[PATCH v2] vhost: Don't set vring call if guest notifier is disabled

[PATCH v2] vhost: Don't set vring call if guest notifier is disabled

[oenhan]

From: Huaitong Han <hanht2@chinatelecom.cn>

The vring call fd is set even when the guest does not use MSIX (e.g., virtio
PMD). This results in unnecessary CPU overhead for handling virtio interrupts.
The previous patch only optimized the condition when query_queue_notifier was
enabled and the vector was unset. However, if query_queue_notifier is disabled,
the vring call FD should also be unset to avoid this inefficiency.

Fixes: 96a3d98d2c ("vhost: don't set vring call if no vector")

Reported-by: Zhiyuan Yuan <yuanzhiyuan@chinatelecom.cn>
Signed-off-by: Huaitong Han <hanht2@chinatelecom.cn>
---
v2: - add Fixes tag
    - cc qemu-stable@nongnu.org

 hw/virtio/vhost.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
index 6aa72fd434..d17e7cc6fe 100644
--- a/hw/virtio/vhost.c
+++ b/hw/virtio/vhost.c
@@ -1342,8 +1342,8 @@ int vhost_virtqueue_start(struct vhost_dev *dev,
     }
 
     if (k->query_guest_notifiers &&
-        k->query_guest_notifiers(qbus->parent) &&
-        virtio_queue_vector(vdev, idx) == VIRTIO_NO_VECTOR) {
+        (!k->query_guest_notifiers(qbus->parent) ||
+            virtio_queue_vector(vdev, idx) == VIRTIO_NO_VECTOR)) {
         file.fd = -1;
         r = dev->vhost_ops->vhost_set_vring_call(dev, &file);
         if (r) {
-- 
2.43.5

Re: [PATCH v2] vhost: Don't set vring call if guest notifier is disabled

[Michael S. Tsirkin]

On Wed, Feb 19, 2025 at 09:52:35AM +0800, oenhan@gmail.com wrote:
> From: Huaitong Han <hanht2@chinatelecom.cn>
> 
> The vring call fd is set even when the guest does not use MSIX (e.g., virtio
> PMD). This results in unnecessary CPU overhead for handling virtio interrupts.
> The previous patch only optimized the condition when query_queue_notifier was
> enabled and the vector was unset. However, if query_queue_notifier is disabled,
> the vring call FD should also be unset to avoid this inefficiency.
> 
> Fixes: 96a3d98d2c ("vhost: don't set vring call if no vector")
> 
> Reported-by: Zhiyuan Yuan <yuanzhiyuan@chinatelecom.cn>
> Signed-off-by: Huaitong Han <hanht2@chinatelecom.cn>


Fails make check:


https://gitlab.com/mstredhat/qemu/-/jobs/9201935243

how was this tested? Pls include data on testing in the commit log.

> ---
> v2: - add Fixes tag
>     - cc qemu-stable@nongnu.org
> 
>  hw/virtio/vhost.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/hw/virtio/vhost.c b/hw/virtio/vhost.c
> index 6aa72fd434..d17e7cc6fe 100644
> --- a/hw/virtio/vhost.c
> +++ b/hw/virtio/vhost.c
> @@ -1342,8 +1342,8 @@ int vhost_virtqueue_start(struct vhost_dev *dev,
>      }
>  
>      if (k->query_guest_notifiers &&
> -        k->query_guest_notifiers(qbus->parent) &&
> -        virtio_queue_vector(vdev, idx) == VIRTIO_NO_VECTOR) {
> +        (!k->query_guest_notifiers(qbus->parent) ||
> +            virtio_queue_vector(vdev, idx) == VIRTIO_NO_VECTOR)) {
>          file.fd = -1;
>          r = dev->vhost_ops->vhost_set_vring_call(dev, &file);
>          if (r) {
> -- 
> 2.43.5