Re: [PATCH-4.5 3/4] xen/arm: do not request maintenance_interrupts

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

On 02/10/2014 05:03 PM, Stefano Stabellini wrote:
>  
>>> +            inflight = 0;
>>> +            GICH[GICH_LR + i] = 0;
>>> +            clear_bit(i, &this_cpu(lr_mask));
>>> +
>>> +            spin_lock(&gic.lock);
>>> +            p = irq_to_pending(v, irq);
>>> +            if ( p->desc != NULL )
>>> +                p->desc->status &= ~IRQ_INPROGRESS;
>>> +            clear_bit(GIC_IRQ_GUEST_VISIBLE, &p->status);
>>> +            if ( test_bit(GIC_IRQ_GUEST_PENDING, &p->status) &&
>>> +                    test_bit(GIC_IRQ_GUEST_ENABLED, &p->status))
>>> +            {
>>
>> I would add a WARN_ON(p->desc != NULL) here. AFAIK, this code path shouldn't
>> be used for physical IRQ.
> 
> That's not true: an edge physical irq can come through while another one
> of the same type is being handled. In fact pending and active bits exist
> even on the physical GIC interface.
> 

It won't be fired until the previous one is EOIed. The physical GIC
interface will keep it internally.

But ... after thinking the WARN is stupid here because we can have this
following case:

IRQ A fired
    -> inject IRQ A
IRQ A eoied

IRQ A fired
   -> set pending bits

clear lrs
   -> re-inject IRQ A

-- 
Julien Grall