Hi Thomas,
On 22/07/2016 14:44, Thomas Gleixner wrote:
> On Thu, 21 Jul 2016, Auger Eric wrote:
>> On 20/07/2016 10:12, Thomas Gleixner wrote:
>>> On Tue, 19 Jul 2016, Eric Auger wrote:
>>>> +bool msi_doorbell_safe(void)
>>>> +{
>>>> + struct irqchip_doorbell *db;
>>>> + bool irq_remapping = true;
>>>> +
>>>> + mutex_lock(&irqchip_doorbell_mutex);
>>>> + list_for_each_entry(db, &irqchip_doorbell_list, next) {
>>>> + irq_remapping &= db->info.irq_remapping;
>>>
>>> db->info.irq_remapping is set in msi_doorbell_register(). So you can keep book
>>> about that there. No need to iterate here.
>> Yes makes sense to store the info at registration time. Currently this
>> function is not in any fast path but that's cleaner from a general
>> perspective. I will need to do such iteration at un-registration though.
>
> Two simple counter should be sufficient.
>
> nr_registered_bells;
> nr_remapped_bells;
Yes definitively smarter to use counters! mental viscosity.
Thanks
Eric
>
> ....
>
> Thanks,
>
> tglx
>
--
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
