Re: Building Big Ass Linux Machine, what are the limits?

Keith Owens (kaos@ocs.com.au)
Mon, 05 Oct 1998 00:58:57 +1000


On Sun, 4 Oct 1998 13:51:24 +0100,
Carlos Morgado <l39801@alfa.ist.utl.pt> wrote:
>On Sun, Oct 04, 1998 at 01:56:40AM +0100, Alan Cox wrote:
>> All the timing evidence I have from playing with one a bit is that they
>> are not really SMP in the normal sense
>>
>humm .. wouldn't building an SMP machine from non-SMP aware processors imply
>a lot of memory management logic for atomic test and set and such ?

Cross processor locks just need a synchronisation primitive that works
without hardware support, say Deker's algorithm. I have used Deker to
do inter-system locks on two mainframes running under different
operating systems. The only thing they had in common was a shared disk
and the only guaranteed ordering was that single track writes would not
be interleaved, but it worked.

-
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/