last kicked event index must be updated unconditionally:
even if we don't need to kick, we do not want to re-check
the same entry for events.
Signed-off-by: Michael S. Tsirkin <mst at redhat.com>
---
tools/virtio/ringtest/ring.c | 24 +++++++++++++++---------
1 file changed, 15 insertions(+), 9 deletions(-)
diff --git a/tools/virtio/ringtest/ring.c b/tools/virtio/ringtest/ring.c
index 6b70c4e..d044a88 100644
--- a/tools/virtio/ringtest/ring.c
+++ b/tools/virtio/ringtest/ring.c
@@ -196,16 +196,18 @@ bool enable_call()
void kick_available(void)
{
+ bool need;
+
/* Flush in previous flags write */
/* Barrier C (for pairing) */
smp_mb();
- if (!need_event(event->kick_index,
- guest.avail_idx,
- guest.kicked_avail_idx))
- return;
+ need = need_event(event->kick_index,
+ guest.avail_idx,
+ guest.kicked_avail_idx);
guest.kicked_avail_idx = guest.avail_idx;
- kick();
+ if (need)
+ kick();
}
/* host side */
@@ -265,14 +267,18 @@ bool use_buf(unsigned *lenp, void **bufp)
void call_used(void)
{
+ bool need;
+
/* Flush in previous flags write */
/* Barrier D (for pairing) */
smp_mb();
- if (!need_event(event->call_index,
+
+ need = !need_event(event->call_index,
host.used_idx,
- host.called_used_idx))
- return;
+ host.called_used_idx);
host.called_used_idx = host.used_idx;
- call();
+
+ if (need)
+ call();
}
--
MST
On Thu, 26 Oct 2017 04:48:01 +0300 "Michael S. Tsirkin" <mst at redhat.com> wrote:> last kicked event index must be updated unconditionally: > even if we don't need to kick, we do not want to re-check > the same entry for events. > > Signed-off-by: Michael S. Tsirkin <mst at redhat.com> > --- > tools/virtio/ringtest/ring.c | 24 +++++++++++++++--------- > 1 file changed, 15 insertions(+), 9 deletions(-)Acked-by: Cornelia Huck <cohuck at redhat.com> I think virtio_ring_0_9 has the same issue?
On 2017?10?26? 09:48, Michael S. Tsirkin wrote:> last kicked event index must be updated unconditionally: > even if we don't need to kick, we do not want to re-check > the same entry for events. > > Signed-off-by: Michael S. Tsirkin <mst at redhat.com> > --- > tools/virtio/ringtest/ring.c | 24 +++++++++++++++--------- > 1 file changed, 15 insertions(+), 9 deletions(-) > > diff --git a/tools/virtio/ringtest/ring.c b/tools/virtio/ringtest/ring.c > index 6b70c4e..d044a88 100644 > --- a/tools/virtio/ringtest/ring.c > +++ b/tools/virtio/ringtest/ring.c > @@ -196,16 +196,18 @@ bool enable_call() > > void kick_available(void) > { > + bool need; > + > /* Flush in previous flags write */ > /* Barrier C (for pairing) */ > smp_mb(); > - if (!need_event(event->kick_index, > - guest.avail_idx, > - guest.kicked_avail_idx)) > - return; > + need = need_event(event->kick_index, > + guest.avail_idx, > + guest.kicked_avail_idx); > > guest.kicked_avail_idx = guest.avail_idx; > - kick(); > + if (need) > + kick(); > } > > /* host side */ > @@ -265,14 +267,18 @@ bool use_buf(unsigned *lenp, void **bufp) > > void call_used(void) > { > + bool need; > + > /* Flush in previous flags write */ > /* Barrier D (for pairing) */ > smp_mb(); > - if (!need_event(event->call_index, > + > + need = !need_event(event->call_index, > host.used_idx, > - host.called_used_idx)) > - return; > + host.called_used_idx); > > host.called_used_idx = host.used_idx; > - call(); > + > + if (need) > + call(); > }Acked-by: Jason Wang <jasowang at redhat.com>
On Thu, Oct 26, 2017 at 10:36:47AM +0200, Cornelia Huck wrote:> On Thu, 26 Oct 2017 04:48:01 +0300 > "Michael S. Tsirkin" <mst at redhat.com> wrote: > > > last kicked event index must be updated unconditionally: > > even if we don't need to kick, we do not want to re-check > > the same entry for events. > > > > Signed-off-by: Michael S. Tsirkin <mst at redhat.com> > > --- > > tools/virtio/ringtest/ring.c | 24 +++++++++++++++--------- > > 1 file changed, 15 insertions(+), 9 deletions(-) > > Acked-by: Cornelia Huck <cohuck at redhat.com>Actually it seems to sometimes trigger stalls when host and guest are on the same physical core. To trigger: ./ring --sleep --host-affinity 0 --guest-affinity 0 This worries me since it's similar to what vhost does. Still debugging - anyone see anything suspicious?> I think virtio_ring_0_9 has the same issue?I think it does, posted a patch. -- MST