No, the problem is to guarantee that you will always succeed under any
reasonable load (it's obviously quite ok to have fork() fail with ENOMEM
when we really are very low on memory).
The probabilities was to show you that we should not have to swap out very
many thing for that to succeed - ie that it is entirely reasonable to use
a random approach. I also convinced myself with my numbers that something
else is needed on small-memory machines.
The thing is hereby closed as far as I'm concerned. We've discussed this
enough, this is not a committee, I want to see code. I have a solution
that I'm happy with in my tree, you'll see it in 109, and anybody else can
come up with other solutions, but I don't want to be Cc'd on this
discussion any more until somebody has anything more than just more words
to show for it.
I'd be thrilled to hear about results or better simulations, but there is
a point where we can say "enough", as the emails on this are starting to
hide more interesting stuff for me.
Linus
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html