The driver source code released for the Intel 965 Graphics chipset recently isn’t as open as first thought. Keith Packard posted a reply on the linux-kernel mailing list detailing what parts were not available.
The driver source code released for the Intel 965 Graphics chipset recently isn’t as open as first thought. Keith Packard posted a reply on the linux-kernel mailing list detailing what parts were not available.
EDIT : whoops, this is submitted by Binarycrusader — so Kudos Binarycrusader!
Nice journalism job — it’s really great to be notified of such important postings in developer mailing lists, as it’s a tedious job to read them all.
Back to the topic, I still have high hopes for this free Intel driver, though the binary blob is disappointing. After all, that’s still much better than what ATI and Nvidia are currently doing. I guess Intel is bound by some nasty NDAs, but with people like Keith working there, I am still quite confident in their good will.
Edited 2006-08-13 08:56
Nice journalism job — it’s really great to be notified of such important postings in developer mailing lists, as it’s a tedious job to read them all.
The credit should go to the OSNews crew, who have provided a relatively focused technologly site that keeps the community informed.
Also, I probably should have said “are not available” instead of “were not available.” I am certain that one of my teachers from long ago is suddenly annoyed with me and doesn’t know why. I am also certain there is a comma missing in my text as well somewhere
See this post for a few more hints about what the binary blob may contain:
http://marc.theaimsgroup.com/?l=linux-kernel&m=115534356117835&w=2
Edited 2006-08-13 09:29
So bad if they produced these ‘binary blobs’ for more than just Linux; there are those of us who run FreeBSD, who would love to see a decent level of Xorg support for more than just Linux.
So Intel gives us this open source driver, issues the kind of LKML post that is guaranteed to hit all of the Linux-related news sites, and fails to mention that this doesn’t actually mean that we’d be able to program the hardware.
We have to use a binary hardware abstraction layer or miss out on the highly-touted advanced features. I especially like the part where Keith says that the new driver, sans blob, will have all of the features of the previous driver versions… but presumably not many of the new features. Thanks guys!
This whole trade secrets thing has to stop. Patented technology is, by definition, not a trade secret. The only reason why Intel would have trade secrets is if they can’t patent them, and the best explanation for that is because they are already patented by somebody else.
That’s right folks, we don’t get to see the interface specs for Intel’s graphics hardware because they’re infringing other people’s patents. Two wrongs make a right?
From TFA:
It’s optional, so if you don’t want to use a binary module, you don’t get to use code written by Intel agents for these features.
We can’t exactly fast-forward to a software-patentless world. It seems that this is an important step in the right direction. When manufacturers see that there is indeed market advantage to getting the code out there in plain sight, there will be progress.
Let’s include a little carrot with the stick.
———-This whole trade secrets thing has to stop. Patented technology is, by definition, not a trade secret.————–
As far as I can tell, what wasn’t open sourced was macrovision,(intel can’t give away somebody else’s technology) DRM bits,(it seems that other posters have also noticed this) and driver optimizations.
———-The only reason why Intel would have trade secrets is if they can’t patent them, and the best explanation for that is because they are already patented by somebody else.————–
That’s a mighty big leap you’ve made there.
AFAIK, driver optimizations are not patentable. It’s one thing to patent “S3TC”, “detonator”, “catalyst”, “digital vibrance control”, “HyperZ”, etc etc etc. I’m not a lawyer, but it seems that if (x company) can re-arrange their driver structure to get better performance in a handful of games or even across the board that it’s not really patentable. But it isn’t something that you’re going to want competitors to know either way.
Maybe they should remove the macrovision garbage, it sucks anyways.
Intel is big enough they could easily get away with it.
Right. Designs are usually patentable, implementations are usually not patentable. They are instead protected by copyright, and the GPL doesn’t really offer much protection.
However, the “big leap” I made corresponds to this idea of “proprietary standards” that has dominated the consumer entertainment space. They’re patented technologies on which the industry has standardized, and they require a proprietary license in order to implement them.
