Did SCO 'Borrow' Linux Code? 688
An Anonymous Reader writes "Apparently someone inside SCO has stated that SCO(actually Caldera) copied Linux code into System V. They did it to build what they now market as Linux Kernel Personality - the ability to run Linux software on their Unix. Now, the open source community(of course they don't mention who) is jumping on this, because they didn't return the changes to the OS community or give the community credit. Of course, SCO says it's a misunderstanding and, get this 'SCO also never used any of the Linux kernel code.'"
This could be very bad for SCO... (Score:3, Informative)
This is exactly the kind of scenario that Cringely pointed out in his latest column about the SCO vs. IBM situation on his PBS.org website:
I, Cringely: Technician, Steal Thyself [pbs.org]
Related past columns:
May 22, 2003 [pbs.org]
May 29, 2003 [pbs.org]
For those of you who haven't already (Score:5, Informative)
Does eWeek's source understand the GPL? (Score:3, Informative)
Sounds as if the source doesn't understand the difference between the BSD license and the GPL.
Re:"Someone inside SCO" (Score:1, Informative)
GPL (Score:3, Informative)
Err... hello? If you distribute it (SCO did) it must be given back; I can't find anything in the GPL which mentions a copyright notice as an alternative.
Re:Two Words (Score:3, Informative)
The original poster has a point, but it has little to do with the article. Please read the article again. This is a new claim. If substantiated, it basically constitutes a large torpedo headed in the direction of SCO.
Re:Get this! (Score:3, Informative)
Re:The other way ... (Score:3, Informative)
You can assign a date for when a line entered linux pretty easily, you just search the kernel list archives. Since they're public and mirrored sufficiently they can be considered reliable - even if someone wanted to fake them it would be impossible to get away with it. You can set a date on calderas end but less easily, because you must have a check - they can fake anything they want since it's proprietary code. Essentially you would need to take the code they claimed came from a certain date, compile it, and compare it with a properly attested binary that actually matches the claimed date.
Note also that this method isn't perfect - they can change the comments all they want and there's no way you could possibly prove that, since it doesn't affect the compiled code.
Re:In two weeks no one will care. (Score:5, Informative)
These limitations of liability are no different that those included in Sun's Solaris BCL, despite SCO's claim above the contrary:
All Sun's guaranteeing is that the media is good and that they'll replace your media if defective. All current Linux distributions that distribute on physical media will, of course, do the same. Sun is essentially saying that they'll give you your money back for the license fee paid if and only if a court determines that it's limitations of liability (read: disclaimer of all liability) are held to be invalid under the law. And the only reason they're saying that is so that their whole limitations of liability (read: disclaimer of liability) doesn't wholesale get tossed out on the basis that it violates contract law. (IANAL, but I did have a business law class
This is the same for Linux, except, in all cases regarding the licensing of the kernel, there is no license fee paid, hence, you get no money back. Duh.
which part of "parts of the Linux kernel code..." (Score:2, Informative)
do you not understand? The article states:
Their source does not understand the GPL (Score:4, Informative)
From the article:
A source close to SCO, who spoke on condition of anonymity, told eWEEK that parts of the Linux kernel code were copied into the Unix System V source tree by former or current SCO employees.
That could violate the conditions of the GNU GPL, which states that any amendments to open-source code used in a commercial product must be given back to the community or a copyright notice must be displayed attributable to Linux, he said.
That would be the BSD license. The GPL requires a project that incorporates GPL code to be GPLed as well, which means the source must be made available to people who get binaries and their rights to distribute the program under the GPL cannot be infringed. If this does not happen the right to distribute the GPL code is revoked and its distribution is therefore a copyright violation. Therefore if SCO really stole GPL code for its Linux Kernel Personality it has a serious problem on its hands.
Re:Congratulation on a USA today reading level! (Score:5, Informative)
Re:YOU HAVE IT BACKWARDS! (Score:5, Informative)
The SCO code in question was in the process of being written and could not yet have been copied into the Linux kernel unless someone on the Linux team had a time machine. I repeat, the article is about a SCO engineer encountering supposedly NEW SCO code that appeared to have been cribbed from already existing Linux code.
Please re-read the following quote from the article and evaluate how it fits with your interpretation of the article:
This is not about old SCO code finding its way into Linux, this is about supposedly new SCO code written to implement Linux kernel functions that looked suspiciously like code taken straight out of existing versions of Linux.
You're misunderstanding the article. (Score:5, Informative)
Yes, this is what the unnamed "source close to SCO" says he was doing -- re-implementing certain Linux kernel APIs in the UnixWare kernel. And you're correct that that's perfectly legal.
But what he's saying is that while he and the other programmers on the project were implementing the LKP, they discovered that portions of the UnixWare kernel were already very similar to portions of the Linux kernel -- to the point of having identical variable names (presumably non-trivial ones), etc. In other words, although the LKP project is perfectly legal, this anonymous source says that while working on it, he and the other programmers on the project uncovered evidence of prior code-copying by SCO.
