zero runtime when negative

[Adam Hoka <adam.hoka%gmail.com@localhost>]

Hi!

Can you guys please check if the following makes sense:

http://www.netbsd.org/~ahoka/patches/timebackwards.diff 

The use case is the following:

After starting NetBSD in vmware I see a lot of RLIMIT_CPU reached
messages overflowing my dmesg. At some time during the boot one or more
processes calculate a negative runtime, but as nothing resets them to
zero they keep reaching the cpu limit and the printouts are spamming the
console. Also the printout is bogus, because they are not really killed
in this case.

Any comments?


Cheers,
Adam