Intel can’t release the code, and neither can they patent the technology. It’s already patented by somebody else. That statement is correct.
The following statement, where I insinuated patent infringement, is not. I should have instead said that Intel’s code is _encumbered_ with other people’s patents. My bad…
———-Designs are usually patentable, implementations are usually not patentable. They are instead protected by copyright, and the GPL doesn’t really offer much protection.————-
You may be right. I’m not sure.
But considering that only Intel makes intel chips, that specific chip design being unique and different from Nvidia’s, Ati’s, and etc chips it would be rather useless to patent a driver design/implementation that doesn’t apply to another type of chip design/implementation.
However, Intel doesn’t want to give pointers to it’s competition. There could be other reasons as well. You may be right, I’m not saying you’re 100% wrong. We could both be right to a degree.
————-However, the “big leap” I made corresponds to this idea of “proprietary standards” that has dominated the consumer entertainment space.———
If you paid attention a little more, you’d see that it dominates more than just the CE space. People are generally big on ownership. That’s mine. From our homes and land, to our children and spouses, and yes, even the things that we invent. That’s mine. It’s everywhere.
And most of us expect to get proper compensation. Now if some want to give it away, that’s their freedom. But if they don’t want to give it away, it’s a little preposterous to automatically think the worst, to think that it’s because they’ve already committed the crime.
“Innocent until proven guilty”, remember?
————-Intel can’t release the code, and neither can they patent the technology. It’s already patented by somebody else. That statement is correct.———-
It’s not. It can’t be. Why? You’re jumping from one extreme to the other.
You don’t know what’s in the middle any more than I do.
Actually, we probably don’t get to see the code for the blob because Intel has licensed the code from somebody else, and is not allowed to distribute that code, they aren’t infringing, they are abiding by thier contractual agreements to other companies. Loosen up the tinfoil hat, there is no conspiracy here
Actually, we probably don’t get to see the code for the blob because Intel has licensed the code from somebody else, and is not allowed to distribute that code, they aren’t infringing, they are abiding by thier contractual agreements to other companies. Loosen up the tinfoil hat, there is no conspiracy here
Thank you. That’s a point that seems lost on many people here. If Intel violated the GPL to distribute code in an unlicensed manner, people would scream bloody murder. Yet many of these same people are intolerant of the fact that Intel is respecting license terms with third parties. Can’t have your cake and eat it to, if you expect people to respect the GPL, you have to expect them to respect all licensing terms they agree to.
DRM is here to stay, at least for the near term. And media conglomerate DRM methods will never ever be compatible with OSS licenses. There’s no way around it.
At the very least, Intel has taken steps to seperate the DRM functionality and their legal requirements from the driver’s core functionality. That’s more than any of the other vendors have done.
The free software zealots are supposed to be morally opposed to DRM anyways, so there’s no issue there. The GPL-bound distros, commercial and community, and their users, will ultimately have access to a compatible license that can ship with the code, so no issue there. And Joe Average, who likely doesn’t care, will have likely have a simple method for downloading the blob and having full DRM-entrenched functionality, so pretty much status quo there.
I’m cautiously optimistic that Intel may have managed to trump nvidia in finding a way to support the OSS community while still managing to balance real-world business requirements and concerns.
Edited 2006-08-13 16:14
Although I agree with you, just one silly question from my side:
What if Intel licensed some technology (hardware/software) from someone else under certain contracts (like not open sourcing the software part), no patents involved?
One way or another, we’ll have a lot less to reverse engineer. If an entire R300 driver can be build by reverse engineering, we can be confident that it’ll get done with that intel_hal.so, if there is anything of interrest in it.
Edited 2006-08-13 09:55
1 – never believe intel’s lies
2 – never buy intel anymore
When will all this binary blob madness end ? When will kernel coders enforce the GPL ?
1 – Intel is not lying. They can’t open-source some pretty insignificant things like Macrovision technologies (used in some DVDs). They have provided a complete driver here.
