. But not that sexy.

seems like a good idea

Seems like a good idea, but it also seems like it would give the device manufacturers an out. "I'm sorry, but we don't officially support the linux operating system". This way they get drivers written for them for free, and don't need to provide any tech support for the device to those users who purchase it for linux. Anyone else see this happening?

Re:seems like a good idea

No, those drivers would most likely be written by the community anyways.

To me it seems more like an initiative to figure out which companies use "we don't have the staff/resources for an open driver" to keep their drivers proprietary.

Re:seems like a good idea

Another way of looking at it would be as formalising the rule that "if you give us specifications, the driver will get written". A lot of the problems with free software drivers is lack of information on how a device works; if this makes it better known that all they have to do is provide some specification, it might encourage companies to submit more of them, and encourage customers to ask people to submit more of them.

Re:seems like a good idea

You'll need a big cluebat to get a lot of these companies to wise up. The real reason that they don't give out programming interface information is because they're listening to a lot of IP Lawyers that tell them they have to keep everything secret or it might affect future patentability of future devices (YES, I've seen that A LOT lately, doing Linux driver consulting for some of the crowd willing to do proprietary driver work...), etc.

It's a mixture of worries about revealing possible Patent infringements, trying to slavishly follow the lawyer's advice, and a confusion as to what business they are precisely in (Software versus hardware- a lot of companies, because of the advice of their IP lawyers are confused as to what they should be doing...).

Re:seems like a good idea

It might give the device manufacturers an out but, more importantly, won't it equally give the Linux family an 'in'?

The point isn't, so far as I can see, to make any profit from the scheme other than to spread the word of Linux and increase the potential compatibilities/reduce the incompatibilities.

Plus, as a bonus for the device driver writers, it's an impressive CV when you consider the varieties of hardware that are supported by the various Linux distros and the work, and potential elegance, that goes into solving the various demands.

It seems win-win for everyone, really. And a good, and generous, idea.

Re:seems like a good idea

The point isn't, so far as I can see, to make any profit from the scheme other than to spread the word of Linux and increase the potential compatibilities/reduce the incompatibilities.

In fact, this is how it's always worked --- people have been asking companies for device information for years. (I did, once; I wanted the specs for a SIM reader device so I could do a Linux driver. Did I get a response? Did I hell.) The only difference is that this announcement rephrases things in a rather more marketspeak and official manner. Instead of the companies doing us a favour, by providing hardware specs, we are now doing them a favour, by writing their drivers for them.

It's a rather neat bit of lateral thinking.

Re:seems like a good idea

In fact, this is how it's always worked...

Exactly! I think Greg made a tongue-in-cheek post on his blog and the submitter and /. editor unwittingly took it at face value (no big surprise there). It will be wonderful/funny if some corporate shills also take the blog post at face value and actually start feeding the kernel developers the information they need.

Re:

This way they get drivers written for them for free, and don't need to provide any tech support for the device to those users who purchase it for linux.

I'd say that's pretty much the state of play for the majority of Linux drivers anyway. To be fair I h

Re:

As far as printers go, HP has a driver out now thats pretty slick (HPLIP). It doesn't do everything, but it does most of the personal printers. (Not to mention they make damn good printers).

http://hplip.sourceforge.net/ [sourceforge.net]

Re:seems like a good idea

I've decided that HP pisses me off. They go out of their way to make sure you're not refilling ink carts (and in some cases toner) and frankly I don't think their printers are all that. Not to mention that they've all but done away with duplex modules, now you have to buy duplex printers. So if you bought a non-duplex printer and you need duplex, now you have to buy a whole fucking printer! I want a company that gives me choice, not a company that scares me into buying the full-featured product by not offering me a way to add the functionality later. And they're straight up lying if they claim that no one buys those things. Anecdotally, I've got a Laserjet 5550n behind me and to the left for which we would buy a duplex module if it were offered. Since it is not, I have to do it manually, and since I'm lazy, that makes me a sad panda. :(

Hardware ?

