Re: [tip:irq/numa] x86: read apic ID in the !acpi_lapic case

[Yinghai Lu <yinghai@kernel.org>]

Ingo Molnar wrote:
> * Yinghai Lu <yinghai@kernel.org> wrote:
> 
>> -	return hard_smp_processor_id() >> index_msb;
>> +	return initial_apic_id >> index_msb;
> 
>> -	unsigned apicid = hard_smp_processor_id();
>> +	unsigned apicid = cpu_has_apic ? hard_smp_processor_id() : c->apicid;
> 
>> -	int apicid = hard_smp_processor_id();
>> +	int apicid = cpu_has_apic ? hard_smp_processor_id() : c->apicid;
> 
> Doesnt such a pattern call for the pushing of that distinction into 
> hard_smp_processor_id() instead? (after a careful review of all 
> hard_smp_processor_id() call-sites)

don't want to burden system that have APIC to check that bit.

YH