2 – because AMD is better? They’ve said that they have absolutely NO intention of providing open source drivers because they don’t want to let their competitors (nVidia) see the optimizations made for ATi cards.
So Intel is open-sourcing everything they can for this graphics driver and AMD is stating in blatant terms that you won’t get an open-source driver from them unless you pry it from their cold dead hands.
Since you only have two options in processors, you can buy Intel who is trying to help and contributing a great driver or AMD who is trying to keep everything closed. I know there are AMD fanboys, but Intel is the OSS friend from the graphics standpoint at least.
Intel is not lying. They can’t open-source some pretty insignificant things like Macrovision technologies (used in some DVDs). They have provided a complete driver here.
Depends on what you mean by ‘complete’. If you have a TV-Out or HDMI interface on your card the odds are you’re going to need this to get it wo work, otherwise you’ll get nothing out of it.
because AMD is better? They’ve said that they have absolutely NO intention of providing open source drivers
You have to remember that ATI and AMD are still two separate companies at the moment. ATI have categorically said that they wouldn’t open their drivers, and they want to give us the quality we’ve all come to know and love. AMD have certainly stated, off the record, that opening their drivers is something they’d like to do.
…or AMD who is trying to keep everything closed. I know there are AMD fanboys, but Intel is the OSS friend from the graphics standpoint at least.
Errrr. AMD hasn’t exactly been a graphics company for very long. ATI is still very much ATI. There are Intel fanboys as well you know? 😉
Edited 2006-08-13 11:53
Depends on what you mean by ‘complete’. If you have a TV-Out or HDMI interface on your card the odds are you’re going to need this to get it wo work, otherwise you’ll get nothing out of it.
My TV-Out works without Macrovision and indeed lets me do more with its output because of that, and HDMI doesn’t use it. Not really an issue anymore is it?
My TV-Out works without Macrovision and indeed lets me do more with its output because of that, and HDMI doesn’t use it.
I’m afraid that’s not the issue. Of course your TV-out works, or appears to, unless you have Macrovision output put through it. The fact that HDMI doesn’t use it is irrelevant, because HDMI will be using different protection.
Not really an issue anymore is it?
It will be an issue. That’s the point.
Its not going to be an issue for open source software… the first app that requires macrovision support will quickly get forked.
Any app that requires macrovision will require a closed source driver… Those who use close source apps probibly aren’t going to be so insistant that the macrovision source be gpl’d, so I doubt this is really going to affect many people.
…will have disassemblers available, as will Intel and nVidia and everybody else.
Effectively, there are no secrets hidden from the major players, so this kind of posturing is almost entirely done from a legal-appearances front rather than necessity.
This is not exactly Intel-specific, since most tech companies would be effectively bound by it.
I’ve a feeling that nVidia can work out how any optimisations work from the released binary drivers, that’s a very poor excuse.
I’m encouraged that there is a core driver there under GPL which can be used regardless of the binary blob situation. Nice work intel.
Well then what’s the big deal anyway. So you can’t get the open software to play a closed (in so many evil ways) format DVD (I know the macrovision isn’t about getting it to play but improving the sampling, but I think you can see my meaning).
The important thing is that what they did release is truly orthogonal to what they didn’t and fully functional without it. Here fully would mean it does everything it’s written to do without it, so no extra abilities lost.
Intel can *NOT* provide code patented by someone else or related to DRM (DMCA).
Blame the US law.
Fedora (no libcss, mp3,…) do the same. Intel and Fedora do NOT have the choice.
Intel can *NOT* provide code patented by someone else
That really depends on the patent, the patent holder and the source licence.
For example Fluendo has an MP3 licence agreement which allows them to make source available for a decoder plugin (under MPL or an X11/MIT variant IIRC)
I think even the GPL allows use of patented techniques as long as anyone else who uses this code under GPL is getting the same rights regarding the patented parts.