TheFrood
A lose-lose situation for SCO (Score:2, Informative)
I don't agree that SCO produces "nothing," because they are clearly selling server computers and such on their website. Yet I would agree that they produce nothing unique, new, or innovative (quite like Microsoft... sorry, couldn't help it). I spent a while looking over their "products and services" and I couldn't find a single thing a sysadmin couldn't get somewhere else. And, correct me if I'm wrong, but sysadmins are SCO's main customers, and even more specifically, Unix sysadmins.
Now I got thinking: most Unix sysadmins are Linux and Open Source advocates. At least this has been my experience. So it seems to me that even if SCO won this lawsuit, they would lose most of their customers. As I said before, SCO doesn't provide anything unique, new or innovative in any form, so there are plenty of viable alternatives to their services us sysadmins can move on to. In conclusion: this lawsuit is a lose-lose situation to SCO no matter what. Either they lose the lawsuit and their company collapses, or they win the lawsuit and angry system administrators simply move on to any of the billion other companies that provide exactly the same things they do.
A short note before I wrap this up: I sent a message like this to SCO early on in their lawsuit against IBM, urging them to cancel it quickly. After clicking the send button, I noticed a little message that said something like "Thank you for your input, you will hear from us soon!" I realized at that moment that I was stupid enough to write in my personal email address in that email, rather than my standard spam distraction. Well, I did get a message from SCO very quickly, in a way, for the next day my inbox was littered with spam... something I had never gotten before I wrote that message.
Thanks SCO. Thanks.
No, not microsoft. (Score:1, Informative)
16 million in licensing fees the last
quarter, mostly from two sources. One
of them was acknowledged to be
Microsoft, and the other did not wish
to have their identity disclosed.
Microsoft was estimated to have given
them ~$8.8 million dollars. That leaves
another unnamed player on the stage.
Re:In two weeks no one will care. (Score:5, Informative)
Besides the article you site supports these facts. The $90k was paid to Microsoft. It was afterward, when MS made EB out to be a poster child of licence abuse, that the EB Executives themselves decided to give their own IT department 120 days to become Windows-free.
Real good read, tho!
Re:Proving the code (Score:4, Informative)
Seems to me that even compiling it to prove the binaries are identical will be difficult, whether you are compiling SCO or Linux code.
Re:Two Words (Score:4, Informative)
I found a November 2002 article [com.com] talking about SCO, high end computing work that they had done with Compaq in the clustering arena and a brief touch on LKP.
I found a February 2001 article [vnunet.com] just about Linux and SCO integration and LKP.
I found a 2002 SCO Newsletter [wimal.com.pl] touting LKP.
I also found Simon Baldwin's resume [geocities.com] who has a long history at SCO and who was the "Lead Kernel Engineer and Architect for the Linux Kernel Personality (LKP)" from February of 2000 to "present".
So the LKP stuff was going on quite some time ago. Before or after IBM allegedly put the offending into Linux? Inquiring minds want to know.
Closeley protected? (Score:2, Informative)
UNIX source isn't some close-held secret; everyone's implementation is, of course, but there are tons of licensees; it is certainly possible that "generic" UNIX source could be passed on to Linux for various reasons (stupidity, ignorance, arrogance, maliciousness against either party). If it's SCO's own customisation of what they license to other parties, then it's one of their people (or IBM, of course).
That's all assuming that we believe this anonymous story, which could well have been perpetrated by a slashdot reader, of course....
And then, SCO could be making it all up, as has been said enough times before to be not worth repeating.
Re:Two Words (Score:2, Informative)
This exact possibility HAS been mentioned before NUMEROUS times in other SCO threads. Now if you actually followed them you wouldn't use your "excellent" account to mod me down then comment with your hater. (READ: INSUBSTANTIAL CLAIMS SIMILAR TO YOUR INSUBSTATIAL CLAIM OF MY NOT RTFA)
SO, as I was saying, this WAS brought up before. AND I won't believe this until I see it in court or SCO settles for peanuts.
NOW, if you wanna hate cause I got FP, please do, but don't yell at me for misreading my post.
Hmm. No. (Score:5, Informative)
If they distribute code derived form a GPL work, they aer not BOUND by the gpl; they are BOUND by copyright law, and the copyright holders who's rights are being violated can sue. The GPL is simply something they could cite to demonstrate they had permission to do what they do.
IT's not a GPL violation, it's code theft. There is a difference.
Yes, there is a clause in the GPL that says "By distributing siad work, you accept this license"... but that can only apply if you have READ the license in the first place. So it's dodgy.
