[ ... an alternative way to run profiling ... ]

> What I'm used to is having the clock interrupt perform a backtrace in
> the interrupt routine, and fill in the call-graph information at clock
> interrupt time. In return, you do *not* need a profiled kernel. No
> mcount function. No performance hit on every function call. No separate
> kernels for profiling and regular use.

This sounds really interesting.  Can NetBSD developers/users try to
see sample implementation, preferably for NetBSD/i386, to measure and
judge the implemenation?  Sould continue in tech-kern@netbsd.org

Tohru Nishimura