Blaming Intel for how the world is (Moblin Zone)

Posted Oct 28, 2009 19:09 UTC (Wed) by LinuxDevices (guest, #51613)
Parent article: Blaming Intel for how the world is (Moblin Zone)

Howdy, author Henry here. Yep, MoblinZone is astroturfing. But, it's also paying my bills during a career transition. And, they let me write whatever I want, and it's about Linux, so what the heck.

I edited LinuxDevices for six years, so I know a thing or two about closed drivers in the embedded world. No one at Intel suggested I write about this topic. In fact, I'm sure I'll get in trouble over it. But, it's something I feel needs to be clarified, because too many Linux folks buy these Poulsbo devices and then learn too late of the graphics driver issue. Heck, I know -- I was one of 'em.

Still don't think it's Intel's fault, though. They have to use best-of-breed graphics, if they want to compete in this new market. And, it wasn't their decision that the drivers be closed. And, in that market, at that value point, it's hard to imagine the drivers being open. My own view, not that of Intel.

Cheers, thanks everyone for your feedback.

Blaming Intel for how the world is (Moblin Zone)

Posted Oct 28, 2009 20:07 UTC (Wed) by ncm (guest, #165) [Link]

Well, shucks, then.

Blaming Intel for how the world is (Moblin Zone)

Posted Oct 28, 2009 20:22 UTC (Wed) by rahulsundaram (subscriber, #21946) [Link]

The article is wrong in several different ways:

* People do want to hack devices and hackable devices are very popular for that reason. Even vendors who sell routers are realizing it and marketing it as such.

* The proprietary portions are really the X bits rather than the kernel bits

* Intel contracted out the development of a proprietary driver. It was a choice and not some kind of free market law that they should do that. They definitely deserve to a share of that blame.

* I could before recommend any Linux user to just verify that it is a Intel chipset and be guaranteed that it would work out of the box well or atleast that bugs can be fixed by other people. The brand damage and loss of trust from this move is definitely going to hurt Intel considering things like ATI driver in Fedora 12 getting 3D support and Intel isn't the only vendor anymore to understand the benefits of free and open source drivers and good documentation for the hardware specs.

Blaming Intel for how the world is (Moblin Zone)

Posted Oct 28, 2009 20:37 UTC (Wed) by jejb (subscriber, #6654) [Link] (7 responses)

> I edited LinuxDevices for six years, so I know a thing or two about closed
> drivers in the embedded world. No one at Intel suggested I write about
> this topic. In fact, I'm sure I'll get in trouble over it. But, it's
> something I feel needs to be clarified, because too many Linux folks buy
> these Poulsbo devices and then learn too late of the graphics driver
> issue. Heck, I know -- I was one of 'em.

I think there are two issues here. The first, as you say, is the damage done to the impression we all created that intel graphics was the safe default choice for a desktop that just worked. The problem is that there are many articles available on the internet stating this and suddenly its no longer true ... unfortunately, we can't erase the articles, so many people buy the hardware expecting it to work and get annoyed when it doesn't. Intel actively encouraged the creation of that impression, so they share some of the blame for making it a lie.

The second issue is your contention that binary drivers are appropriate for some markets (and even beneficial for some products). I could do a lengthy refutation, but I think the words of an intel engineer are more eloquent in this situation:

http://lwn.net/Articles/162686/

Blaming Intel for how the world is (Moblin Zone)

Posted Oct 28, 2009 23:48 UTC (Wed) by LinuxDevices (guest, #51613) [Link] (1 responses)

Hey, no one's arguing that open drivers have their advantages, even for device developers. As the PC world and device world collide, I'm certain that'll be one of the benefits. We just ain't there quite yet. The device world has a lot to learn from the openness of the PC market, that's for sure.

Oh, say, that reminds me... are there any open source implementations of OpenGL ES yet, or is everyone still waiting for Nokia to do it? Dunno if it's true, but I've heard it sort of implied once or twice that that might free the way to open drivers for this sort of hardware IP.

