Re: [PATCH v8 09/13] xen/arm: second irq injection while the first irq is still inflight

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

On 23/05/14 18:24, Stefano Stabellini wrote:
> On Fri, 23 May 2014, Julien Grall wrote:
>> On 05/23/2014 03:50 PM, Stefano Stabellini wrote:
>>> The following change works:
>>>
>>> diff --git a/xen/arch/arm/domain.c b/xen/arch/arm/domain.c
>>> index 33141e3..2a8456f 100644
>>> --- a/xen/arch/arm/domain.c
>>> +++ b/xen/arch/arm/domain.c
>>> @@ -644,6 +644,8 @@ int arch_set_info_guest(
>>>       else
>>>           set_bit(_VPF_down, &v->pause_flags);
>>>
>>> +    vgic_vcpu_inject_irq(v, v->domain->arch.evtchn_irq);
>>> +
>>
>> This is racy, we may not clear the _VPF_down bit in this function
>> (depending if VGCF_online is set or not).
>>
>> Hopefully for ARM, libxc is setting this flags by default but it's not
>> always true.
>
> I could change the code to call vgic_vcpu_inject_irq only if VGCF_online
> is set, but on second thought, would the code actually be more readable?
> Or less error prone?
>
> I think that the original patch is better. At least the hack is present
> in a single very obvious place (vgic_enable_irqs).

Hmmm ... right. I know that this code will likely change (with GICv3 
support). Can you add a comment in the code explain this issue?

With this change:

Acked-by: Julien Grall <julien.grall@linaro.org>

Regards,

-- 
Julien Grall