On Fri, 31 Oct 1997, der Mouse wrote:

: I have a comparatively little amount of time, but I do know the SPARC
: instruction set fairly well.  (I have a multi-machine disassembler that
: I added a SPARC module to, in case you want it.  I also have a text
: file describing it.)

On Fri, 31 Oct 1997, Christos Zoulas wrote:

: Please let me know what is left to do... I can do the md portion if you'd
: like...

First, grab the file
ftp://ftp.duh.org/pub/NetBSD-hacks/linux_compat/linux_compat.tar.gz and
extract it per the README.txt in that directory.  That's as far as I have
gone thus far (splitting syscalls and errnos into their own architecture
dependent files).  After I go out of town this weekend, I'll be doing a
one-syscall-at-a-time "argument/struct verification" that will round down
all function arguments, structs, sysctls, ioctls, and fcntls that are not
consistent between Linux/x86 and the other architectures.  Once done, and
any MD support syscalls are written, that should complete the preparations
for dividing up Linux compat and split it nicely into MI and MD segments.

As to the MD part; I'm not _entirely_ sure (as I haven't gone thought i386
linux_machdep.c yet) how much stuff needs to be written yet, but it's likely
to be just low-level signal and context save/restore handlers, along with
the little bit of locore code that will be necessary to go from a syscall at
the low level to the syscall dispatching code.  I can bundle the Linux
machine dependent kernel sources into a tarball if desired....

Sigh; I never thought I'd be more condescending to an OS than I am to
Microsoft's, but the Linux kernel just seems to be sooooo disorganized.... 

=====
== Todd Vierling (Personal tv@pobox.com; Business tv@lucent.com)
== I know you like the Internet, Bobby.  Now go eat your Frosted Flakes.