Linux kernel maintainers removed hooks from a semi-binary driver (half open, half closed) for Philips webcams. This has angered the developer of the driver who have worked on the driver for 5 years, resulting on removing the driver completely from the kernel and his site. The developer has an NDA with Philips and he can’t fully open source the driver, even if he wanted to. The losers from the whole story, are the users.My Take: Here’s an example of some linux kernel maintainers that put who-cares politicalities over practicality & user’s interests. Being Free is a good thing, but if I can’t be 100% Free, I’d take that 50% any day. Common users (the kind of mammal species that the Linux platform is trying to captivate away from Windows/Mac) want cheap and easy solutions, and some Linux maintainers fall short in materializing this need because of this or the other (irrelevant to most users) reason.
Strict Open Source Kernel Policies Give an End to a Driver
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
The authors point seems like a slippery slope to me. I say Linux should stay the course.
I agree.
It isn’t NVIDIA by creating drivers outside of the kernel. ( I know, I know, video drivers are usually kept out of the kernel , but still… ).
I think in the case of hardware, OSS devs are going to have to figure out how to deal with Hardware vendors that do not want to share their IP secrets. This is why I believe that an OS independant driver framework (like what sci-tech has done) is the way for future OSs to go.
Unless there is some sort of Open hardware movement that I am unaware of… I am under the impression that the Free Software Movement is all about Software – specifically.
I don’t understand what the problem is here. Nvidia doesn’t have any Nvidia specific parts of the kernel that relate to their video driver, yet they release a binary only module without a problem. There are plenty of other companys that release binary only modules also. Why doesn’t philips just do the same thing or completely open source it? It seems ridiculous to remove it from the linux kernel now, but really should we let hardware companys add pieces to the kernel that doesn’t do anything without the binary piece anyway? What functionality would it add to the kernel aside from aiding this paricular binary driver that wouldn’t be shipped with the standard kernel anyway?
This isn’t the kernel developers fault at all, the driver shouldn’t have been designed to work in this retarded way in the first place. Philips and the developer of this driver should make the driver it’s own module just like every other company out there, no special treatment for philips and obviously not 50 seperate parts of the kernel that help 50 manufacturers binary only drivers. If this developer decides he’s not going to work on the driver now, that’s his fault, and if philips can’t offer a way for me to use their products in the same manner everyone else can, then i just won’t use their products, simple.
“Common users (the kind of mammal species that the Linux platform is trying to captivate away from Windows/Mac) want cheap and easy solutions, and some Linux maintainers fall short in materializing this need because of this or the other (irrelevant to most users) reason.”
Linux the kernel is not trying to captivate or steal away Windows/Mac users. What ever gave you that idea ???
BTW, I think drivers should be seporate from the kernel (say waht you want about performance, but Windows runs great). It’s way easier for joe user that way, IMO.
“if you really want to make Linux a good user experience, it should be easy to install, and have good hardware support. The harder you make it to get your stuff working under Linux, the quicker people will turn away.”
So tell it to the manufacturer. This is why I am VERY careful about what hardware I buy to use with Linux. I have never had an issue such as this, since I don’t rely on hardware manufacturers who refuse to release the specs.
One of the biggest pet hates I have with Linux is that the core kernel developers insist on open source drivers, and they change regularly change the binary driver layer to force manufacturers to update driver packages, until they give in and open source the driver. This results in binary closed source drivers not working on several kernel versions/distributions. I take the side on the manufacturers, and would prefer closed source binary drivers (than none at all).
This piece demonstrates the common OSS-commentator’s fallacy: an attribution of the personal prejudices of the author onto some nebulous “community” of linux folk. Fact is, while the author may want linux to “take normal users away from win/mac”, that does not mean any “community (which is to say, collection of individuals) has any interest in this at all.
The people who actually write the linux kernel may have all sorts of reason for doing what they do, such as enjoying writing code for its own sake (gosh). They may well see (and I would agree with them) a lot of self-serving drivel about “weaning linux and mac users” as just a political triviality, and irrelevance.
This same line of argument goes for all the “why does the community write 1000 text editors instead of working on what I want?” and “why don’t KDE and Gnome unite so they can do what _I_ want and go on some political smash and grab against Mac and Windows users.”
Commentators need to try and understand that the linux “community” is a complete fiction, they need to be more open about their personal biases and stop trying to make out they are the default and that everybody else shares them.
People work on kernels and write code for various prjects because they enjoy it in their free time or because (sometimes) they are paid to do so. Unless you are paying them, you have no particular right to expect them to implement your personal agenda. Their work is a gift. Take it or leave it.
Linux should remain 100% free. It’s up to distributions to decide to include non-free drivers or software.
Is not going to help the end users of a system! This is clearly one of those cases. I did not see the opposing side’s view of why they chose to mutate the work of another to remove functionality. I’d like to think they had a far better reason than “purity to open source” as their motivation, such as perhaps that of security or stability. I get the impression, however, that if the driver has been happily used for this long, that they can’t claim such a thing. Are they paranoid that the portion of the driver that is closed is sending Philips information as a spyware application? Truly silly!
To whomever made the choice to cut off a body part that wasn’t theirs to cut off: what made you think that was wise?
Drivers should work on some sort of Hardware Abstraction layer, which THEN is hooked into the kernel.
It seems to me that a bad driver directly hooked into the kernel could cause all sorts of problems.
However, not maintaining a binary consistency across kernels, just seems idiotic to me.
Finally, when are we gonna get a game API that just works, like direct x? Damn, if only linux had more games, I’d never boot windows. Instead, I still log on to win2k, cuz I don’t have time to hobby OS with Linux, etc.