Actually, it looks like Intel did open everything that they legally could and that this is not a crippled driver. So, they didn’t open-source their Macrovision decryption because they legally couldn’t.
I think the message is very clear that the new Intel open-source driver is complete and functional and free.
I’m still waiting for the feature matrix with binary blob and without binary blog.
But I think all this BS around “open-source drivers” without even mentioning the binary blob is not being honest with the (ex)customer I am.
Somebody gave me a link to this site:
http://www.vendorwatch.org/
It shows the friendlyness per vendor, and gives more information after following the links to the vendor names.
Reading through the thread there looks as if there’s possibly some suspicious stuff in intel_hal.so like Macrovision and other pointless DRM stuff. Calls to things like intel_hal_set_content_protection are not very reassuring. The way to get around having to close this stuff in a binary blob is simply not to use it.
I’m also slightly confused by Keith’s post. He then goes on to tell us that the closed binary blob has not been released yet, and if you don’t use it there’s no reduction in functionality ‘from
previous driver releases‘. I take that to mean that you’ll need it to make Intel graphics chipsets actually work fully, certainly in the future, and if you’re using TV-Out or piping stuff over something like HDMI and the binary blob finds something it doesn’t like then you’re going to get a nice black screen.
Needless to say, this has got absolutely nothing to do with patents or trade secrets. It’s all about monkeying about with DRM, and presumably there are various ‘understandings‘ with people like Macrovision that this stuff will be closed.
I agree with Butters. It’s a bit much that they’ve blasted the open source news site world with this announcement, some people then have a look at what’s in it and ask questions and then Keith Packard comes along and says ‘Oh well, it’s not all open source’.
In the last few days, we’ve done a complete circle.
1. Intel Release Open Source Drivers, hoards of open source zealots post about using only intel graphics for their next box..
2. Rumour that AMD is OS’ing ATI drivers, hoards of open source zealots go estatic at such new, buy AMD/ATI!
2. Rumour that AMD is OS’ing ATI drivers DISMISSED, hoards of open source zealots snarll at ATI – It’ll ONLY BE INTEL (or Nvidia maybe) FOR ME!
1. Intel Release Open Source Drivers Which Arn’t Actaully Open Source (in Code), hoards of open source zealots post about intel is evil, and how there going to use <oh wait, who now?> instead.
Face it, VIDEO DRIVERS will NEVER be Open Source, GPU’s are what will sell computers in a future, and no company wants to give anybody, and secrets to whats in their GPU sauce.
I like your roundup, but IMHO you draw the wrong conclusion.
The free software community is literally on pins and needles waiting for open source 3D graphics drivers or for the hardware specs to implement them ourselves. We over-analyze every announcement, we have wild mood swings as positive and negative news emerges from any camp, and we’re a screaming loud minority.
You better believe that the graphics vendors are listening. They may not know exactly how to get us on their side, but they are very well aware that we are sick and tired of proprietary drivers. Our ceaseless plees for open 3D GPUs will be in the back of their heads when the content and consumer electronics industries try to sell them on proprietary standards.
Face it, computer hardware of all kinds has been forced into commoditization by hordes of consumers that have the same basic feature requirements. The staying power of proprietary graphics is really remarkable, but all mainstream graphics hardware will eventually commoditize and become open.
In fact, part of the reason why the nVidia/ATi duopoly is still closed is because the two designs are becoming increasingly identical with every generation. At the moment, the only real difference between the two camps is unified vs. non-unified shader architectures. The graphics companies have realized that everybody wants the same features and capabilities; this is the reality of the massive consumer PC industry.
And the corporate PC industry? They couldn’t care less. All they want is basic graphics capabilities that keep their workforce productive.
DRM is failing. The record industry had a major “hey, wait a minute!” moment when they realized that FairPlay didn’t give them more control over their portfolio, it gave _Apple_ more control over their portfolio. HDMI/HDCP seem like they’re never going to gain wide market penetration. The HD-DVD/Blu-Ray turf wars, especially with first-gen Blu-Ray devices not being able to play HD content, is nothing but misery for consumers.
