Re: [PATCH v3 7/8] Move IO APIC to its own lock.

[Gleb Natapov <gleb@redhat.com>]

On Thu, Aug 13, 2009 at 12:49:45PM +0300, Avi Kivity wrote:
> On 08/13/2009 12:48 PM, Gleb Natapov wrote:
> >On Thu, Aug 13, 2009 at 06:13:30AM -0300, Marcelo Tosatti wrote:
> >>>+++ b/virt/kvm/ioapic.c
> >>>@@ -182,6 +182,7 @@ int kvm_ioapic_set_irq(struct kvm_ioapic *ioapic, int irq, int level)
> >>>  	union kvm_ioapic_redirect_entry entry;
> >>>  	int ret = 1;
> >>>
> >>>+	mutex_lock(&ioapic->lock);
> >>>  	if (irq>= 0&&  irq<  IOAPIC_NUM_PINS) {
> >>>  		entry = ioapic->redirtbl[irq];
> >>>  		level ^= entry.fields.polarity;
> >>But this is an RCU critical section now, right?
> >>
> >Correct! Forget about that. It was spinlock, but Avi prefers mutexes.
> 
> Well, I prefer correct code to mutexes.
> 
> >>If so, you can't sleep, must use a spinlock.
> >Either that or I can collect callbacks in critical section and call them
> >afterwords.
> 
> There's also srcu.
> 
What are the disadvantages? There should be some, otherwise why not use
it all the time.

--
			Gleb.