"All that is needed is some kind of specification that describes how your device works". They also need some real hardware to test the brand new written drivers. Specifications are not enough. Who will test the real hardware with the fresh drivers in a real-world operation ?

Re:Hardware ?

from tfa

A few sample devices might be good to have so that debugging doesn't have to be done by email, but if necessary, that can be done.

Re:Hardware ?

Well, why not us beta testers?

Ballmer: "We refer to them as customers."

/try the veal

Dedicated

Whatever you might say about the Linux community - that it is elitist or sanctimonious or whatever - it is impossible to ignore their commitment to what they believe in. That somebody would be willing to write device drivers for nothing, apparently just to forward the cause of a free operating system, is pretty impressive. Microsoft and Apple can match this devotion only in the ferocity with which they defend their control over their customers, in anti-trust trials and by imposing DRM.

Peter

Quick Scan

A really quick scan [google.ca] of the price of windows driver development, demonstrates how much actual value this is for business. Now all you would need to do is pay someone to extend the drivers to other platforms! Eureka!

Example of extending to other platform

As an example of extending to other platforms, we may cite the 3DFx Voodoo board.

After the company collapsed, users were left with no drivers for recent windows version (XP, XP64 and Vista).
But, the Linux drivers happened to be open source.

So most of the work you may see on websites like http://3dfxzone.it/ [3dfxzone.it] for Windows, is mostly based on libglide and Mesa3d for linux.
(This is also another proof that open-source enable something to survive beyond the death of it's parent company)

Another example may be the linux USB stack, which was later ported to both the Cromwell xbox bios and ReactOS (opensource clone of the Windows NT system, cousin of Wine project).

This is needed

I think this kind of action, and offer for help is needed by companies. I hope it will be touted enough. What I know is that, companies having really hard times finding skilled coders for developing Linux drivers. Most of them does not care about the specifications, as they have already patents pending for their works, but they can't actually find people to code for Linux and/or they don't willing to pay more than Windows developers for Linux developers for a smaller market.

Wonderful

What an outstanding idea! I especially like this (from TFA):

If your company is worried about NDA issues surrounding your device's specifications, we have arranged a program with OSDL/TLF's Tech Board to provide the legal framework where a company can interact with a member of the kernel community in order to properly assure that all needed NDA requirements are fulfilled.

This is intelligent, it means they're covering their backs, and even more importantly the manufacturers haven't got an excuse!

Is this realistic though? Are companies actually going to take this offer up? If they do, the impact could be awesome (hardware compatibility that could rival Windows and/or Mac OSX)...

Nice one!

Re:Wonderful

(hardware compatibility that could rival Windows and/or Mac OSX)

Hmmm? Linux already supports more hardware out of the box than Windows does. I'm not talking ancient SCSI cards either; I mean components like an onboard Intel PRO 10/100 NIC from a few years ago that requires an extra driver on XP SP2, but works automagically with e100 on Linux. The only segment where Linux falls down is on very new hardware.

Re:Wonderful

Hmmm? Linux already supports more hardware out of the box than Windows does. I'm not talking ancient SCSI cards either; I mean components like an onboard Intel PRO 10/100 NIC from a few years ago that requires an extra driver on XP SP2, but works automagically with e100 on Linux. The only segment where Linux falls down is on very new hardware.

I am sure this is the case statistically, but anecdotally I've noticed that things I don't expect to "just work" (TM) with linux oftentimes do. My girlfriend's digital camera working without a driver install really sort of surprised me as I expected it not to work at all. The same camera often requires driver installs on windows (differing per version). I'm actually quite impressed with Linux's hardware support.

Re:Wonderful

http://www.sane-project.org/sane-mfgs.html [sane-project.org]

Re:Wonderful

Except Linux has supported _more_ devices than any version of Microsoft-windows for some time now. Okay so most of those drivers are for older hardware that is no longer supported by new versions of Microsoft-windows... but that doesn't change the facts. You need to qualify your statement, and say what you mean. I guess something like "Microsoft Windows gets support for some new devices more quickly than Linux"... thats about it. I am not even sure there is any truth to OSX supporting more of anything than Linux, Apple-mac hardware is all the same after all.

