Intel — Only "Open" For Business

Michael Knudsen writes, "Intel still refuses to work with open source projects such that they can provide their users with proper support for Intel's hardware products. As he has done before, Theo de Raadt once again asks users to take action by contacting Intel, telling them what they think of their current policy of not releasing hardware documentation and granting open source projects the right to distribute hardware firmware with their products. Failing to do so only harms users in the way that they risk having unsupported or malfunctioning hardware in their operating system of choice." Read more below.

It's really important that people understand that Intel is only trying to cooperate just enough to make people believe that they're open and doing the right thing. Don't fool yourselves: They are not.

What we need all users of open source software to do is contact Intel and let them know what you think of their current behaviour. If you run a big department and chose another vendor's products over Intel's because it doesn't work in your operating system, let them know, along with how many units they could have sold you. If you are an end user who has had problems when using Intel hardware because of poor support, let them know.

Let them know that their current lack of support will only harm them in the long run because you will be avoiding their products. Let them know that you want your hardware to work out of the box when you have installed your operating system of choice, and how Intel is preventing this with their lack of support.

Intel is not doing you a favor by requiring you to go to a website and download firmware for your hardware. You paid for the hardware, and Intel is thanking you by making it difficult for you to use it. Let Intel know what you think of this.

Comments

damned if you do...

[macadamia_harold]

Failing to do so only harms users in the way that they risk having unsupported or malfunctioning hardware in their operating system of choice.

So we get unsupported or malfunctioning hardware with our operating system of choice, or we get supported and functioning hardware with a malfunctioning operating system. cool.

Yes, but:

[kripkenstein]

I couldn't agree more with the goal here, but the approach seems a bit unproductive. I refer to the parts like this:

James is a big fat liar

(It's in TFA, believe it or not.)

This is no way to get the other side to play nicely with you.

Re:Yes, but:

[udippel]

If you need to keep IP closed source (for example some whiz-bang algorithm), document the hardware sufficiently that the community can provide their own.

James Ketrenos, Intel Open Source Technology Center

Agreed, I wouldn't call him a big fat liar as long as I haven't given him the chance to respond and do something about the situation.
I hope and guess this was done.
James ?

Re:Yes, but:

[pimpimpim]

well, theo deraadt seemed clearly pissed indeed, but was also smart enough to realise that, and for a correct way to contact intel, he suggests the careful post written by another person that was done to TI as an example how to write to Intel.

Re:Yes, but:

[ari_j]

It's ether that or Theo is a refugee from some parallel universe Bizzaro World, where "up yours, scumbags!" is a polite way to open a formal dialogue and "good morning, gentlemen" is a killing insult. One of the two, certainly.

Well, he is Canadian.

Re:Yes, but:

[GauteL]

"well, theo deraadt seemed clearly pissed indeed, but was also smart enough to realise that, and for a correct way to contact intel, he suggests the careful post written by another person that was done to TI as an example how to write to Intel." .. and strangely enough stupid enough not to realise that everything he publishes on the Internet can have consequences. If Intel-executives have read this story (and it may have been presented to them by one of their employees) then no amount of polite letter from De Raadt is going to help, because the Intel people would have already made up their minds about De Raadt.

His writing was unhelpful, unproductive, unprofessional immature, and downright slanderous.

Re:Yes, but:

[kv9]

His writing was unhelpful, unproductive, unprofessional immature, and downright slanderous.

yet, his style always produces results, while perhaps offending some prudes. can you really blame him for that? i know i can't. RMS usually is very polite when crusading, and still people talk shit. i'm glad FOSS fellows like these are around, which just do their thing, and don't pay attention to the naysayers. i wish him the best of luck in this endeavour.

Re:Yes, but:

[jketreno]

Well, if by 'giving him the chance to respond' you mean posting the original article to the net where he states what he stated, then yes.

If, however, you mean that he engaged in a conversation with me where we discussed anything related to those slides (or anything else in the last year), then no.

Oh well. At least my name will show up more on Google now :)

James

Re:Yes, but:

[evilviper]

His writing was unhelpful, unproductive, unprofessional immature, and downright slanderous.

It's only slanderous* if it's not true.

*(libel actually)

Be professional!

[HRbnjR]

I tried to keep it short and to the point, my email to them read:

Subject: Linux Wireless Firmware Distribution

I was very happy to hear that Intel is working with the community to
ensure that G965 graphics will work out of the box under Linux.