Stumbling toward efficiency,

-Henry

Blaming Intel for how the world is (Moblin Zone)

Posted Oct 29, 2009 1:59 UTC (Thu) by pabs (subscriber, #43278) [Link]

There are several free OpenGL ES implementations (including the reference implementation - MIT-licensed), the latest is the OpenGL ES Gallium3D state tracker in Mesa IIRC.

http://www.khronos.org/developers/resources/opengles/
http://www.phoronix.com/scan.php?page=news_item&px=Nz...

Blaming Intel for how the world is (Moblin Zone)

Posted Oct 30, 2009 17:45 UTC (Fri) by LinuxDevices (guest, #51613) [Link] (4 responses)

Hah, oh yeah, Arjen...

I'll see your Intel engineer, and raise you a Linus.

If I write a driver, I might have good reasons for keeping it closed (companies do). Or, I might have no reason at all. But at the end of the day, the code is my property, and it's my decision to make.

That was pretty much Linus's line of thinking, if I recall correctly, in deciding to tolerate binary drivers: he wasn't comfortable telling people what they had to do with *their* code.

Had Linus decided differently, I have no doubt that Linux would not be top device OS today. Razor thin margins and feverishly paced product development cycles aren't that conducive to code license audits and kicking changes up to the kernel mailing list. I haven't seen them myself, but have heard from reliable sources that the third-world engineering sweatshops feeding our collective consumer lust have quite a tartarus-of-maids-like atmosphere. [Reference: http://etext.virginia.edu/toc/modeng/public/MelPara.html]

Andrew Morton and the embedded maintainers are doing a great job educating device companies that over the longer view, *in cases where an OS update -- either by the company or users -- is likely down the road*, there can be technical advantages of doing exactly that. But it's an ongoing process.

I think it's great that so many of the posters here on this list understand that from a technical point of view, closed drivers *are* evil. Yet, if you believe in individual property rights, they're kind of a necessary one.

-Hick-from-the-sticks Hank

Blaming Intel for how the world is (Moblin Zone)

Posted Oct 30, 2009 18:10 UTC (Fri) by pboddie (guest, #50784) [Link]

Had Linus decided differently, I have no doubt that Linux would not be top device OS today.

So why isn't NetBSD more popular, to take an example of a credible alternative? I remember in 1994 or so when a bunch of people decided to port a free Unix variant to ARM (on the Acorn Risc PC - a convenient target environment) that the first choice was indeed NetBSD; this formed the basis of Oracle's second generation network computer, supposedly. As I recall, ARM Linux landed after RiscBSD (the port of NetBSD, later known as the acorn32 port), yet you'd be fortunate to find many of the embedded ARM board vendors pushing NetBSD instead of Linux, or even offering it as an option.

If one bought into the tiresome arguments about vibrant big-money development only being supported by permissive licensing, one would have to look away from the success of Linux and how stuff like NetBSD has been largely ignored.

Blaming Intel for how the world is (Moblin Zone)

Posted Oct 30, 2009 18:32 UTC (Fri) by jejb (subscriber, #6654) [Link]

> That was pretty much Linus's line of thinking, if I recall correctly, in
> deciding to tolerate binary drivers: he wasn't comfortable telling people
> what they had to do with *their* code.

Erm, actually not really. There have been many clarifications to this; e.g. here:

http://lkml.indiana.edu/hypermail/linux/kernel/0312.0/067...

The bottom line is that for Linus to be happy with a binary module it mustn't be a derived work of linux; specifically must not be designed for linux or poke about with linux internals.

> If I write a driver, I might have good reasons for keeping it closed
> (companies do). Or, I might have no reason at all. But at the end of the
> day, the code is my property, and it's my decision to make.

Actually no: the code is only your property if it's not derivative of anything else. If you derive, you have to comply with the licence allowing that (in this case the GPL). What constitutes derivation is a fairly nebulous legal question, but you'll see in the above that Linus thinks it's anything designed to fit in with Linux kernel interfaces. That's pretty much every linux driver out there in the embedded space.

> I think it's great that so many of the posters here on this list
> understand that from a technical point of view, closed drivers *are*
> evil. Yet, if you believe in individual property rights, they're kind of
> a necessary one.

