Paul Brook wrote:
> On Tuesday 23 September 2008, Jes Sorensen wrote:
>> So what you're saying is that we need to distinguish between number of
>> possible CPUs and how many we allow for a given type of emulation? It
>> seems to me that using MAX_CPUS to be the maximum possible for an
>> architecture is fair and we should then maybe put it into the machine
>> description to set the upper limit for actual runtime ones?
> 
> Something like that, yes.
> 
> Currently there's no real hard limit on the number of CPUs. I'd be nice to 
> keep it that way if possible.

Hi,

Here's an attempt of implementing a per-machine max_cpus variable.
This means developers can limit smp support to match the actual board
sizes. It also makes the static MAX_CPUS check in vl.c obsolete which
I think is a good thing :-)

I think I managed to get the numbers right for all the boards, but my
apologies if I got any of them wrong. It should be easy to fix, as it
will spit out a message explaining why it failed.

Let me know what you think of this approach.

Cheers,
Jes