Looking Deeper at the LKP -- Did SCO Copy Linux Code to Open UNIX?

Wednesday, September 24 2003 @ 12:36 AM EDT

Contributed by: PJ

One unanswered question in the SCO vs. Linux conflict is the allegation that SCO directly copied Linux kernel source code into the Linux Kernel Personality feature of Open UNIX. eWeek first raised the question in an article printed in June, "Did SCO Violate the GPL?". The article is still available here.

In the article, an anonymous source "close to SCO" who spoke on condition of anonymity said that "parts of the Linux kernel code were copied into the Unix System V source tree by former or current SCO employees."

Blake Stowell was quoted as denying it: "SCO also never used any of the Linux kernel code in the LKP and thus has not violated the GPL."

First, code might be in other places than the LKP itself, but leaving that aside, let's see what some newly discovered evidence, found by Groklaw reader lightsail, indicates as to who told the truth. He directs us to a regional SCO website, SCO Benelux, the regional office for Belgium, The Netherlands and Luxembourg, where you will find a Caldera Powerpoint slideshow by a senior Caldera engineer entitled,"Linux Kernel Personality Internals -- Making a UNIX Kernel Prozac-Ready."

SCO management may be reaching for the Prozac when they see these slides have been discovered. You won't find them on the main SCO website. There are some particularly interesting slides. Slide number 30, for example, says this:

"Open UNIX can run Linux applications and binaries without recompilation, and without marking the binaries in any way

"All it takes is around 40,000 lines of code to build:

- A kernel module for handling system calls
- Three totally new Open UNIX filesystems
- A new system daemon
- And the infrastructure necessary to install all of these, and the whole of Linux, onto Open UNIX"
[emphasis added]

Install the whole of Linux onto Open UNIX, you say?

The presentation includes several other interesting points:

1. Slide 6 shows UNIX source code. Excuse me. Isn't that supposed to be such a deep, holy trade secret they can't show any of it even to allow Linus to remove any allegedly infringing code? Yet, there it seems to be in its birthday suit, for all the world to see. This slide is also interesting because is shows a side-by-side comparison of UNIX and Linux source code. So, developers of LKP were using Linux source when developing LKP. Did they copy? Or simply view code as needed?

2. To help us find the answer, consider Slide 11, which states “copy this file to UNIX” when discussing the file: /lib/ld-linux.so.2. Doesn't that make it appear the LKP developers were copying file directly from Linux to UNIX? Otherwise, what does that sentence mean?

3. Slide 12: “To work, this needs a full Linux distribution in /linux”. The kernel is part of a full distribution. Does LKP contain a full Linux distribution in the /linux folder of Open Unix, as the slide says it "needs" to?

Here's hoping IBM asks to see the source code of LKP and Open UNIX as part of the discovery process, to get to the bottom of this question. It has been reported in the news that SCO has recently altered the LKP code, so it would be important to gain access to it prior to the changes, as well as to its current version.

So, take a look at the slides, especially you Linux kernel coders, and tell us what it looks like to you. I am a researcher, not a programmer, and lightsail says he isn't a kernel coder, so he asks for your input:

"On the face of this, it seems that SCO was quite willing to 're-use' any Linux source code needed to create LKP. It is my hopes that this information may be a help to some writer of Linux kernel source code, who can cement down the code copied into LKP and answer the question: Did SCO copy Linux into UNIX?"

Thanks, lightsail. If this isn't a smoking gun you found, and by itself it isn't, it is the next best thing, a string to begin to pull.