I think it’s wise to keep the kernel itself ‘pure’. The Linux kernel should be 100% open source without compromises. It’s not the job of the kernel to have plug-and-play support for every hardware and software out there. That’s what distro providers are for.
It should be the kernel maintainer’s responsonsibility to make sure the kernel itself is 100% open source and easy to port. The support for non-OSS drivers should be left to the distro makers like Redhat, Suse, Mandrake, etc.
Whatever happened to the GKHI (Generalised Kernel Hooks Interface)? It was designed to allow just this kind of thing. Did IBM drop it in the end?
Did anyone even read the title of this post?
“Strict Open Source Kernel Policies Give an End to a Driver”
The title itself gives you the answer. How is a driver that requires a binary-only file open source?
Again, leave the kernel 100% open source, let the distros handle the binary-only/partial-OSS-partial-binary drivers.
WTF! Sheesh. I mentioned the problems the author of this driver was having a couple days ago and everyone thought it wasn’t a big deal. Now look what happened!
So this guy dedicated his free time for years to create drivers for a lot of the most popular webcams and then he gets shafted. And WTF for? Some stupid idealogical reason about binary drivers?
Damn this stuff gets me riled up. I have the very popular and very good QuickCam Pro 4000 which works great with these drivers. So I guess I’m not upgrading past kernel 2.6.8. It was so easy to install these drivers before, I’m sure now it’s gonna be a hell of patching and compiling.
I like Linux but stuff like that is completely incomprehensible for me. We need a kernel developed in the style of KDE, f*ck ideology and just concentrate on making the best possible product for the users.
Finally, when are we gonna get a game API that just works, like direct x? Damn, if only linux had more games, I’d never boot windows. Instead, I still log on to win2k, cuz I don’t have time to hobby OS with Linux, etc.
SDL, which makes use of OpenGL, and OpenAL. Handles all sorts of stuff and works under unix, windows and osx. Really, there’s no reason to limit yourself to just windows anymore by using direct x.
I have never had an issue such as this, since I don’t rely on hardware manufacturers who refuse to release the specs.
So I guess you’ve got the specs for your BIOS? Why stop at your PC, perhaps you should get the code to you TV box and your car’s onboard computer. If you’ve got principles, make sure you don’t have exceptions.
Do you think that the Nemosoft programmers are idiots who can’t program themselves out of a brown paper bag? Why do you think that ONLY open source programmers are good at writing drivers?.
best regards
Dev Mazumdar
This module had this hook for 3 years with no problems. And now it’s suddenly a big problem?? I would have little problem with this decision if there was a solid, technical reason why the hook was removed, but this is pure shit.
Seriously, who the hell cares if it’s not 100% clean or 100% open source. Is it not more important that some of the best products in a class (webcams) are well supported?
Grrrr.. Maybe if enough people write to the USB maintainer (Greg Koah), he may reverse this decision. I guess I’ll do that once I’ve calmed down a bit.
But really, is there any good reason for this decision?? Why isn’t anyone more pissed about this? I guess noone here has a webcam.
Great post. I too am tired of seeing the same arguements you describe, also being made by journalists and posters all over the net. Unfortunately, some misguided souls in this thread must have passed over it.
Where’s the other side of the story?
It seems to be a LEGAL ISSUE not a POLITICAL ISSUE. Your rant (“my take”) adds nothing to the issue. There are already enough people who have “that take”.
“Why do you think that ONLY open source programmers are good at writing drivers?.”
Strawman. Read the original post.
Why the hell did he pull it?…so they removed the hooks. Make a patch for your driver and I bet many distro’s would include it. The Linux kernel devs were right IMO..the linux kernel is open source and it makes very little sense for the kernel devs to keep hooks around for a non-open binary driver.
The author however is doing nothing but throwing a hissy fit. I am also a dev and I realize that this would be frustrating…but is that frustration really worth throwing away all your work and pulling the driver totally? I thought we were doing this to make a great OS/for fun, not to stroke our own egos by having hooks in the kernel.
On another note: it would be nice to have a generic support for loading binary drivers. Doesn’t FreeBSD have something like this?
Nobody cares that it’s not 100% open source. However, not being 100% clean is a good reason to remove anything. Hacking up the kernel to retain binary compatibility is the fastest path to the dark side. Microsoft can barely limp along doing that sort of thing, but they have 10x as many people hacking on Windows. The kernel developers have better things to do.
There’s a good reason to keep everything GPL compliant. The fact is that it isn’t just Linus that gets to decide if binary things get into the kernel. Every single person whose code is in the kernel can bring a suit against the kernel for including binary-only stuff. The original developer isn’t the only person that has GPL-enforcement privledges and the people who made this decision have obviously contributed to the kernel and could enforce the GPL against this stuff any time they wanted in a court of law. Of course, they don’t have to because they are allowed to make decisions like these. Either way, if a contributer wants binary-only stuff out of the kernel, it doesn’t matter who wants it in there.
“Finally, when are we gonna get a game API that just works, like direct x?”
When you decide to stop complaining and finally code one yourself.
“Damn, if only linux had more games, I’d never boot windows. Instead, I still log on to win2k, cuz I don’t have time to hobby OS with Linux, etc.”
Umm…unless you get payed to play games all day, then I would consider that a hobby. And since you boot into Windows to enjoy your favorite hobby(games), wouldn’t that make Windows a hobby os in your case?
People work on kernels and write code for various prjects because they enjoy it in their free time or because (sometimes) they are paid to do so. Unless you are paying them, you have no particular right to expect them to implement your personal agenda. Their work is a gift. Take it or leave it.
Your statement would have merit in 1991 to 1994. Greg KH is paid by IBM to write linux code. Sorry, but he’s not doing USB support to scratch his own itch.