You are being mistaken (Score:5, Informative)
There is no need to prove the legality of GPL. If it was invalid, or if someone doesn't want to agree with its terms (because no one has to accept GPL, as it is stated in the GPL itself -- you don't have to sign it, after all) then the only rules there are, are those imposed by the copyright law, which makes it illegal to distribute copyrighted works. If anyone wants to distribute GPL'ed software while saying that she doesn't agree with the GPL, or while questioning its legality, then all she has is a copyright law, which clearly states that what she's doing is illegal. I have posted a comment [slashdot.org] about it, but it has only Score:2, so obviously no one has read it. Basically, if SCO thinks GPL doesn't mean anything from the legal point of view, then, after rejecting the GPL, when they look at the Linux kernel, what they have is just a piece of software with "Copyright (C) 1991-2003 Linus & Co. All right reserved."
Re:The other way ... (Score:3, Informative)
Next time, please credit Bill Hicks (Score:5, Informative)
Article seems bogus (Score:3, Informative)
"We were even surprised by the identical variable names of both codes !!" Honestly, who talks like that?
And socketcall() is listed as an example of a syscall that is "fairly hard to wrap." socketcall() is a cheap socket API syscall... Most UNIXes have socket(), listen(), send(), sendto(), etc., all as different syscalls, but not Linux. Linux has:
int socketcall( int call, unsigned long *args );
"call" is an integer representing which socket syscall. "args" is effectively a stack pointer..
Now... This call is INSANELY easy to wrap. Create a table of function pointers:
void *funcs[] = { sys_socket, sys_connect,
Then, essentially, copy "args" to the stack, and call funcs[call], after verifying that "call" is legal and good... This could be done with a touch of assembler that manipulates the stack pointer. Or, you could do it the long way in straight C, the way the Linux kernel itself does.
This would all be very little work. I don't even know much about kernels and I could do it. For that to be described as "hard" is a bit much.
I can think of a lot of other places in the Linux syscall interface that are much harder to wrap than "socketcall()"... Also, it shouldn't surprise anyone that most syscalls map one-to-one across 2 different types of UNIX systems. It's that way for a reason.
This article is... Somewhat confusing... And by all means, sounds bogus.
Article doesn't add up (Score:5, Informative)
Especially to any one who worked as an engineer at SCO.
My suspicions were raised by the quote:
The source, who has seen both the Unix System V source code and the Linux source code and who assisted with a SCO project to bring the two kernels closer together, said that SCO "basically re-implemented the Linux kernel with functions available in the Unix kernel to build what is now known as the Linux Kernel Personality (LKP) in SCO Unix."
Unless they were a real newbie no engineer who worked at SCO would refer to SCO Unix because this would be internally confusing. SCO had two flavours of unix - SCO Openserver and Unixware.
LKP was really an enhancement on the basis of the lxrun application developed originally by Mike Davidson - I think Sun now doesn't some work with this - also I think it is open source. By implementing a system call trap handler you can implement a system call handling interface for our linux binaries. The LKP was really about making this system call handling and environment emulation more realistic and efficient. You don't 'necessarily' need to put parts of the linux kernel in your kernel to do this.
It is also quite suprising how much lxrun could actually do without all the LKP stuff.
I get the impression that the 'SCO source' didnt really understand how this emulation worked.
This statement is dodgy: Parts of linux were copied in to "The Unix System V tree".
Presumably this refers to OpenServer, it certainly doesn't refer to Unixware. This is inconsistent because UnixWare is OS on the LKP was implemented - and it certainly would not be refered to as Unix System V.
The facts seem a bit muddled to me. It might be that the engineer was telling truth but some facts got lost in translation - and just don't ring true to me. I left SCO before the LKP project was in full flight and I guess they would want to engineer some system calls into the kernel.
There was a lot of crap (usually ignorant or laughably incorrect) on slashdot about SCO before all of this stuff happened. But I do know that SCO had plenty of customers who were very happy with the products and that it was a great place to work.
And by the way, I and no one I've talked to since have seen any Unixware source in the Linux kernel.
Its a massive shame to me that a decent company was taken over by the bandits and shysters called Caldera. It beats me why they had change the company name back to 'SCO' before launching this pointless action.
Re:The continuing saga of SCO's suicide. (Score:4, Informative)
Re:YOU HAVE IT BACKWARDS! (Score:1, Informative)
Furthermore, reading the code while "reimplementing" it is strickly a no-no as far as getting anyone to believe you didn't use it in your program. In fact IBM, the subject of SCO's stupidity, is very strict in this regard. Something that it is obvious that SCO isn't!
Did anybody notice the arrow from Linux 2.2.16 (Score:3, Informative)
Re:Next time, please credit Bill Hicks (Score:3, Informative)
Since when was a simple statement of the facts a joke? Rumsfeld arranged the sale of several types of plague, plus Anthrax and West Nile Virus, to Saddam, so what's the joke in pointing this out?
TWW
Re:It has happened before . . . (Score:3, Informative)
Only true of the newer BSD license: the original contains the "advertising clause", which is incompatible with the GPL. Alan Cox had a problem with this: he wanted to integrate a BSD IP stack (back before Linux had one), but at the time the BSD license wasn't GPL compatible. So, he wrote his own GPLed one and integrated that instead.