I've done an implementation of the following sysctl mib-variables:

kern.maxsyscall.{emulation_name}
kern.syscall.{emulation_name}.{syscall#}

as well as changed sysctl(8), and kdump(1) to take advantage of them.
It all works like a charm.

The only remaining static dependence is that on what emulations are
available in the current kernel and what mib-numbers they are mapped
to.  For the moment I'm doing this in sys/sysctl.h much like all other
sysctl mappings.

However, what I'd like to do is to have a central in-kernel table of
emulations with place-holders for LKM emulations (provided by exec
packages) so this info could be dynamically exported via sysctl as
well.  Are there any opinions on how this is done?  Is there interest
for this enhancement at all?  At least *I* like when kdump gets the
correct picture of what emulations and syscalls are available in a
kernel.  It can also be used for runtime decision of available kernel
features.

If I don't get any comments I'll cook this up myself, and file a
change-request for it.

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