Re: [PATCH v3 3/6] xen/arm: gic: Use the correct CPU ID

[Julien Grall <julien.grall@linaro.org>]

On 09/20/2013 04:44 PM, Ian Campbell wrote:
> On Fri, 2013-09-20 at 16:03 +0100, Julien Grall wrote:
> 
>> +static unsigned int gic_cpu_mask(const cpumask_t *cpumask)
>> +{
>> +    unsigned int cpu;
>> +    unsigned int mask = 0;
>> +    cpumask_t possible_mask;
>> +
>> +    cpumask_and(&possible_mask, cpumask, &cpu_possible_map);
> 
> Based on the other subconversation doesn't this need to be online_mask?
> Or is the cpu area setup for cpus which are possible but not online?

cpu area is initialized a little bit before the cpu is bring up:
  * initialize per cpu data (via the notifier_call_chain CPU_UP_PREPARE)
  * signal the cpu to boot (__cpu_up)
  ...
  * call start_secondary
      * initialize the gic cpu interface (gic_cpu_init)
      * route ppis (which used gic_cpu_mask)
      * set the cpu online

If we use the cpu online mask, Xen won't be able to route the different
ppis to the processor.

> Should the check (whichever it is) be an assertion?

If we stay with the cpu_possible_map, I think we are fine without an
assertion.

>> +    for_each_cpu(cpu, &possible_mask)
>> +    {
>> +        ASSERT(cpu < NR_GIC_CPU_IF);
>> +        ASSERT(__per_cpu_offset[cpu] != -(long)__per_cpu_start);
>> +        mask |= per_cpu(gic_cpu_id, cpu);
>> +    }
>> +
>> +    return mask;
>> +}
> 
> 


-- 
Julien Grall