On Mon, Jul 13, 2009 at 09:26:21AM -0400, Gregory Haskins wrote:
> Gleb Natapov wrote:
> > On Mon, Jul 13, 2009 at 04:02:56PM +0300, Michael S. Tsirkin wrote:
> >
> >> On Sun, Jul 12, 2009 at 03:03:53PM +0300, Gleb Natapov wrote:
> >>
> >>> Use RCU locking for mask/ack notifiers lists.
> >>>
> >>> Signed-off-by: Gleb Natapov <gleb@xxxxxxxxxx>
> >>> ---
> >>> virt/kvm/irq_comm.c | 20 +++++++++++---------
> >>> 1 files changed, 11 insertions(+), 9 deletions(-)
> >>>
> >>> diff --git a/virt/kvm/irq_comm.c b/virt/kvm/irq_comm.c
> >>> index 5dde1ef..ba3a115 100644
> >>> --- a/virt/kvm/irq_comm.c
> >>> +++ b/virt/kvm/irq_comm.c
> >>> @@ -179,18 +179,18 @@ void kvm_notify_acked_irq(struct kvm *kvm, unsigned irqchip, unsigned pin)
> >>> break;
> >>> }
> >>> }
> >>> - rcu_read_unlock();
> >>>
> >>> - hlist_for_each_entry(kian, n, &kvm->irq_ack_notifier_list, link)
> >>> + hlist_for_each_entry_rcu(kian, n, &kvm->irq_ack_notifier_list, link)
> >>> if (kian->gsi == gsi)
> >>> kian->irq_acked(kian);
> >>> + rcu_read_unlock();
> >>> }
> >>>
> >>> void kvm_register_irq_ack_notifier(struct kvm *kvm,
> >>> struct kvm_irq_ack_notifier *kian)
> >>> {
> >>> mutex_lock(&kvm->irq_lock);
> >>> - hlist_add_head(&kian->link, &kvm->irq_ack_notifier_list);
> >>> + hlist_add_head_rcu(&kian->link, &kvm->irq_ack_notifier_list);
> >>> mutex_unlock(&kvm->irq_lock);
> >>> }
> >>>
> >>> @@ -198,8 +198,9 @@ void kvm_unregister_irq_ack_notifier(struct kvm *kvm,
> >>> struct kvm_irq_ack_notifier *kian)
> >>> {
> >>> mutex_lock(&kvm->irq_lock);
> >>> - hlist_del_init(&kian->link);
> >>> + hlist_del_init_rcu(&kian->link);
> >>> mutex_unlock(&kvm->irq_lock);
> >>> + synchronize_rcu();
> >>> }
> >>>
> >> This is done under kvm->lock still, which means the lock might be held
> >> potentially for a very long time. Can synchronize_rcu be moved out of
> >> this lock?
> >>
> >>
> > Only if kvm_free_assigned_device() will be moved out of this lock.
> > Device de-assignment is not very frequent event though. How long do you
> > think it may be held? KVM RCU read sections are very brief.
> >
>
> Note that the delay imposed by the barrier is not only related to the
> length of the critical section. The barrier blocks until the next grace
> period, and depending on the type of RCU you are using and your config
> options, this could be multiple milliseconds.
>
> I am not saying that this is definitely a problem for your design. I
> am just pointing out that the length of the KVM-RCU read section is only
Yeah I understand that other RCU read section may introduce delays too.
The question is how big the delay may be. I don't think multiple
milliseconds delay in device de-assignment is a big issue though.
> one of several factors that influence the ultimate duration of your
> kvm->lock CS. IIUC, in an ideally designed subsystem, the read-side CS
> will be dwarfed by the length of the grace, so the grace (and barriers
> against it) are really the critical factor of this type of design.
>
> I am CC'ing Paul in case I am saying something dumb/untrue. ;)
>
> Kind Regards,
> -Greg
>
> >
> >>> int kvm_request_irq_source_id(struct kvm *kvm)
> >>> @@ -246,7 +247,7 @@ void kvm_register_irq_mask_notifier(struct kvm *kvm, int irq,
> >>> {
> >>> mutex_lock(&kvm->irq_lock);
> >>> kimn->irq = irq;
> >>> - hlist_add_head(&kimn->link, &kvm->mask_notifier_list);
> >>> + hlist_add_head_rcu(&kimn->link, &kvm->mask_notifier_list);
> >>> mutex_unlock(&kvm->irq_lock);
> >>> }
> >>>
> >>> @@ -254,8 +255,9 @@ void kvm_unregister_irq_mask_notifier(struct kvm *kvm, int irq,
> >>> struct kvm_irq_mask_notifier *kimn)
> >>> {
> >>> mutex_lock(&kvm->irq_lock);
> >>> - hlist_del(&kimn->link);
> >>> + hlist_del_rcu(&kimn->link);
> >>> mutex_unlock(&kvm->irq_lock);
> >>> + synchronize_rcu();
> >>> }
> >>>
> >>> void kvm_fire_mask_notifiers(struct kvm *kvm, int irq, bool mask)
> >>> @@ -263,11 +265,11 @@ void kvm_fire_mask_notifiers(struct kvm *kvm, int irq, bool mask)
> >>> struct kvm_irq_mask_notifier *kimn;
> >>> struct hlist_node *n;
> >>>
> >>> - WARN_ON(!mutex_is_locked(&kvm->irq_lock));
> >>> -
> >>> - hlist_for_each_entry(kimn, n, &kvm->mask_notifier_list, link)
> >>> + rcu_read_lock();
> >>> + hlist_for_each_entry_rcu(kimn, n, &kvm->mask_notifier_list, link)
> >>> if (kimn->irq == irq)
> >>> kimn->func(kimn, mask);
> >>> + rcu_read_unlock();
> >>> }
> >>>
> >>> void kvm_free_irq_routing(struct kvm *kvm)
> >>> --
> >>> 1.6.2.1
> >>>
> >>> --
> >>> 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
> >>>
> >
> > --
> > Gleb.
> > --
> > 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
> >
>
>
--
Gleb.
--
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
