RCU, ULS, IPC, ELF and Intensive Care

Message ID: 152989
Posted By: stats_for_all
Posted On: 2004-07-11 17:51:00
Subject: RCU, ULS, IPC, ELF and Intensive Care

Lawyer John Harrop submits a declaration in support of the SCO memorandum against CC10. He makes extensive reference to a SEALED declaration by Sandy Gupta, SCO Engineer on SCO IP in Linux. (Note #1)
IANAGeek, and I would greatly appreciate the skills of others in debunking this list.
This list is extracted verbatim from the Herrop declaration:
.. 1. substantial similarity of the Read-Copy-Update (“RCU”) routine in Linux to a routine in UNIX
.. 2. copying of UNIX System V init (SYS V init) code in Linux version 2.6
.. 3. substantial similarity of the user level synchronization (ULS) routines in Linux and similar routines in UNIX;
.. 4. copying of SCO’s UNIX System V IPC code in Linux 2.4.20;
.. 5. copying of SCO’s copyrighted UNIX “header and interfaces” in Linux
.. 6. copying of SCO’s UNIX Executable and Linking Format (ELF) codes in Linux

This list is carefully crafted to conflate “substantial similarity”, line-by-line copying, and duplication of file names and variables.

1) The RCU element occupies first place in this list. Harrod also indicates they will depose IBM engineers Paul McKenney and Brent Kingsbury to demonstrate derived code.
Web documentation supports an explicitly GPL’d implementation of RCU with code published in 2001 (#2 and #3).

2) I have no information regarding the SYS V init allegation.

3) The Unix style ULS routines were deprecated by engineer Hubertus Franke, who submitted a Linux 2.4.17 patch based on entirely new architecture (#4)

4) IPC code is a required specification by the Open Group (#5). Copying may refer to system codes at the heart of the Open Group’s specification.

5) “headers and interfaces”: I am not sure what is meant here, header file names ?

6) ELF codes, this is a published Unix System Lab standard. Copying here may refer to the published standards for C language functions ? (#6)
Sources:
#1 www.sco.com/ibmlawsuit/20040708_John_Harrop_Declaration_Final_1.pdf
#2 www.rdrop.com/users/paulmck

#3 www.linuxjournal.com/article.php?sid=6993

# 4 www.uwsg.iu.edu/hypermail/linux/kernel/0202.1/0745.html
#5 www.dwheeler.com/secure-programs/Secure-Programs-HOWTO/sysv-ipc.html

#7 www.skyfree.org/linux/references/ELF_Format.pdf

Message ID: 152992
Posted By: nobbutl
Posted On: 2004-07-11 18:15:00
Subject: Re: RCU, ULS, IPC, ELF and Intensive Ca

1. Others have written extensively about the origin of and patents regarding RCU. It originated outside SysV, any line for line similarity is likely due to shared ancestry.

2. There is no SysV init code in the kernel. An external 'sysvinit' package is distributed by Miguel van Smoorenburg IIRC. The init process is rather simple and the invocation of boot scripts is even more trivial. Similarities will be inevitable due to common functionality.

3. The acronym 'ULS' is new to me, I suspect it may be a SCO-ism invented to FUDdify mechanisms that have hitherto been nameless because they are not unique to Unix and date back to the stone age.

4. Lord knows why 2.4.20 should be singled out. SysV IPC is standardised in Posix, I guess we are in non-literal FUD territory here, where form follows functionality, which in turn follows the published standard.

5. The ABI and header file bullshit has been debunked ad nauseam, eg on Groklaw. Again, these files simply encapsulate published standards. If Linux's signal.h infringes, then so did Windows NT's until Ballmer paided SCOX.

6. ELF is a standard. Presumably a case of 'form follows function follows public standard' again.

Is that enough yet?

Message ID: 153003
Posted By: nobbutl
Posted On: 2004-07-11 18:53:00
Subject: Re: RCU, ULS, IPC, ELF and Intensive Ca

3) The Unix style ULS routines were deprecated by engineer Hubertus Franke, who submitted a Linux 2.4.17 patch based on entirely new architecture (www.uwsg.iu.edu/hypermail/linux/kernel/0202.1/0745.html)

Brilliant! Good find. Hubertus is/was @watson.ibm.com, so that's SCOX's angle right there, this is presumably the only example of Hatch's "non-public contributions" that they've identified. (Or was it Heise. Phooey.)

BINGO! I think this is one of the contributions which eventually morphed into Futexes.
http://ds9a.nl/futex-manpages/futex2.html
(Which is why we've never heard of ULS.) There is a *HUGE* amount of Linux innovation and well documented independent development in Futexes, which both pre-dates and post-dates the contribution from Hubertus that the indefatigable Mr Stats_for_all cites.

Hubertus' other area seems to be GCOV, which is stupefyingly innocuous.

Message ID: 153006
Posted By: nobbutl
Posted On: 2004-07-11 19:30:00
Subject: Re: RCU, ULS, IPC, ELF and Intensive Ca
Recs: 2
Rating: 0
Raters: 0

<< 2) I have no information regarding the SYS V init allegation.>>

This was one of the *non-kernel* packages in which SCOX identified, with specificity, lines 1 to the end of every single file as being theirs, oh yessss precioussss.
sco.tuxrocks.com/Docs/IBM/IBM-157-28-G.pdf
(page 16 of 17)

Hope this helps. Presumably Mr Harrop ran out of paper before he got onto binutils, glibc, LiS, AIO, and COFF, all of which apparently belong to SCOX in whole or part.

The texts of these Yahoo Message Board posts have been licensed for copying and distribution by the Yahoo Message Board user "stats_for_all", "nobbutl" under the following license: License: CCL Attribution-NonCommercial-ShareAlike v2.0.