This is the answer I got from Larry McVoy about some of his smp ideas.
I wanted to throw these ideas on the table for the list to toss
around. Larry agreed. The original idea was HOW (not whether, so lets
no point this in that direction) to implement a multi-kernel os (using the
current linux base).
On Mon, 14 Feb 2000, Larry McVoy replied:
> : Dear Mr. McVoy,
>
> Call me Larry - I'm not very formal - but thanks for being polite.
>
> : I have been reading your input to the recent smp/scheduling threads on the
> : l-k mailing list. I have also read your papers. I am convinced that your
> : concept is sound (and neat).
>
> Cool, that's one down, 3.999999 billion to go :-)
>
> : I am really asking how one would implement something like this?
>
> Well, I've been thinking about that quite a bit lately.
>
> : Let me recap so you can correct any misconceptions on my part. You seem to
> : be explaining a concept that entails small kernels running on specific
> : sets of proceccesors on smp boxes. These kernels are then light enough to
> : still run effectivly on smaller machines (i.e. non-smp machines.)
>
> Right, exactly, and well put too.
>
> : another, and make each kernel hardware-equal to the others. The first
> : kernel would have to decide how many brothers to spawn based on the
> : hardware it finds and the launch them while releaseing resources for
> : them. This is a neat idea, but shouldn't a protocol be discussed to ensure
> : that the proper materials can be passed back and forth between kernels? I
> : can't even begin to know where to start with that one.
>
> That's the right idea.
>
I am thinking that there may be another protocol that would fit the need
and could be implemented vitually (tcp/udp) to fill the communication
need between the individual kernels. But maybe this is the cart before
the horse sort of thing.
Where do we have to start to get this up and running on an intel
platform. That is probably most common (although I would bet its the least
common smp implementation.
The first challenge I see is to get the "intial" kernel to do the system
analysis and load its sibling.
Thoughts?
Matt Berglund
-
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.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Feb 15 2000 - 21:00:27 EST