Re: [PATCH v3 14/18] xen/arm: IRQ: Store IRQ type in arch_irq_desc

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

On 04/16/2014 04:45 PM, Ian Campbell wrote:
> On Tue, 2014-04-08 at 15:44 +0100, Julien Grall wrote:
> [...]
>> @@ -282,7 +286,8 @@ int setup_dt_irq(const struct dt_irq *irq, struct irqaction *new)
>>           * TODO: Handle case where SPI is setup on different CPU than
>>           * the targeted CPU and the priority.
>>           */
>> -        gic_route_irq_to_xen(desc, level, cpumask_of(smp_processor_id()),
>> +        desc->arch.type = irq->type;
>> +        gic_route_irq_to_xen(desc, cpumask_of(smp_processor_id()),
>>                               GIC_PRI_IRQ);
>>          desc->handler->startup(desc);
>>      }
> [...]
>> @@ -341,10 +345,9 @@ int route_dt_irq_to_guest(struct domain *d, const struct dt_irq *irq,
>>          goto out;
>>      }
>>  
>> -    level = dt_irq_is_level_triggered(irq);
>> -    gic_route_irq_to_guest(d, desc, level, cpumask_of(smp_processor_id()),
>> +    desc->arch.type = irq->type;
>> +    gic_route_irq_to_guest(d, desc, cpumask_of(smp_processor_id()),
>>                             GIC_PRI_IRQ);
> 
> 
> When I asked why these two assignments weren't using irq_set_type you
> said you were going to add an assert.

The arch.type in setup_dt_irq will be removed in next patch. It's only
here for bisection.

For the second one, I was planning to add an ASSERT in irq_set_type not
here. But, I forgot to take into account your comment from V3 on this
patch :/.

Here it's fine because the function will bail out if the IRQ desc is
already setup it (see patch #12).

Regards,

-- 
Julien Grall