I am very sad to hear that Intel isn't doing the same for their wireless
products WRT freely distributable firmware.

I am a developer in the Computing Services department at a 30 thousand
plus student university. Community enabled Linux support is a huge
factor in the purchasing decisions of our department.

Re:Be professional!

[arivanov]

There is a reason for that.

2D (and 3D) algorithms are commodity, they are well known (most are published 20+ years ago) and large part of the card is designed towards a rather old, but still valid common standard (VESA). In addition to that there is no regulatory regime to deal with it. Having "super duper secrets" in a low-to-mid sector video parts makes no sense whatsoever.

Wireless chipsets operate with a mix of commodity and private algorithms, there is no common spec regarding the way the platform sees them and there standards specify only the external side and nothing on the OS side. In addition to that there is list of Frequency Nazies to deal in every country. All of them insist that any power, frequency and tuning parameters are private and inaccessible to Joe Average Luser. In a modern chipset these are done in firmware and having them secret and limited makes all the sense in the world to a manufacturer. They have to distribute it under strict conditions which limit its possible uses and forbid tampering. If they do not they will lose their license. This forces the license terms on Intel, Atheros, etc. They have no choice on the matter and writing billions of letters to them will be pointless. There will be no change of mind and the firmware will always be under a license that is OSS incompatible. The right addressee for the mail is FCC (and its analogues). It is their business to enforce frequency bands and they are taking the easy way out by passing this responsibility to the manufacturers. If we really want wireless OSS firmware (I doubt that) the enforcement method of the current FCC regime must change and FCC must allow the manufacturers to release such firmware.

Until then, no point to bother and Theo should vent some steam elsewhere. Plenty of new crypto processors around without support in OpenBSD (or elsewhere).

Re:Be professional!

[HSpirit]

If we really want wireless OSS firmware (I doubt that) the enforcement method of the current FCC regime must change and FCC must allow the manufacturers to release such firmware.

You really need to read what Theo and others are asking, rather than making assumptions. If you did you would realise there are two things OpenBSD asks of Intel, and neither of them relate to your conclusions:

1. Documentation for their hardware so that drivers can be written for it without reverse engineering.
2. A licence for the firmware which allows redistribution of it in OpenBSD without restriction.

Note we're not talking about open-source firmware, we're talking about opening up the licensing conditions on distribution of the closed-source firmware.

Intel open enough for me

[Jeffrey Baker]

It should be noted that Intel manufactures the only technologically-current graphics processor which can claim to have open source drivers, and then Intel series of gigabit ethernet NICs is by far the best choice for use with Linux. Intel's wireless chips, the subject of the article, are not completely open but are rather more open than some of the competition.

Re:Intel open enough for me

[TheRaven64]

Intel does seem really odd on this issue. They employ Eric Anholt, for example, who is the maintainer of the FreeBSD DRI port, and so they have the best supported (blob-free) 3D cards for FreeBSD. But they don't let you distribute the firmware blob for their cards. The documentation is quite important, but the OpenBSD team has become pretty good at reverse engineering WiFi cards now (I almost feel like giving them documentation would spoil their fun). Not allowing the distribution of the firmware is inexcusable, however. Without the firmware, the card is completely useless. Theo is not asking for the source code, he is just asking for permission to distribute a binary that is on the drivers CD for the card so that users don't have to hunt for the CD (and probably use Windows to decompress the installer) or download it from the Intel web site.

Just to re-itterate; he is not asking them to open source the firmware. He is asking for them to allow OpenBSD to distribute the binary. This is not a binary blob in the OpenBSD sense, since it runs on the WiFi card, and not in the kernel. On older cards, it would have been stored in ROM on the board, but modern cards save money by making the driver load it at runtime.

Re:Intel open enough for me

[LWATCDR]

Intel is doing this for a good reason I am afraid.
They believe, or their lawyers believe, or the FCC has told them that they can not release the full programing specs to those adapters.
Those wireless adaptors use soft radios. You can change the frequency, transmission power, and goodness knows what else by just changing a register.
Intel runs a huge risk of law suits, products loosing certifications, and possible criminal actions if they release those drivers.
Will Theo indemnify Intel for all damages and criminal liability if they release the information?
The FCC requires unlicensed devices to be not easily modifiable to operate out of band. The ability to go in and change a const or DEFINE MAXPOWER from 0xFE to 0xFF may be considered easily modifiable by the FCC.

