On 07/06/2015 05:57 PM, Paolo Bonzini wrote:
>
>
> On 06/07/2015 17:35, Eric Auger wrote:
>>>> diff --git a/kernel/irq/bypass.c b/kernel/irq/bypass.c
>>>>>> index 5d0f92b..efadbe5 100644
>>>>>> --- a/kernel/irq/bypass.c
>>>>>> +++ b/kernel/irq/bypass.c
>>>>>> @@ -19,6 +19,42 @@ static LIST_HEAD(producers);
>>>>>> static LIST_HEAD(consumers);
>>>>>> static DEFINE_MUTEX(lock);
>>>>>>
>>>>>> +/* lock must be hold when calling connect */
>>>>
>>>> If a lock must be held while callbacks are called, you have to document
>>>> that producers and consumers must _not_ call back into the IRQ bypass
>>>> manager. (If they have to, you have to document explicitly "This
>>>> function can be called from producer and consumer callbacks" whenever
>>>> relevant).
>> OK Thanks
>
> Also, please document on functions that take the irq bypass mutex that
> they can sleep. In fact irq_bypass_{,un}register_{producer,consumer}
> need kerneldoc comments.
>
> The good thing is that this helps a bit forming a lock hierarchy across
> the subsystems, for example irq bypass mutex outside vfio_platform_irq
> spinlock, because you cannot have a spinlock inside the mutex. I think
> that all of your six callbacks are fine.
arghh, no that's wrong then. I have plenty of them in the KVM/arm vgic
part :-(
Eric
>
> Paolo
>
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