The Linux development train has moved out of Linus’ and Greg’s bedrooms to a multi-billion dollar business with a lot of interests that need to be addressed. This is no longer Greg’s hobby and he cannot arbitrarily impose his will.
best regards
Dev Mazumdar
Where does it say that this is legal issue?
PS. I wish I could run all the Linux userspace apps on an NT kernel. Would solve so many problems without going back to the crappy windows interface.
What kind of BS is this?
‘STRICT’ OSS idealogical politics?
Let me quote Havoc Pennington:
“I think a lot of people feel that the open source world is being unreasonable in insisting on open source for all dependencies…It’s simply the premise of the GNOME organization that we’re building an open source desktop. That’s the whole point of the undertaking; otherwise we’d use Windows or Mac or BeOS. If you don’t stick to the premise of the organization, you don’t have a reason to exist; in this case we’d simply become what we’re trying to replace.”
He was talking about GNOME, but his same arguments apply for the Linux kernel. The Linux kernel is meant to be an FOSS kernel which promotes FOSS! Telling the Linux kernel developers that they are unreasonable for wanting to keep the GPL is like going to a closed source app company and telling them they are unreasonable for making a profit.
After all if the developers wouldn’t care if its strictly FOSS or not, they could of gone and worked on BSD instead or something like that, right?
Also, its not like the module can’t still work as an external module, its just that he dosn’t want it to be an external module. Meaning no actual fucntionality is lost. So now basicly, what is happening here is that the kernel developers don’t want there to be non-free kernel in thier FOSS kernel that they ave worked so hard on creating a FOSS kernel? What a news flash! In tommorow’s headlines, the BSD kernel dosn’t want non-BSD code in its kernel.
“One of the biggest pet hates I have with Linux is that the core kernel developers insist on open source drivers”
You think that’s out of license purity? The kernel maintainers do not like non-open source drivers because they can’t see the code. If that binary only driver makes the system unstable whereas users file a bug report that creates FUD.
“and they change regularly change the binary driver layer to force manufacturers to update driver packages”
I have no idea on this, but i see Rayiner Hashem gave an explanation on page #2.
Reminds me of the Atheros wireless driver. The major part of that is closed source. It actually HAS to be closed source to comply with various legal requirements on radio signals. However there are plenty of folks that would say the driver shouldn’t be used because it isn’t open source. Sad, but true.
It seems to me that a bad driver directly hooked into the kernel could cause all sorts of problems.
Welcome to the land of the monolithic kernel, any loaded module can crash the system should it want to (Note that I’m not saying another kernel scheme is any better, just pointing out the facts).
Well lets quick dump it just in case we might lose it!!!!
What a dumb mentality.
However, not being 100% clean is a good reason to remove anything. Hacking up the kernel to retain binary compatibility is the fastest path to the dark side. Microsoft can barely limp along doing that sort of thing, but they have 10x as many people hacking on Windows. The kernel developers have better things to do.
I would agree if this hook was creating big problems, but obviously it wasn’t anything major or they would have removed it years ago.
And Microsoft is barely limping along? Please.. Microsoft has a lot of problems but the kernel is decent. (at least from a black box point of view). It’s stable, 99% of hardware works with it, and there’s no huge gravitating problems for most users.
Also this is not hacking up the kernel to retain binary compatibility, this is not touching a couple lines of code that enable a ton of users to actually USE that popular camera they bought.
Nowhere. I read it wrong. It’s not unpossible either though, and depends on how you see it (e.g. “author wants to release as GPL but isn’t allowed to by Philips). But you’re right, based on the article one can’t reasonably be sure its a legal issue.
The rest of my comment, namely about the rant and about the lack of objectivity because the other side of the story is not covered still stands.
What’s the difference between having an OSS driver with hooks to a binary driver and an OSS driver loading a firmware like there are a bunch right now?
You have to ask the question.
Is the webcam driver critical to the installation of the machine?
No.
Could a driver be added later?
Yes.
If you truely want a way to impress would be for the necessary drivers to be in the kernel. The rest would need driver disks and/or automatic downloads from the internet.
Either way I could care less about a webcam driver in the kernel. MB, Keyboard, Mouse, Video, Hard Drive, CDROM, Network are the critical peices.. The rest is purely fluff.
http://kerneltrap.org/node/view/3024
And I quote:
> USB: mark pwc driver as broken, as it is.
>
> And the maintainer doesn’t seem to want to fix it
I use this driver with my PCVC690K and it’s great – although practically useless without the closed source PCWX module.
I hope a few people have got recent mirrors of this somewhere.
I can understand not having it in the kernel if it needs a closed source hook, but I can’t understand the author not just making the whole thing a module, like the ATI/NVidia drivers.
I think he’s semi-rightfully throwing a hissy and the kernel guy is being a bit too much of a purist.
I’ve seen way too much good software bite the dust because of this kind of bureaucracy (or is that BS?)
I would agree if this hook was creating big problems, but obviously it wasn’t anything major or they would have removed it years ago.
So basically, you say here:
1) Something doesn’t create big problems
2) Because of 1, It Exists ™
Excuse me?!?
How about this: the issues of binary-only drivers are only recently dealt with. Far more accurate than your point of view. One indication is the NVidia problems (e.g. 2.6 incompatibility, 4k stacks, random crashes the Linux maintainers were unable to deal with) and another one is the new “taints kernel” message to deal with this.