The property rights vs open source is a bit of a false dichotomy. The only way you can comply fully with the GPL without worrying about derivation problems is not to distribute anything based on your changes. The problem with that legally sound approach is that no distribution means no return on investment, even though it absolutely preserves your perceived rights of the code creator.

The object of a modern business isn't preserving individual property rights, it's achieving ROI.

When phrased correctly like this, I think one can make a convincing argument that, even ignoring the costs of the legal risk associated with being closed source, one can achieve more ROI by open sourcing drivers than by keeping them closed. In the case of the GMA500, part of that actual cost was the lack of a "just works" factor that caused market confusion leading to product returns.

Razor thin

Posted Oct 30, 2009 20:18 UTC (Fri) by man_ls (guest, #15091) [Link]

The "razor thin margins" argument is a bit tired. First, consumer products everywhere face the same margins which is why they are sold by the million -- and why those margins are converted into gigantic profits. A 9% profit in a bad year is not so bad.

And second, the costs faced by companies doing code audits were analyzed by LWN.net in a recent series, and they were of the order of thousands of . Not millions, not tens of thousands. The argument could still be made for specialized hardware that sells a few units, but this kind of money is hardly relevant when you sell several million units like Intel -- say, for this horrid Poulsbo thing.

Another tired argument: posting your patches to lkml (or xorg lists) is in no way required in the Free software world, although it can be an advantage in the long run. But these engineering sweatshops just need to post the code somewhere under an appropriate license; if there is enough interest (and in this case I'm sure there is) it will find its way upstream in no time. Anything is better than a closed, proprietary driver.

Blaming Intel for how the world is (Moblin Zone)

Posted Dec 15, 2009 14:36 UTC (Tue) by jmcvetta (guest, #62509) [Link]

I think it's great that so many of the posters here on this list understand that from a technical point of view, closed drivers *are* evil. Yet, if you believe in individual property rights, they're kind of a necessary one.

There are an awful lot of people who do believe in individual property rights; yet don't think intellectual monopoly privileges granted by the government to corporations have any similarity whatsoever to the conventional meaning of 'property'.

Blaming Intel for how the world is (Moblin Zone)

Posted Oct 28, 2009 20:53 UTC (Wed) by martinfick (subscriber, #4455) [Link]

So do you really believe that no one ever wants the legal right to repair their toaster on their own or by an electrician? I have repaired mine, my dad repaired it also 30 years earlier. If it were software, and we had repaired your closed driver device instead of a toaster, we both would have committed illegal acts. You do not lend the appearance of understanding what you are talking about if you can't see that proprietary software is way worse than any "non user serviceable hardware". Just because you likely never go under the hood of your car doesn't mean it's OK to outlaw it.

Blaming Intel for how the world is (Moblin Zone)

Posted Oct 28, 2009 21:37 UTC (Wed) by robert_s (subscriber, #42402) [Link]

"And, they let me write whatever I want, and it's about Linux, so what the heck."

Is that really true? Would they have been fine with you writing an article taking the contrary position?

"I edited LinuxDevices for six years"

I'm a bit disappointed then - I love linuxdevices.

Blaming Intel for how the world is (Moblin Zone)

Posted Oct 28, 2009 22:28 UTC (Wed) by bronson (subscriber, #4806) [Link] (1 responses)

Hi Henry.

> And, it wasn't their decision that the drivers be closed.

If it wasn't Intel's decision, then whose was it? It sounds almost like you want us to believe poor Intel was muscled around by Imagination Technologies??

Intel shipped a broken product and now you're trying to pretend like they had no alternative and that it's somehow someone else's fault. They're a gigantic and influential company -- they definitely could have written their own terms in the deal.

Blaming Intel for how the world is (Moblin Zone)

Posted Oct 29, 2009 0:13 UTC (Thu) by LinuxDevices (guest, #51613) [Link]

Well, I don't know... I hear there's pretty good volume in iPhones and the like these days. Apple and Intel both own little chunks of ImgTec, but no where near a controlling interest.