In some email I received from matthew green, sie wrote:
> 
> there is a known bug in our VM system such that in certain conditions,
> swap pages are lost -- until the parent exits.  this affects long
> running programs like cron, and sendmail.  killing these processes
> often returns quite a lot more vm than ps would have you believe that
> they are using.  shells are another hot spot, i've noticed.
> 
> there are efforts underway to attempt to fix this...

for shells, does doing "exec tcsh" (for example), from tcsh produce the
same result as exiting the shell ?