Btw ATI released 2D drivers for about all their (newer) Radeon’s. They’re actually cooperating from that kind of view. (OSnews doesn’t report it.) SGI contributed something interesting to kernel, “Comprehensive System Accounting” (i’m still learning about it). (OSnews doesn’t report it.) I’d like to see some different Linux news on this site instead of “Linux is (not) ready for the desktop”, “Yet Another Desktop Review”, or “yet another (anti) GPL rant”. Those are getting old.
It seems to me that a bad driver directly hooked into the kernel could cause all sorts of problems.
Sure, but I think the warning that currently exists is good enough. If you load a binary driver like the NVidia one it will warn you that it isn’t GPL. Great, proceed at your own risk. But actively fighting against people trying to develop and use binary drivers is just stupid.
If a user is still stupid enough to complain on the lkml about a bug in a binary driver they can just ignore it. I seriously doubt many people would complain there. Either they are novices and will therefore complain to their distro support, or they are advanced users and can track down the manufacturer of the binary driver and complain to them.
By the way, in Windows we have thousands of binary drivers, each capable of crashing the system, and yet Windows 2000/XP/2003 is still stable.
What’s the difference between having an OSS driver with hooks to a binary driver and an OSS driver loading a firmware like there are a bunch right now?
Read the comments, and you know. In theory, all the advantages of the GPL apply. In practice, the difference is kernel developers who would be able to look into problems of the binary only driver are not allowed nor able to do so. When a user has a problem, and they are using a binary only driver, that is a problem because the problem could lie in the binary driver, or it could have to do with that.
[/i]Thats why corporate interest in Linux is weak. The GPL is viral. The only corporate backing it gets is to ride the popularity bandwagon. With the GPL its basically all or nothing.[/i]
Well gee.. This is like timewarp back to late 90’s. I didn’t know people still say stuff like “GPL is viral”. I’m actually thinking this sorta like slashdot’s BSD is *dying* troll. Anyway I can think of whole bunch of companies that would diagree with you: IBM, HP, SGI, Novell, Computer Associates, Sun, Cisco, Oracle, Sybase. That’s just off the top of my head. Every one of these companies have some stake in success of Linux. All of these companies have lawyers who can read and understand what GPL is and they are all in Linux business.
Here are some links:
http://linuxtoday.com/news_story.php3?ltsn=2001-07-03-007-21-OP-CY
http://www.newsforge.com/software/04/02/12/157253.shtml
By the way, you might want to actually read what GPL says:
http://www.gnu.org/licenses/gpl.html
Also you might be interested in LGPL:
http://www.gnu.org/licenses/lgpl.html
I do not recall Linux ever having a binary layer for closed source modules. Infact I believe Linus is against it, because it would 1) make it easier for companies to use closed source modules, 2) it would always be changing.
And finally, Philips, since a great product of them will not be supported anymore; Linux support may not make them rich, but hey, they’ve given us a chance!
Honestly if Phillips corp gave a damn about Linux they would pay the guy to maintain this driver or release one of their own.
Crying about having to make a patch for a multi billion dollar corperation just doesn`t wash with me, you either support OSS or you don`t.
So basically, you say here:
1) Something doesn’t create big problems
2) Because of 1, It Exists ™
Excuse me?!?
No. I meant it is not creating big problems because it has existed for 3 years with no big problems. In fact, searching the linux kernel archives for messages relating to the closed source pwcx module turns up only 1 or 2 problems related to it that were mistakenly reported to lkml instead of the pwcx author.
So where’s the big problems that require removing support for this driver?
it are legal problems. Which was what i suspected, because they’re not deleted hooks for NVidia and ATI.
Here’s a link to the original discussion:
http://lkml.org/lkml/2004/8/24/280
On the left you can browse.
With that, there’s at least less bias than the current situation. Please consider reading the discussion instead of (only) the conclusions based on it.
Could a driver be added later?
Yes.
Does the driver author have unlimited time and patience to work around problems introduced for no good reason?
No.
He GPL’d one component and then distributed another component that links to that component as binary only. Now he isn’t supplying the binary. It is only a matter of time before someone writes him an email saying “please send me the source code to pwcx.o”. Then he’s screwed. He isn’t legally allowed to give the source to pwcx.o and he isn’t legally allowed to refuse either. He’ll write back that he’s not allowed to give the source out and that people will forward the email onto the FSF who will happily pony up the cash to drag his ass through court. He should have made the binary 100% open source or 100% binary (as the linux kernel license permits binary drivers). By dristributing something as 50/50 he’s forcing himself to give out the source code to the binary only part. The court will rule that he is required to violate his NDA. Philips will sue his ass for revealing a trade secret and that’s when he’ll be declaring himself bankrupt.
Of course, what could also happen is everyone will shrug and say “gee, didn’t really use that camera I bought for $99 anyways” and get on with their freakin’ lives.
This means everyone of us was trolled by the “My Take,” in the original article.
I can understand not having it in the kernel if it needs a closed source hook, but I can’t understand the author not just making the whole thing a module, like the ATI/NVidia drivers.
I think its more a matter of time and patience. It’s not that he couldn’t make it a module, just that it’s a lot of work for no gain in usability of the driver. This is one volunteer, not the driver team from NVidia.
Thanks for the link.. That’s very interesting. Not only the legal stuff, but also that the NDA for the pwcx module has expired already…
Perhaps there is some hope of getting this driver back after all.
I don’t pretend to understand the GPL enough to know if greg’s claims of illegality are true. Could someone confirm this? If it is indeed contrary to the GPL of course I withdraw all my complaints.
He isn’t legally allowed to give the source to pwcx.o and he isn’t legally allowed to refuse either.e’ll write back that he’s not allowed to give the source out and that people will forward the email onto the FSF who will happily pony up the cash to drag his ass through court.
