|| ||Eric.M.Kidd@Dartmouth.EDU (Eric M. Kidd)|
|| ||An open appeal to SCO|
|| ||20 Jun 2003 15:44:36 EDT|
I'm making it easy--15 minutes and 150MB of RAM for them to find code shared
between Linux and Unix. This program uses the rolling hash technique
proposed by Egan at the Inquirer.
I'm trying to make it as easy as possible for SCO (and other copyright
holders) to report wrongdoing to free software maintainers without revealing
any more than necessary about their own code.
Comments (2 posted)
|| ||Searching for Linux code in SCO kernel (or vice versa)|
|| ||Tue, 24 Jun 2003 15:39:37 +0400 (MSD)|
SCO claims that some Linux code is taken from Unix kernel. SCO
also claims that no Linux code ever went into its Unix kernel.
Given SCO's kernel source, that would be easy to check; however,
SCO will not give us the source. So how could we search for similar
code _without_ the sources?
1. Take SCO Unix.
2. Take Linux kernel source.
3. Guess which compiler flags were used by SCO when compiling
4. Compile the suspicious portions from Linux kernel source with
SCO's compiler, making as few modifications as possible.
5. In the generated code, mark global addresses (subject to
relocation), magic constants, and probably some other constants
(struct member displacements?) as irrelevant.
6. Search the SCO Unix binary kernel for chunks matching relevant
portions of compiled Linux code.
Of course, we would be extremely lucky if it worked -- the code
must really be taken as is for such a test to work. But I think
this might be worth a try. (I don't have access to SCO Unix, so
I can't do this myself.)
IANAL, of course, so I don't know whether such procedures would
be legal, in US or elsewhere.
Comments (1 posted)
Page editor: Jonathan Corbet