> Uhm, why are you "reverse" engineering the traps? The Darwin source is
> available on Apple's web site (well their CVS server). So you have
> documentation as to what the traps do. :-)

Darwin sources are indeed usefull, because we can grab message packets
definitions from MiG files, and what mach traps are supposed to do from
the documentation.

But when you have to actually implement things, you end up with a lot of
gaps between the documentation and the reality. It's much easier (at
least for me) to dump Mach messages with gdb than to understand what
value will be stored in a given field from Darwin sources. 

For instance, when escahnging messages between the task and a Mach
server, reply packets id are equal to the request packet id plus 100.
This is not in the documentation (or I did not find the appropriate
document), and it's much easier to observe it in gdb on Darwin than to
dig it from the sources.

-- 
Emmanuel Dreyfus.  
Hiroshima 45. Tchernobyl 86. Windows 95. 
manu@netbsd.org