That’s not how it works. He *is* allowed to refuse requests for the source code. First, the pwcx module was never licensed under the GPL. Second, the only way to take a software license violator to court is to accuse him of copyright infringment. There’s only one person who has standing to bring copyright cases under US law, and that is the copyright owner. Not someone who wants a copy of the source, or RMS, or the FSF, or the EFF, or anone else. Unless the pwcx author wants to sue himself, there can’t be any court cases about him not living up to any of his license terms.
I’m just wondering whether Eugenia or the submitter looked into this in more detail, e.g. checking the LKML or asking Greg K-H what the kernel maintainers’ side of the story was; or whether they just read the page and decided to run this submission with only one side of the dispute represented?
Looking at the LKML archives, I see very little about this, except for occasional issues with the driver being broken and the kernel maintainers not being able to get the driver maintainer to fix it. If that’s what was actually the case, I can definitely understand their no longer going out of their way to facilitate this one driver. Of course, I don’t know whether that’s what’s really going on or not. It’d be nice if this submission could have told the story from both sides, though. Perhaps then what’s actually going on here would be clear.
No. I meant it is not creating big problems because it has existed for 3 years with no big problems. In fact, searching the linux kernel archives for messages relating to the closed source pwcx module turns up only 1 or 2 problems related to it that were mistakenly reported to lkml instead of the pwcx author.
So where’s the big problems that require removing support for this driver?
“In fact”? Lets proof for once and for all its fact! Begone with the doubt.
Are you sure you did a correct search? Perhaps you would like to provide the links to those reports? Then i’ll try to find some more you missed. I’m up for the challenge.
Here’s one.
There are some severe bugs in either your Philips webcam driver, the USB stack or the combination of both, resulting in a “dead” camera within a second of use in some situations. This can only be fixed by a power cycle (reported to you several times btw). — Rob van Nieuwkerk
http://lkml.org/lkml/2004/8/25/310
[…] also that the NDA for the pwcx module has expired already…
Indeed. So wether this license could -from a legal point of view- be contributed as GPL to the kernel exists could be regarded as yes although people don’t seem to agree wether the expiration of the NDA is enough, or not.
(Readers, see the details on LKML.)
“Not having the complete source available makes it unlikely that these problems will be solved (nothing improved wrt this the last years).”
http://lkml.org/lkml/2004/8/25/310
Now what was I saying earlier? Oh yes, this must be it;
“This is why I am VERY careful about what hardware I buy to use with Linux. I have never had an issue such as this, since I don’t rely on hardware manufacturers who refuse to release the specs.?”
“[i]f I can’t be 100% Free, I’d take that 50% any day. Common users (the kind of mammal species that the Linux platform is trying to captivate away from Windows/Mac) want cheap and easy solutions, and some Linux maintainers fall short in materializing this need because of this or the other (irrelevant to most users) reason.”
I’m sure this post will be removed, but I have to say it: Thank the $DEITY that it isn’t up to you.
The 100% Freedom is the crucial distinction that makes GNU/Linux what it is, and it’s an incorrect assumption to believe that the kernel developers are trying to captivate Windows or Mac users.
Bah, they did the right thing. It’s free or it isn’t, there is no middle ground.
> Thats why corporate interest in Linux is weak.
> The GPL is viral. The only corporate backing it gets
> is to ride the popularity bandwagon.
> With the GPL its basically all or nothing.
Oh my God, Linux is viral?! I use Linux? Is it terminal? How long do I have to live?
Seriously though, it’s not an unsurmountable problem for the driver writer. He could put the hooks back in himself since the kernel is GPL. And if there are enough people interested in what he’s doing, someone else might be willing to do it for him. (Sounds like there could be a good chance of this.)
Unfortunately though, if the driver writer does stop maintaining his code, other people won’t be able to keep working on it, because it’s not GPL.
I think more than anything this serves as a reminder not to rely on non-GPL code, you never know when the author/company etc. is going to stop supporting it.
—
James G
PS. I wish I could run all the Linux userspace apps on an NT kernel. Would solve so many problems without going back to the crappy windows interface.
Perhaps you are interested in: ReactOS (Windows implementation in development), VMware (virtual machine), coLinux (runs Linux on Windows, free), Cygwin, Windows Services for Unix or one of the man similar projects.
There are some severe bugs in either your Philips webcam driver, the USB stack or the combination of both, resulting in a “dead” camera within a second of use in some situations. This can only be fixed by a power cycle (reported to you several times btw). — Rob van Nieuwkerk
Nowhere does it say that these bugs were reported to the lkml and annoyed the kernel devs. Where’s the links to people complaining on lkml?
I’m saying that in 3 years, there haven’t been many reports _to_ _lkml_ about the pwcx module. Of course there’s bugs in it, but it doesn’t matter as long as they get reported to the right place (the author).
After reading the whole thread I still think greg’s actions are incredibly unfair to Nemosoft. Sure Nemosoft is pissed, but with good reason! The only acceptable reason for this would be if it was illegal to have the hook from the GPL’d pwc module to the closed pwcx. Greg nebulously mentioned something like that but gave no proof or explanation.
The only hope I have is that the pwcx source will eventually be released since the NDA is now expired. But I guess if nemosoft doesn’t want to then it won’t happen.
“This is why I am VERY careful about what hardware I buy to use with Linux. I have never had an issue such as this, since I don’t rely on hardware manufacturers who refuse to release the specs.?”
Yes, I try to do that too, unfortunately, in the class of webcams, if you want to get a good one, you go for the Logitech QuickCam (uses this pwcx), a high end phillips (uses pwcx), or an Apple iSight (poorly supported). Yes there are a couple others but they are either poorly supported in linux or no longer sold by the company.