You could argue that they should put hardware limits on these chips so they could release them but that wouldn't help for the current adaptors and frankly would add a lot of cost to the chip set while providing no benefit to the users except that they could have "free" as in speech drivers. Of course they can have that now if they use an Orinoco card. And they can have free as in beer drivers now for the Intel chip set. Of course going with hard radios does have some down sides for the user. If the FCC opens up some more space in the band the wifi card works in or changes the power limits a soft radio card can adapt with new drivers while a hard card is stuck.

So what Theo is doing is grandstanding to keep his name in front of people. Why is he complaining. Frankly it is totally with in the spirit of BSD for Intel to grab a BSD driver, modify it, and then offer only a binary of it. And I have to wonder just how important a wifi card set is to the OpenBSD community? OpenBSD is mainly used on servers. Servers don't often use WiFi cards and even if are using OpenBSD on a notebook you have the option of using an Orinoco card.
Theo is choosing to ignore the reality of these regulations because they are inconvenient.
If he really wants this to happen he needs to go before the FCC and what ever the EU, Japanese, Canadian, Australian, and or Chinese equivalent agencies are and get them to rule that releasing the specs will not effect the licence of these devices.
If he can do that then all the wifi chipset makers will have the opportunity to release the specs to the FOSS community.

Other wise he is asking Intel to risk a few million or billion dollars in fines, court costs, and or lost sales so he can have a driver with a license he likes.
Yes it would be great to have totally free wifi drivers. But in this case Intel has some really good legal reasons for releasing only binary blobs.
In other words they are not just trying to be proprietary.

Give them credit for all the documentation that they do release to the FOSS community. From what I hear only the wifi adapters are restricted.

Re:Intel open enough for me

[repvik]

The FCC requires unlicensed devices to be not easily modifiable to operate out of band. The ability to go in and change a const or DEFINE MAXPOWER from 0xFE to 0xFF may be considered easily modifiable by the FCC.

That shouldn't matter. One can still reverse-engineer the firmware and figure it out. This is like banning hammers because they can be used to break into stores. The responsibility should not be on the hardware/software side, but on the user side. I can still choose the wrong country when installing a wireless card to use a frequency that is illegal in my country (Norway).

Who cares?

[t0qer]

I'm not sure how to respond to this one without getting downmodded into the pits of hell, but here goes...

This article was very scant on what exactly intel isn't supporting. All it says is some blurb about requiring folks to download firmware before they can use their OS of choice on intel hardware.

WHAT HARDWARE ARE YOU TALKING ABOUT?

CPU? Chipset? NIC? Router? Switches? What.. What the hell are you complaining about? Bios updates for Motherboards?

I hate to bitch, but when you get some pretty good in depth stories rejected for lame hoopla like this, you get mad.

--toq

Re:Who cares?

[Nikademus]

You should read deeper. Theo was focused here in firmwares for wireless chipsets. There are probably other firmwares needed (as for RAID cards)

Not Holding My Breath

[Jekler]

Intel's behavior won't affect the market one way or another. As a whole, the market is barreling towards an open source model. If Intel opens up, that's great. If they don't, it won't matter because someone else will enter the market that's willing to do so. The market will follow the demand, with or without Intel.

Re:Not Holding My Breath

[Nutria]

As a whole, the market is barreling towards an open source model.

What hard, current evidence do you have that the market is barreling towards open source?

I see it meekly traipsing along, while MSFT earned $1400 every second of every day of fiscal year 2006, and is on pace to earn $1500 every second of every day of fiscal year 2007.

How is this to Intel's advantage?

[jorghis]

I dont totally understand whats going on here. What does intel stand to gain from refusing to publish hardware documentation? The article seems to imply that they are doing something shady and sneaky so that they can make more money but I dont see how this is to their advantage in any way. How do they stand to gain by having people writing software without proper documentation? I would think this would hurt them if anything. Can someone please enlighten me? Although I am ill informed on this issue, calling someone you are trying to influence a "big fat liar" and publishing anothers personal email so that they can be spammed hardly seems like a good idea.

Re:How is this to Intel's advantage?

[Fulkkari]

If the hardware turns out to be extremely buggy, then it might be Intel's advantage not to publish any documentation. Their drivers and firmware code might be full of software based workarounds for hardware flaws that the PR-department would not want the public to see. If this was the case, publishing these to the open source community would make a hole in Intel's credibility as a hardware manufacturer, and possible create monetary losses in selling new products.

