>>> I can't even tell any 'outsider' about the VM problem because that
>>> would make them think NetBSD is completely useless for professional
>>> users.
>> Heh.  No worse than the vendor binary-only OS on my home machine,
>> which is a Mach derivative and - AFAICT - suffers from the same bug.
> SunOS 4 may well have the same bug.  I just added up the process
> sizes from `ps' on my machine and got 97224k, but `pstat -s' reports
> 137612k used.

That may not be meaningful.  I've long suspected that mmap()ed files
aren't counted in the ps-reported size on SunOS; if true, it could
certainly explain this.

I wouldn't suspect anything like the shadow VM object leak unless
long-running processes that fork slowly consume more and more swap
space which is released when they're finally killed.  Just a
discrepancy between ps sizes and pstat swap space used is not, in
itself, enough to imply the bug's presence....

					der Mouse

			    mouse@collatz.mcrcim.mcgill.edu