Re: [PATCH stub] kvm: caching API for interrupts

[Avi Kivity <avi@redhat.com>]

On 07/29/2012 11:00 PM, Michael S. Tsirkin wrote:
> I've been looking at adding caching for IRQs so that we don't need to
> scan all VCPUs on each interrupt.  One issue I had a problem with, is
> how the cache structure can be used from both a thread (to fill out the
> cache) and interrupt (to actually send if cache is valid).
> 
> For now just added a lock field in the cache so we don't need to worry
> about this, and with such a lock in place we don't have to worry about
> RCU as cache can be invalidated simply under this lock.
> 
> For now this just declares the structure and updates the APIs
> so it's not intended to be applied, but just to give you
> the idea.
> 
> Comments, flames wellcome.

I hope you aren't expecting any of the latter.

> 
> +struct kvm_irq_cache {
> +	spinlock_t lock;
> +	bool valid;
> +	struct kvm_vcpu *dest;
> +	/* For now we only cache lapic irqs */
> +	struct kvm_lapic_irq irq;
> +	/* Protected by kvm->irq_cache_lock */
> +	struct list_head list;
> +};
> +

Why an external structure?

Why not add something to kvm_kernel_irq_routing_entry?  It is already
protected by rcu.

The atomic context code could look like this:


  if (kire->cache.valid) {
      kvm_apic_set_irq(kire->cache.vcpu, &kire->cache.irq);
      return;
  }
  queue_work(process_irq_slowpath);

The slow path tries to fills the cache, and processes the interrupt in
any case.

-- 
error compiling committee.c: too many arguments to function