Note that I'm not saying that this is the case, but it might be.

Re:How is this to Intel's advantage?

[portmapper]

> If the hardware turns out to be extremely buggy, then it might be Intel's advantage not to publish any documentation.

This was suspected as the reason why Sun did not release hardware docs for UltraSPARC III. Only very recently did
OpenBSD have working device drivers for UltraSPARC III.

not our enemy

[juventasone]

If you compare Intel to other motherboard, chipset, or processor manufacturers, you'll find they arguably have better documentation and support for end-user and IT people than any of their competitors. They also are one of the only manufacturers I've seen to use open-source projects like FreeDOS and ISOLINUX. In their server lineup they support Linux as much as anyone.

Since I'm not a developer I can't speak from a developer's perspective, but there seems to be a liking in this community to paint Intel with a brush of "evil tight-fisted corporation" when they're actually one of the few who act like they care.

My Suggestion

[Daengbo]

While I actually think TFA is virtually useless, I understand that people want better wireless support for their various open source OS's. Intel's drivers for this are really quite open when compared to most others, but if you want drivers that are more open than Intel's, choose ones with the RT2400, RT2500, RT2570, and RT61 chipsets by RaLink. The drivers were open-sourced last year and have progressed quite well. Find more info at http://rt2x00.serialmonkey.com/wiki/index.php/Main _Page and http://sourceforge.net/projects/rt2400.

Re:My Suggestion

[sirambrose]

I've bought a rt2500 and the drivers are not really good. The code is messy enough that the kernel developers won't accept the driver and the driver is missing features such as WPA. They are rewriting it, but the new version will not be accepted into the kernel until the devicescape framework is. I still can't get WPA to work with the beta driver.

I'm not saying that people shouldn't buy the card. I bought the card because there are no better option. It is one of the few that will be supported out of the box with linux since it doesn't require a binary firmware or a binary kernel module. It just isn't ready quite yet.

Re:My Suggestion

[Daengbo]

I'm not claiming anything is perfect here, just that people who care about open drivers (what the article is about) should choose this card if they can, because it is completely open, with no binary firmware. If the community works on this one well enough, it will have all those "missing pieces" soon enough. Incidentally, WEP, WPA and WPA2 should be handled by the 2x00 Beta4, due out "real soon now."

Okay, I'll Try

[Anonymous Coward]

Dear Intel,

I am humble Nigerian prince with a great wealth of BSD users in a locked-out community...

Aw, shucks.

Re:Intel DOES provide some OS drivers

[portmapper]

What OpenBSD asks for is hardware documentation, not source code.

They also ask for the right to distribute firmware under an acceptable license, but Intel refuses. Ironically your link
above describe exactly the Intel attitude: http://ipw2100.sourceforge.net/firmware.php

Upon selecting a link above you will be taken to the firmware license agreement. Agreeing to the terms
presented on that page will direct your browser to the firmware download.

Re:Why Firmware?

[portmapper]

OpenBSD want to distribute firmware along with the OS under an acceptable license. They are not asking for the source
code of the firmware. Intel are instractible here, so owners of Intel wireless devices needs to personally accept a license
before downloading the firmware. As an example: http://ipw2100.sourceforge.net/firmware.php

As for open source drivers: OpenBSD wants hardware documentation, not a Linux driver, so that they can write their own drivers.
Intel claims that they are open source friendly and gives out documentation, but the last is clearly a lie since OpenBSD had to reverse
engineer several Intel wireless chipsets.

Giving the appearance of beeing friendly to open source, while not beeing so, is the latest fad in business. Intel is an example
of this fad.

Better article on the story.

[Karellen]

http://kerneltrap.org/node/7184

Re:Intel DOES provide some OS drivers

[sirambrose]

Intel does distribute the firmware for the newer ipw3945 driver under a sane license. Unfortunately, nobody distributes that since it requires a binary daemon to function. One has to wonder why Intel has not relicensed the other firmware files. They have acknowledged that the ipw2100/2200 license is too complicated and doesn't meet the needs of distributors, but they don't want to fix that problem. It would seem that Intel does not want their drivers supported out of the box on open source operating systems.

If I were to write to Intel, I would ask that all wireless firmware be released under the ipw3945 license. Intel legal has already approved it and it provides a clear description of exactly what we want.