I agree with Nacs, the kernel itself needs to stay 100% pure.
No propriatory junk in the kernel, please!
If people can’t see the surce code then there is no OSS.
Soon, we’d have Windows like Linux full of propriatory junk.
And nobody could fix anything because they couldn’t work with the source code.
Hardware drivers can be compiled and installed as modules just like Nvidia driver does it.
It works well and doesn’t affect the OSS nature of the kernel.
Although one can, and some do, argue that once propriatory driver is installed then the kernel is poluted.
But at least one had always the choice to remove the ofending driver at will.
Eugenia: >> “f I can’t be 100% Free, I’d take that 50% any day. Common users (the kind of mammal species that the Linux platform is trying to captivate away from Windows/Mac) want cheap and easy solutions, and some Linux maintainers fall short in materializing this need because of this or the other (irrelevant to most users) reason.”
Dave: > I’m sure this post will be removed, but I have to say it: Thank the $DEITY that it isn’t up to you.
The 100% Freedom is the crucial distinction that makes GNU/Linux what it is, and it’s an incorrect assumption to believe that the kernel developers are trying to captivate Windows or Mac users.
After having investigated the other side of the story, i fall on cheap and easy. Apparantly, according to Eugenia, stability is of no concern to the users. Ofcourse, we all know is is. Had she investigated the other side of the story, she’d have taken a more pragmatic point of view on the issue taking the stability aspect into account.
Perhaps you are interested in: ReactOS (Windows implementation in development), VMware (virtual machine), coLinux (runs Linux on Windows, free), Cygwin, Windows Services for Unix or one of the man similar projects.
ReactOS – Not ready to use
VMWare – Virtual machine, not an OS
coLinux – This is linux on top of windows, so now you have the worst of both worlds
Cygwin – Yeah I guess. Good for running linux apps but you still have to load all of windows to get there
I wish there was a way to rip out the Linux kernel and replace it with the NT kernel without changing anything else. Hehe. Everyone has their pipedreams I guess
I think my best bet will be if DriverLoader (http://www.linuxant.com/company/press_dldr.php) or something similar ever gets extended to support all drivers.
Nowhere does it say that these bugs were reported to the lkml and annoyed the kernel devs. Where’s the links to people complaining on lkml?
I’m saying that in 3 years, there haven’t been many reports _to_ _lkml_ about the pwcx module. Of course there’s bugs in it, but it doesn’t matter as long as they get reported to the right place (the author).
The bug i quoted from Rob van Nieuwkerk is extremely severe. As you have also read in that message, he also says nothing has changed the past 3 years. I assume the bug got reported to the author precicely because of the severity. How am i able to proof my assumption?
However, another question to wonder about is: is this risk to damage hardware worth taking for innocent (?) users who end up with broken hardware or should we take precautions?
Before i’m gonna search, i’d like to know what you exactly found. I’d like to know what you’ve find out first, as i already asked in a previous post. I stated that, so that you cannot apply your same conclusions you stated here on different and/or multiple sources.
Yeah, i agree, they all got (possibly negative) side aspects. There are projects which help you to strip down Windows. If you’d use one of these in conjunction with ports of “Linux” applications for Windows then perhaps that would make you happy.
Apparantly, according to Eugenia, stability is of no concern to the users.
No, it is of concern to users. You are just mistaken in the view that binary drivers == instability. My linux box with pwcx and the binary NVidia driver is stable. The one time it wasn’t it was because of a bug in the 2.6.5 usb code (yes I tested it without any binary modules, didn’t change a thing).
My windows install is just as stable with all binary drivers.
I don’t think all NDA’s should be open ended durable. meaning, I believe most should have a condition or experiation.
There is few cases where an NDA is required for a very long time, but there are good cases.
The phillips driver? Phillips should have open sourced that whole driver about 4 years ago. Intellecutual Property is supposed to be a tool, not a slaving whip.
The number of complaints that “My webcam stopped working!” when someone’s distro upgrades is going to be hiliarious!
Do we need an Intellectual rights court of arbitration?
The bug i quoted from Rob van Nieuwkerk is extremely severe. As you have also read in that message, he also says nothing has changed the past 3 years. I assume the bug got reported to the author precicely because of the severity. How am i able to proof my assumption?
You don’t understand what I mean. I don’t deny that it is a serious bug, it doesn’t matter. The argument taken by most against binary drivers is that people using them will report kernel crashes that are caused by the binary drivers to the LKML. I agree that the kernel devs shouldn’t be bothered by problems caused by binary drivers. I was just saying that the vast majority of people do not harass the kernel devs with crash reports that are caused by binary drivers. So having people use binary drivers doesn’t impact the kernel devs significantly.
And anyway, you act as if an old crash bug is unheard of in open source. There’s tons of bugs like this in open source projects that just haven’t been fixed because the devs didn’t have the time or they couldn’t reproduce the problem.
Kudos to the kernel developers – they have done the *right* thing. This is the only way to stop hardware manufacturers giving Linux a bum steer with driver support. The sooner Open Source makes it’s own hardware (yes, you read right) the better. Then we can use that, have it work perfectly with our favourite o/s and *iss off the crappy hardware vendors who don’t care. They’re loss, our gain. I keep saying this, the only thing that hardware vendors and software developers care about is loss of money from loss of sales. Then they sit up and take notice. Put your money where your mouth is and don’t support these types of companies. Period.
Dave
They stuck to their guns and didn’t back down.
Binary modules *ARE* bad. The sooner H/W manufacturers understand that the Linux kernel-hackers won’t be manipulated, the better.
However, another question to wonder about is: is this risk to damage hardware worth taking for innocent (?) users who end up with broken hardware or should we take precautions?
