> from within the kernel, would it be possible to cause an arbitrary
> function in an arbitrary process execute asynchronously with
> arbitrary arguments?

Yes, under certain reasonable assumptions (such as, the process hasn't
trashed its stack pointer, or you know where the process has some
memory that can be used as stack for the duration).

> (and then restore the process to the proper state after this is done)

> like a signal handler with a twist...

Exactly.  You would just need to generalize (a duplicate of?) the
signal delivery mechanisms.

> assuming that this is possible, would there be a way to do it with MI
> code?

No, not even if you count using existing MD signal delivery code as MI.
(At least I can't see any.  If anyone else can, I'd love to hear what
I've missed.)

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B