The content and consumer electronics industries have tried their very best, but they’ve only managed to trip on their own feet with their attempts to bring strongly copy-protected HD content to consumers. One of these days, they’re going to decide that open was better.
In short, I don’t buy into your theory that the future of graphics is closed. Graphics will become open, it’s just a matter of time.
“You better believe that the graphics vendors are listening. They may not know exactly how to get us on their side, but they are very well aware that we are sick and tired of proprietary drivers.”
I think you are seriously overestimating the power of the open source community here. What percentage of computer buyers base their purchasing decisions on open source 3D graphics driver availability?
Same percentage that think about amount of the packaging, political stance of the manufacturer, waste produced over product lifetime, longevity, waste produced by product usage, destruction at end and overall environmental impact before they buy anything? More and more people are starting to care how their actions echo and impact the masses for the better or worse all the time, it’s a trend that will hopefully keep growing. That consumerism doesn’t stop at the grocery store. A lot of open source faithful may be nuts and zealots but they’re also a little nutty and a little overzealous in and about everything they do that impacts the greater good of the entire planet no matter how insignificant some may feel those things are.
for what i understand, the 3d accleration will (or is?) be open… for me macrovision and stuff like these are irrelevant.
I always wonder why so many guys mention patents as a reason for _not_ opensourcing.
On the contrary, patents were initially created so that you did not have to hide your inventions because you could now patent and thereby protect them.
One reason I see for not opensourcing stuff ist that they do NOT have patents on important parts of their technology. There are two possible reasons:
a) They use stuff patented by someone else.
b) They simply don’t have the TIME to patent everything.
Think about it:
How long does it take to develop a new graphics chip?
At which point of the development process are you ready to file a patent?
How long does it take you to actually GET the patent?
How long does it take to develop a _stable_ driver?
And finally: What good is a driver that’s two years late?
There is also the limited utility of patented technology without licensing the use of the patented technology to others. You don’t release the source code for such things because you do not intend to permit others to take the source code and infringe on your patents. You also don’t want to create an arbitrarily-large number of violations of your patents for you to track down by releasing the source code in question for others to incorporate. While the idea of a patent is to disclose the details of the invention for the public, if you have ever read through patents you’d know that they can be pretty vague and of marginal use for implementing something, unless they cover inanely simple tasks.
There is also the matter of other people’s patents and copyrighted material, and all sorts of other licensing details between organizations over intellectual property.
There is also the possibility of making it easy for people to jump out of the woodwork and claim patent infringement. You may be quite unaware of patented methods that are in use in the more esoteric portions of your software, and you do not want to deal with the expense of any inconvenience surrounding it since you are not necessarily obtaining much value from releasing the source code.
Companies whose principle basis is intellectual property are not in the habit of giving such property away. They have no particular overriding reason to do so as part of their business philosophy. When approaching issues of this sort one shouldn’t approach it from “Why isn’t Intel giving me everything?” it should be “Why is Intel giving me anything? Really, why? They surely don’t care about my personal ideology or that I want to run 3D-accelerated FreeBSD on my home computer.”
There has long been speculation that the graphics adapter industry has for years been, shall we say, lax in honoring the patents of other industries and each other. That is to say, most vendors have substantial bits of their products that they either believe to be infringing on a patent, or at the very least questionable to the point that it would draw and expensive suit that would prevent them doom them to financial ruin (whether it’s infringing or not).
As a result, they provide as little information as possible regarding the workings of their hardware.
It’s not very likely that, with the patent system as it is today, that these firms will ever feel that their products are completely free-and-clear of issues.
It should be interesting to see if the Open Graphics guys ever get anywhere. I don’t think the industry takes them seriously, but they should. There’s a subtext of liability that should be keenly aware of with regard to the video hardware being shipped with many OEM systems these days.
Well, I would like very much to have all specs of hardware vendors open but lets face it, it probably will not happen.