What does this have to do with closed drivers? Off the top of my head, the modules that have damaged hardware that I know of are open source.
Before i’m gonna search, i’d like to know what you exactly found. I’d like to know what you’ve find out first, as i already asked in a previous post. I stated that, so that you cannot apply your same conclusions you stated here on different and/or multiple sources.
I don’t claim to have done an exhaustive search, I just typed “pwcx” into the search for the lkml archives here:
http://marc.theaimsgroup.com/?l=linux-kernel&r=1&w=2
And got this result:
http://marc.theaimsgroup.com/?l=linux-kernel&w=2&r=1&s=pwcx&q=b
30 results back to 2001.
Of these results, the first 4 are related to the current issue. The only ones that I can tell that are problems explicitly related to pwcx are here:
http://marc.theaimsgroup.com/?l=linux-kernel&m=107196984709257&w=2
http://marc.theaimsgroup.com/?l=linux-kernel&m=108246186124456&w=2
http://marc.theaimsgroup.com/?l=linux-kernel&m=101530580900914&w=2
The second message received no reply, and the third bug report was reproduced without the pwcx module loaded.
So in 4 years, the kernel devs had to deal with ONE valid report. WOW! Look at that flood of bug reports because of a binary driver!
Please.. Prove me wrong.
Binary modules *ARE* bad.
Why? Asterixes do not constitute an argument.
Sure is bad how I can use my hardware at it’s full potential instead of with severely crippled features like I could have with a 100% open source approach. Read my post above for why the argument against binary drivers doesn’t hold up.
Posted on the LKML:
Everybody should never forget, as history tells, that extremist positions quickly lead to destruction.
I hope that open source movement will never become fundamentalism.
Amen
You are just mistaken in the view that binary drivers == instability.
Actually, no. I’m using one myself (ATI’s proprietary driver). Great for when i play Quake2 (i have another IRIX computer which plays it as well both with good performance so imagine its fun to do cooperative style or play on the Internet in the weekends). In short, this is not my position.
And anyway, you act as if an old crash bug is unheard of in open source. There’s tons of bugs like this in open source projects that just haven’t been fixed because the devs didn’t have the time or they couldn’t reproduce the problem.
Not my point of view either. I’m aware there are bugs, and applaud usability efforts such as this http://kerneltrap.org/node/view/3695 (sadly, not reported on OSnews.)
To clarify my point of view,
* Binary drivers have lead to changes in the kernel. I have already stated 2 earlier implications in the discussion.
* Binary drivers do lower innovative changes or make them impractical for some users. The 4k stack problem and ABI changes are examples of this.
* What matters is not a majority of people complaing, what matters is the signal-to-noise ratio which did lead to changes in the kernel. IOW: the kernel developers decide on what obstructs their work. Their perception matters; not ours. They try to work on the kernel; not we.
PS: for the 3rd time where is your research / findings on the issue of stability of the PCWX drivers in question? Where are the references?
Thats why i quit using Linux as my primary OS, either the driver I needed was unavailable or it was buggy as hell. I quit trying to guess and cutting myself short on hardware. Sometimes for use with Linux I would have to choose cheap hardware because it was the only thing supported. With Windows XP and Server 2003 my hardware works great 100% of the time.
No, the looser is Philips, they just lost one customer.
* Binary drivers do lower innovative changes or make them impractical for some users. The 4k stack problem and ABI changes are examples of this.
Of course, if there’s an open source and a closed source driver for the same hardware, then open source is better. And yes, binary problems, in general are not as nice to use as open source drivers. BUT, and this is a big one, for hardware that has shoddy support with the opensource driver, the binary driver is the ONLY choice. So the choice is not between open source and impractical closed source, the choice is between nothing or closed source.
By the way, binary drivers lower innovative changes? Bullshit, drivers aren’t about innovative changes, they’re about complete, fast access to hardware. All the innovation in the world is not going to make the nv driver better than the nvidia driver, or the pwc driver better than pwc+pwcx.
PS: for the 3rd time where is your research / findings on the issue of stability of the PCWX drivers in question? Where are the references?
For the first time, hit refresh.
If Linux is touted as 100% free you cannot and should not allow any proprietary apps – which means you cannot and should not push Linux in a situation that requires the use of an Oracle or a Websphere application. You have to tell IBM and Oracle that they must open source their apps or just run them on closed source UNIX operating systems like Solaris or AIX.
Why is it ok to have an application that can be closed source but not the kernel?. Where does your set of principles get compromised?.
As I’ve said before, the Linux train has left the train station and people still hung up on philosophy of free software need to get pragmatic – people say linux wouldn’t have gotten where it’s at without being open source, I can claim that it’s the closed source guys who have made linux – Oracle, IBM, Sun, Novell, Intel, HP are all closed source guys. Redhat was able to got IP because of investments by closed source companies like Oracle, IBM and Intel.
Best regards
Dev Mazumdar
The kernel must be protected against LEGAL THREATS ! Sure it may not be a problem today but who knows down the road things may change. I am glad they pulled this driver. Also the guy who puts it out is nothing but a big baby. Grow up and patch it and make it a module like NVIDIA or ATI !
> Why is it ok to have an application that can be closed source but not the kernel?.
> Where oes your set of principles get compromised?.
You are confusing the kernel with an Operating System. Linux isnt an Operating System, it’s a kernel. What is allowed to be *part of* the kernel and what the kernel is allowed to be *used for* are two completely different things.
While I agree that the big corps helped linux a lot, they did not make linux.
Without the big companies, linux would still exist, without the open source hackers, it wouldn’t.
Their job is not to appease the average user, its to scratch their itch, and build a free (as in freedom) OS that accomplishes their goals.