Infact Linux supports more devices that any other operating system ever... and thats one of the advantages of open-source kernel drivers... they are maintained with the Kernel, so they remain usable through kernel architecture changes with zero effort from the original contributer of the device-driver. I am sure Microsoft would love to do this with windows, but of course they cannot, as they don't have the source code to the drivers they did not write themselves.

Re:Wonderful

No-one seems to have commented on the fact that if NDA requirements are met the drivers cannot be open source. This doesn't mean fewer binary blobs, it means more.

And what about Vista's new requirement that all hardware mustn't be compromised by hackers or else the drivers will be remotely disabled? Might a company which produced hardware which is part of the DRM stack risk being more likely to be seen as compromised if it has collaborated with the OSS community?

Re:Wonderful

No-one seems to have commented on the fact that if NDA requirements are met the drivers cannot be open source. This doesn't mean fewer binary blobs, it means more.

I'm far from being an expert but I've got the impression that for example the open source 3D accelerating drivers for ATI's R200 series were written under an NDA. It really depends on the NDA. The specs can reveal things that aren't immediately apparent from the code, and the NDA may be written to protect those parts. While some companies would probably require NDAs that would effectively prevent an open source driver from being written, it doesn't automatically have to be so.

Re:Wonderful

This reminds me of Theo de Raadt's letter to the OLPC project [theaimsgroup.com]. What good is code that contains an array of bytes consisting of basically pre-compiled source code? What happens when a bug is discovered which crashes your system? How do you go about fixing those bytes if the person who wrote it and was under NDA is no longer available?
 

Re:Wonderful

No-one seems to have commented on the fact that if NDA requirements are met the drivers cannot be open source. This doesn't mean fewer binary blobs, it means more.

Open Source drivers have been written under NDA before. What this typically means is that they are write-only code. The NDA will prevent things like properly labelling constants and helpful comments, so you end up with code full of magic constants and seemingly random operations. It's basically impossible for anyone to maintain without the NDA'd documentation, so you are pretty much screwed if you want to port it to another OS or maintain it when the original author gets bored or dies.

On the plus side, a badly written driver is marginally easier to reverse-engineer than a black box. There are a couple of WiFi chips that had Linux drivers written in this way, and now have much better OpenBSD drivers written by a combination of reverse engineering and examining the Linux code. Some of these drivers have been ported to FreeBSD and, I believe, back to Linux again.

Re:Wonderful

No-one seems to have commented on the fact that if NDA requirements are met the drivers cannot be open source. This doesn't mean fewer binary blobs, it means more.

Open Source drivers have been written under NDA before. What this typically means is that they are write-only code. The NDA will prevent things like properly labelling constants and helpful comments, so you end up with code full of magic constants and seemingly random operations. It's basically impossible for anyone to maintain without the NDA'd documentation, so you are pretty much screwed if you want to port it to another OS or maintain it when the original author gets bored or dies.

Some NDAs require that, true; the resulting drivers look a lot like the "nv" driver for X, which does indeed look like write-only code. (And as you suggested, it did still help the authors of nouveau [freedesktop.org].)

However, in many cases a hardware company NDA just requires non-disclosure of the hardware documentation itself, and in particular the documentation of the hardware's internal workings. In these cases, the resulting driver generally looks like most other drivers in Linux, including useful constants and helpful comments.

(Not commenting on the ethics of NDAs in general; just presenting information.)

Standard Driver Model?

I have never written drivers so I may be way out in left field here, but how close are we to being able to specify a standard driver model, with compatibility across operating systems? It seems to me that drivers are one of the biggest impediments to OS adoption. They are also a huge cost center for device manufacturers. Imagine if 99.9% of the driver code could be the same across platforms. Is this even remotely possible? Or perhaps the Linux Kernel driver developers could figure out a way to adapt Windows drivers to run, perhaps in an interpreted or emulated fashion, on Linux (ala Virtual PC). Just a thought.

Re:Standard Driver Model?