Most of us, for one reason or other, like the FOSS movement. It give us a huge opportunit to cooperate, build upon and also to learn like no other way, but for business its is more a strategic move, they will use it to accomplish their own agenda. Not that is a bad thing, there is a lot of synergy on this, but it is just that. They will release strictly what is needed, no more, no less.
Take IBM for example, in many areas the OS is like a commodity for them, they need a stable and dependable base upon what they can build their customer solution (services), if they need to colaborate to reach a satisfactory point, they will, and this is exactly what they do, as also Intel, AMD, Oracle and many others. This way they also share the costs what, of course, all they like.
Do their clients need also an office suit? Yes! Lets share the costs also and everybody will be happy, it is not their main target anyway and the competition on that markets are very tough and the investiment should be prohibitive and very risky. Everybody except Microsoft (and very few other companies) of course. For them the OS plus Office stiil is the main chash cow and they will try to keep their contral on the market as long as they can, what the others companies, again for strategic reason, don’t like. Could they, Microsoft, get most of their money from other services and they would for sure eambrace the FOSS community.
In other words, if the tecnology invelved is not directly responsible for keep money coming, lets share the costs, otherwise try to keep it secret as long as it is possible or lose its big attractives.
So, I don’t think that is realistic to expect they will publish all specs if the secrecy of tecnology is vital for them to have an edge over the competition after all. This is true for Intel as it is true for ATI and many other companies.
I have to add my vote in for full disclosure. They can still patent it with the GPL or etc. I don’t like stuff in my home that is Black Box. Maybe I’m paranoid but my general rule is that I like to know what’s in my domain. My furniture and everything else I should be able to bring in for testing for bad chemicals etc. It’s called the Open Source home based on renewable materials.
What if a chip has a dangerous chemical in it. Do you think a componay will tell you. No. Because of shareholdars or whatever. We have to be very wary especially of nano-tech and DRM harware that can physically harm.
And why do we need money anymore with the Internet. Just do it all for free. Go in the store take what you want and that’s it. No crime. Done.
I think Intel is stepping in the right direction here and AMD is taking a step back. I will probably buy this but only if Linux approves at dev time. If not then it’s the Open Graphics project for me.
So Intel does not provide everything, even though you can still compile a complete driver. AMD chooses not to release their driver to open source. Well guess what folks, these are businesses and they have every damn right to do as they please. All those cry-babies just need to shut up and go out and make their own video cards then. Besides pleasing a few geeks who really have no need anyways, these companies have absolutely no reason what so ever to open source their software. Fact is if you are truly one with an actual need, and will actually do something about it these companies have always allowed others to view their IP.
Frankly too many geeks take things way too far without realising that these closed-source businesses are the ones that allow the whole tech community to grow. Not every damn think in the world needs to be “free”. Grow up.
Are you sure you know what this thread is about? The geeks are asking for an open driver, not for intel to give away their wares for free. Calm down.
It has nothing to do with growing up, and everything to do with principles and wanting Free Software – YOU stop whining and grow up. We’re not going anywhere, and we’re not softening our stance.
It is those who the compromisers always whine about that got free software as far as it has come, and as mentioned before – we’re not backing down. You’ve had 20 years, get used to it.
..in terms of noisy idiots per square inch.
Look, it’s this simple. The only part not released is the module that implements Macrovision. What does Macrovision do? Well, if you’re playing a DVD on your PC it messes with the signal that comes out of the TV Out so you can’t videotape the signal.
So this driver comes without Macrovision support which means that the TV Out signal is not degraded. It is a more functional driver than it would be without the module.
So how did you all fail to realise this? It says in the original mail what the implications are. The only conclusion to draw is that the vast majority of people posting here are too stupid to read properly or think for themselves.
To all the people who posted long screeds about OSS zealots and patents – you have wasted your time. You may as well have shouted it into space.
I suggest you all go back to talking about BeOS, a subject where nobody actually cares enough to correct you.