It appears that [kern/2585: time accumulated in subprocesses often is
bad] fixes my pr [port-pmax/2573:  CPU resource utilisation not
accurate].  At least I'd made it to looking at kern_resource.c :-)

I think it's still interesting to note that I have _not_ seen this
problem on the i386 port, but it's extremely easy to reproduce on the
pmax (and [kern/2585] originated on a pc532).

BTW: Is this the right forum to discuss pr fixes?

Simon.