>>>>> "Ronald" == Ronald Khoo <ronald@cpm.COM.MY> writes:

>> Either "kdump" source should #define all of the options before
>> including sys/kern/syscalls.c or sys/kern/syscalls.c #ifdefs should
>> be changed to pick up on something defined by kdump source.

Ronald> kdump should be able to decode everything, regardless of
Ronald> whether the system its running on supports them, as long as
Ronald> the version of the OS is compatible, no ?  (so that I might
Ronald> have a chance of dumping someone else's ktrace.out, perhaps?)

Ronald> If you go along with that, then you'd probably make kdump
Ronald> define everything by itself, regardless of whether config put
Ronald> it into your kernel's Makefile or not.

I thought of that early on, but that is an impossible goal,
considering LKMs.  OK, One might use "der Mouse" tactics to let ktrace
files be self-contained, which is easy to do with my scheme, just let
ktrace instead of kdump extract the currently used emulation/syscalls
in a kernel and prepend it as a header to ktrace.out.  Perhaps that is
a hybrid solution all can accept?

Niklas

Niklas Hallqvist	Phone: +46-(0)31-40 75 00
Applitron Datasystem	Fax:   +46-(0)31-83 39 50
Molndalsvagen 95	Email: niklas@appli.se
S-412 63  GOTEBORG	WWW:   Here
Sweden