Re: [PATCHv3 RFC 1/2] kvm: implement kvm_set_msi_inatomic

[Gleb Natapov <gleb@redhat.com>]

On Wed, Jun 13, 2012 at 07:12:26PM +0300, Michael S. Tsirkin wrote:
> > > > > @@ -134,10 +141,30 @@ int kvm_set_msi(struct kvm_kernel_irq_routing_entry *e,
> > > > >  	irq.level = 1;
> > > > >  	irq.shorthand = 0;
> > > > >  
> > > > > +	/* Multicast MSI doesn't really block but might take a long time. */
> > > > > +	if (unlikely(noblock && kvm_msi_is_multicast(irq.dest_id,
> > > > > +						     irq.delivery_mode)))
> > > > delivery_mode? Should be dest_mode.
> 
> Yes. Good catch, thanks.
> 
> > But you probably need to check that
> > > > delivery_mode is not ExtINT either.
> 
> It does not look like anything happens with ExtInt
> if you try to trigger it from MSI.
> 
Currently no, but it should appear as if interrupt comes from PIC.
I wouldn't allow anything but fixed mode here just to be on a safe side.
Lowest prio will have to loop even after introducing irq cache.

--
			Gleb.