Maciej W. Rozycki wrote:
> On Mon, 9 Jun 2008, Glauber Costa wrote:
> 
>> Do it, instead of keeping in io_apic_32.c. This is the way x86_64
>> already does.
> 
>  This change looks wrong -- is native_smp_prepare_cpus() at all run on
> !CONFIG_SMP?  I don't think so.  You have to do this differently.
> 
>   Maciej

How about this one instead?

This one does enable_IO_APIC in the init_uniprocessor too, and should 
account for the !smp case.