Re:My letter

[Cal Paterson]

Shite. Forgot about html ;)

From: Cal Paterson
To: majid.awad@intel.com, peter.engelbrecht@intel.com
Date: Oct 1, 2006 1:06 PM
Subject: Intel Firmware for the Wireless chips

As an OpenBSD user and "Intel Wireless PRO" owner, I would like you to
release your firmware for the "Intel Wireless PRO" chipset. I have an
IBM/Lenovo Thinkpad that uses this chipset, and I am unable to use it
without the binary blob firmware you provide.

You often say at conferences that you are committed to Open
Source/Free Software, and that you release sourcecode to that effect,
but often times you fail to release critical code, or even
documentation that would make it possible for the community to
re-create that code.

At the Open Drivers Summit, James Ketrenos said:

"If you need to keep IP closed source (for example some whiz-bang
algorithm), document the hardware sufficiently that the community can
provide their own."

This is a fine statement, but it would probably be more meaningful if
Intel would actually do so. The wpi driver for OpenBSD is currently
suffering for lack of documentation from Intel.

Lies and double standards are the currency of your commitment to Open
Source/Free Software as it is.

However, this is a issue that is easily solved. Release the
documentation for this chipset (or, even better, the original code).

[Next time, I'll use the preview button ;)]

Not going to happen

[vladimir_putin]

I work for some company developing wireless firmware for our ( completely unrelated ) product. Opensourcing the firmware or HW specs below it is not going to happen, ever, by any company. The reason is that wireless devices must comply with wireless standards. The firmware plays an important part in creating this compatibility. Opening the HW specs would mean that the original company would have to support some random hacker "optimizing" the algorithms in firmware to work better with his scenario forgetting other features that he does not think really matter, but are necessary for the wireless devices working toghether. HW specs are not designed to be easy to understand. They are designed to kick ass in performance or save 0.001mm2 from the silicon area. Usually the savings in silicon area come with the penalty of the interface being "interesting" to say it nicely. Also the HW versions change quite often and HW bugs are worked around in firmware. The amount of work to document all the bugs for open source firmware writers would be humongous. There are not really that many people working with the firmware. Gaining complete understanding of how our own firmware works takes years for for any novice entering the team. Nobody from our team wants to get into scenario where we must try to understand tens of different versions of the firmware and what are the implications of running each of them. The published interfaces need to be - quite stable across HW revisions. - Must not be able to compromize wireless standards compatibility Open source can work with drivers, but never with firmware. This is life. Deal with it.

or Dont buy Intel

[zenst]

I will voice my opinion in the tried and tested way of consumer protesting. I will just not buy Intel for my OpenBSD box's.

    I will buy hardware that has an open support commitment and prove those vendors right in there move.

Difference in Paradigm

[Chabil Ha']

See, this community bases everything off of a paradigm that everything ought to be free "as in libre". However, the current paradigm (which I admit is slowly shifting) is modeled around the though that information is worth money and power. Because there is secrecy surrounding the code that they use to guard their property, then they have control over how their property is used. They are able to make money off of it. This is their motive. This is how capitalism works.

They see releasing that information as a threat to their MO. They think that if they start handing this stuff out for free their turning into a bunch of commies. And even though this community knows that isn't true, it doesn't help using ad homonym attacks against them by calling them 'big fat liars'. It looks childish and immature.

As for emailing? I don't think they give hoot whether a few geeks boycott them because they don't get open source drivers, mostly because there will always be someone else who will buy their product without qualms. Only if someone like Dell dropped Intel for such reasons would they begin to notice. What would happen if Apple and HP dropped them too? Sure they would wake up. But you know why none of them will do that? Because, they operate under the same paradigm. And 99.99999999999% of their customer base doesn't care because (to them) it's irrelevant.

Intel Wireless Motes are OpenSource

[cabbi]

I work at UC Berkeley on ecological monitoring using wireless sensors. We have been collaborating with Intel Lab, Berkeley for the last 2 years and their wireless hardware, "motes", use an entirely open source OS/firmware: tinyOS. They made this a deliberate strategy, by collaborating with the university they get high quality fast developing firmware and they make the money on the hardware design. So far it has worked well for all of us. Intel is a big company. Not all of their divisions play badly.

Doublespeak

[el borak]

From the article:

We do not believe that Intel is not special enough that they can take people's money and their rights.

I can't misunderstand why Intel won't refuse to not release the unavailable documents.