> Each time you write code that runs SMP threaded you must insert the proper
> barriers and you must consider each case separately. That's why writing
> SMP threaded code is harder ;).
It is harder since you want to use everything that exists. You must base
SMP on a very fiew mechanisms you know about being implementable on
everything. For synchronisation between CPUs, you can build everything
with SMP semaphores (spinlocks), for example, provided that you base the
kernel code on a _right_ set of kernel primitives.
Remember some recent mail from me about the way to doing things obviously
correct for SMP and not significantly more complex than UP. ;-)
Gérard.
-
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/