You've just invented OpenFirmware [wikipedia.org].

The only small problem is that it requires slightly more intelligence (and some flash memory) in the individual device - something which manufacturers have spent the last 20 years doing their best to avoid.

Re:Standard Driver Model?

Not really. No one actually uses OpenFirmware drivers much past boot time if they can avoid it. A better example would be I2O [wikipedia.org], which proposed a split driver model. Half of the driver would be hardware-specific, and half would be OS-specific. For a graphics card, for example, the OS would load a hardware driver that would translate something like OpenGL into device commands, and an OS-specific driver that would translate whatever the native graphics API was into generic graphics API calls.

Xen implements something like this for block and network devices, and the USB and Bluetooth specifications do something similar for a few categories of device. The problem comes with things like GPUs where each new generation provides some extra functionality that the last one didn't; you'd need to constantly update your driver model to work with the new functionality. It's not impossible, but it does require a standards body that can quickly specify interfaces to the new functionality, which is quite improbably.

Re:Standard Driver Model?

Well, there's paravirtualization. The drivers go in the hypervisor, which then provides a simplified and unified interface to guest OSes. The guest OSes still have to implement drivers for the exokernel, but there are a lot fewer of those than there are, say, Ethernet cards.

Re:Standard Driver Model?

There are huge advantages to going the driver route rather than wrappers for windows drivers. For a start, the community can update any open drivers, whereas they cannot touch closed windows drivers. There are many to whom open drivers over closed ones is a big deal.

Nothing new?

Other than the public announcement, how is this any different from the way things already work?

The community already writes free drivers for vendors who provide specs and (even better in some cases) loan some hardware.

There are already situations where Linux devs have been able to work out NDA-acceptable solutions.

Really, all the announcement is saying is, "Look, we do this. We've been doing this for years. Just letting you know how things work over here."

This is definitley new

Other than the public announcement, how is this any different from the way things already work?

Actually this really is something new, and quite an announcement. It was never the case before that any old random driver would get created by the open source community. The way OSS development generally works is there has to be a strong need, strong backing, or a high fun factor, for things to get done.

Prior to this announcement it's not like there was a group of people dedicated to writing drivers -- just waiting for companies to release new hardware, then they'd scurry to reverse engineer it and write a driver. Nor do companies (generally) release hardware specs in the hopes that others will provide a driver for their product.

A significant portion of initial open source driver development comes from the device manufacturers themselves, and smaller companies without the resources to spearhead these developments simply don't have the ability to have Linux support.

Your conception that "The community already writes free drivers for vendors who provide specs and loan some hardware" isn't true in the vast majority of cases.

This really is a big change, because now anyone can create a hardware device and actually have formal linux support, and have this printed on the box. This creates a formal avenue for companies to easily, reliably, and cheaply have Linux support for their products.

How will the NDA work ?

The Kernel code will be publically visible, so how is ''confidentiality'' maintained ? The only ways that I can think that this will be done are:

1. Uncommented Kernel code - Yuck!
2. Spaghetti/obscured Kernel code - Yuck!
3. Binary blobs in the Kernel - Yuck!

Re:How will the NDA work ?

Ranked in order of preference:

a) no driver for your hardware
b) binary blob kernel patch created by hardware munfacturuers
c) binary blob in kernel tree created under NDA by the kernel team (who have private access to the source)
d) obfuscated code in the kernel tree (with original kept private to those kernel devs that have signed the NDA)
e) uncommented code in the kernel tree (with commented code kept private to those kernel devs that have signed the NDA)
f) fully open source driver

Personally I'll accept anything b or above - I'd prefer d or above, would settles for c but would really like f!!

I wonder where the compromises will be made? How far will kernel devs go? How far will companies go?

Re:How will the NDA work ?

b) binary blob kernel patch created by hardware munfacturuers

Widely believed to be a license violation.

c) binary blob in kernel tree created under NDA by the kernel team (who have private access to the source)

Almost certainly a license violation. (Can't be distributed with the portions of the kernel written by others who have released their code as GPL)

d) obfuscated code in