It is simply not to provide a Windows replacement for the masses. If it evolves into that one day, that will be just fine, but I’m not holding my breath and don’t really care – Windows and Microsoft are utterly irrelevant to me, and whether current Microsoft users switch to Linux or not is of no consequence to me.
I don’t know where this whole ‘how do you expect people to switch to Linux if…’ comes from – I dont expect anyone to switch to Linux. Stay with what works for you right now.
All the people that are bitching about how the current kernel maintainers’ philosophy negatively impacts on the computing experience of the average user should band together, fork the kernel or use a FreeBSD kernel or similar, and build a better system – to show us the way it should be done.
Oh but you don’t want to do that – you want somebody else to whore out their beliefs and principles out so you can be fat and happy.
Microsoft has already done that for you, so I don’t see what’s left to complain about – it’s not like you don’t have a choice.
The kernel must be protected against LEGAL THREATS ! Sure it may not be a problem today but who knows down the road things may change.
Do you have any more insight into the legality of this hook? I’m genuinely interested if there is fact behind greg-kh’s claim.
Now you are just being childish here and grasping at strawman arguements. There is a big difference between a userland application/ and the CORE OF THE OS..IE THE KERNEL ! If a company turns around and says you can’t use their application for free anymore and wants to charge you for it then you just dump it and create your own application to fill the void. You can’t do the same with a heart and sould of the Linux OS the kernel without making major overhauls and bringing the OS and it’s development down to a crawl ! If you can’t see the difference between the two then there is no reason to argue with someone as blind as you.
Well time to crash for me, I’m sick and have to work early tomorrow. I’m interested to see this evidence of noise in the lkml because of the pwcx driver that you say you will dig up.
The OSNews article from the 25/8/04. Phillips does not care about Linux users of their products otherwise they would have released either a better driver or all the specs. Phillips is the one who “sells” the product and it is they who the complaints should go to.
If Phillips released a crappy driver for Windows how would users react? When they release nothing for Linux how do you expect users to react?
The same as Windows Users?
Or some other way?
If Phillips released a crappy driver for Windows how would users react?
Poorly, and phillips would soon have to withdraw from that market.
When they release nothing for Linux how do you expect users to react?
Poorly, and phillips will be largely unaffected.
You can’t really fault hardware manufacturers for not catering to the demands of linux users. Smart companies don’t spend time on things that do not result in profit for them, directly or indirectly. Good luck changing that reality.
Are you implying that Linux users don’t buy or pay for hardware ? Should they not be supported if they are paying for a piece of computer hardware ? Creating a simple module and adding it to their drivers cd or allowing comercial distros to include it via a liscensing deal wouldn’t hurt them one bit.
good for him, bad for Linux in general, particularly the kernel maintainers. Every little sting a dedicated developer and corporation takes, puts us one step back.
Man, i got to say that is very sad. Not because of the hook or not to hook, just because there’s no other option for the user. This guy is gonna quit, and Phillips itself isn’t going to develop a stand-alone binary for the webcam. I think NVidia got it giving their own driver. Maybe thats the way it has to be, just to avoid another diseagreeable episode like this. The kernel? well, just keep it ready for the “outsiders”(company’s binary drivers).
God Bless U all.
The guy signed a NDA with Philips and including hooks into the kernel would put the linux kernel in a position to be attacked by this company if they wanted to. Also fellow kernel hackers could sue for the breach in the GPL license since they have no access to this hook or driver. Not to mention that if you include this then what else should be allowed ?? Where do you draw the line ? It just slippery slope from there on down if you allow this into the kernel. Like others have said it’s not about attracting users or about supporting hardware vendors, it’s about keeping the kernel safe from legal threats and open to all who want to contribute and share code.
If Philips wants to create a closed-source driver then they can create a kernel module like NVIDIA and ATI and retian their intelectual rights. Nothing is stoping them from doing this to begin with if they truely wanted to do support Linux users.
No I am implying that they DO pay for hardware and thus should expect at least some level of service from Phillips. Unless Phillips makes the claim that the webcam is Windows(TM) only, then they should provide some level of service to alternative operating systems on the PC.
That is also why I do not buy Hardware that does not work with Linux. Because as a informed consumer I know when a company is giving me the service I need from their products and in Linux that is in the form of driver support. Phillips should either offer driver transperancy ( the specs ) or a binary driver module like Nvidia does. If they do not then they should make sure that their product lists the OS’s it supports and that support is only for them. Since I have nver bought a webcam and probably never will this is irrelevant to me, but if I did I would check for compliance first.
Can you guarantee that no malicious user (SCO sponsored, say) will not “taint” the kernel so that all further users of linux will have to pay the SCO-tax?
As long as the kernel is kept free of all proprietory code, we don’t have to lose sleep over if the OS we all have come to love will be hijacked by a nasty corporate such as SCO … or even Microsoft!
How is this application different from other stream applications like RealPlayer which also has a binary compressor/decompressor stage but not this problem.
The losers from the whole story, are Philips. No linux users will buy thier webcams. And I prefer useless webcam then non-free driver!
Can you guarantee that no malicious user (SCO sponsored, say) will not “taint” the kernel so that all further users of linux will have to pay the SCO-tax?
You must have gone to the same corn-flake-box come law school as the boneheads at SCO who wrote that little piece of FUD. Say Ford steals the designs for a new water pump from Toyota. Suppose Toyota finds out. Who do they sue? Ford. Who has to pay Toyota if they lose the case? Ford. The idea that everyone who owns a Ford which contains Toyota’s water pump design would have to give back their car or pay anything to Toyota ever is not only wrong it is stupid.