[PATCH] virtio: add num_vf callback to virtio_bus

[PATCH] virtio: add num_vf callback to virtio_bus

[Yui Washizu]

Recent QEMU versions added support for virtio SR-IOV emulation,
allowing virtio devices to expose SR-IOV VFs to the guest.
However, virtio_bus does not implement the num_vf callback of bus_type,
causing dev_num_vf() to return 0 for virtio devices even when
SR-IOV VFs are active.

net/core/rtnetlink.c calls dev_num_vf(dev->dev.parent) to populate
IFLA_NUM_VF and IFLA_VFINFO_LIST in RTM_GETLINK responses.  For a
virtio-net device, dev.parent points to the virtio_device, whose bus
is virtio_bus.  Without num_vf, SR-IOV VF information is silently
omitted from tools that rely on rtnetlink, such as 'ip link show'.

Add a num_vf callback that delegates to dev_num_vf(dev->parent),
which in turn reaches the underlying transport (pci_bus_type for
virtio-pci) where the actual VF count is tracked.  Non-PCI transports
are unaffected as dev_num_vf() returns 0 when no num_vf callback is
present.

Signed-off-by: Yui Washizu <yui.washidu@gmail.com>
---
 drivers/virtio/virtio.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/virtio/virtio.c b/drivers/virtio/virtio.c
index 5bdc6b82b30b..299fa83be1d5 100644
--- a/drivers/virtio/virtio.c
+++ b/drivers/virtio/virtio.c
@@ -435,6 +435,14 @@ static void virtio_dev_shutdown(struct device *_d)
 	dev->config->reset(dev);
 }
 
+static int virtio_dev_num_vf(struct device *dev)
+{
+	struct virtio_device *vdev = dev_to_virtio(dev);
+
+	return dev_num_vf(vdev->dev.parent);
+}
+
+
 static const struct bus_type virtio_bus = {
 	.name  = "virtio",
 	.match = virtio_dev_match,
@@ -444,6 +452,7 @@ static const struct bus_type virtio_bus = {
 	.remove = virtio_dev_remove,
 	.irq_get_affinity = virtio_irq_get_affinity,
 	.shutdown = virtio_dev_shutdown,
+	.num_vf = virtio_dev_num_vf,
 };
 
 int __register_virtio_driver(struct virtio_driver *driver, struct module *owner)
-- 
2.47.3

Re: [PATCH] virtio: add num_vf callback to virtio_bus

[Jason Wang]

On Mon, Mar 9, 2026 at 10:37 AM Yui Washizu <yui.washidu@gmail.com> wrote:
>
> Recent QEMU versions added support for virtio SR-IOV emulation,
> allowing virtio devices to expose SR-IOV VFs to the guest.
> However, virtio_bus does not implement the num_vf callback of bus_type,
> causing dev_num_vf() to return 0 for virtio devices even when
> SR-IOV VFs are active.
>
> net/core/rtnetlink.c calls dev_num_vf(dev->dev.parent) to populate
> IFLA_NUM_VF and IFLA_VFINFO_LIST

Patch looks good but,

If I understand correctly, ndo_get_vf_config is needed for
IFLA_VFINFO_LIST but not implemeted in this patch?

> in RTM_GETLINK responses.  For a
> virtio-net device, dev.parent points to the virtio_device, whose bus
> is virtio_bus.  Without num_vf, SR-IOV VF information is silently
> omitted from tools that rely on rtnetlink, such as 'ip link show'.
>
> Add a num_vf callback that delegates to dev_num_vf(dev->parent),
> which in turn reaches the underlying transport (pci_bus_type for
> virtio-pci) where the actual VF count is tracked.  Non-PCI transports
> are unaffected as dev_num_vf() returns 0 when no num_vf callback is
> present.
>
> Signed-off-by: Yui Washizu <yui.washidu@gmail.com>
> ---
>  drivers/virtio/virtio.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
>
> diff --git a/drivers/virtio/virtio.c b/drivers/virtio/virtio.c
> index 5bdc6b82b30b..299fa83be1d5 100644
> --- a/drivers/virtio/virtio.c
> +++ b/drivers/virtio/virtio.c
> @@ -435,6 +435,14 @@ static void virtio_dev_shutdown(struct device *_d)
>         dev->config->reset(dev);
>  }
>
> +static int virtio_dev_num_vf(struct device *dev)
> +{
> +       struct virtio_device *vdev = dev_to_virtio(dev);
> +
> +       return dev_num_vf(vdev->dev.parent);
> +}
> +
> +
>  static const struct bus_type virtio_bus = {
>         .name  = "virtio",
>         .match = virtio_dev_match,
> @@ -444,6 +452,7 @@ static const struct bus_type virtio_bus = {
>         .remove = virtio_dev_remove,
>         .irq_get_affinity = virtio_irq_get_affinity,
>         .shutdown = virtio_dev_shutdown,
> +       .num_vf = virtio_dev_num_vf,
>  };
>
>  int __register_virtio_driver(struct virtio_driver *driver, struct module *owner)
> --
> 2.47.3
>

Thanks

Re: [PATCH] virtio: add num_vf callback to virtio_bus

[Yui Washizu]

On 3/9/2026 12:33 PM, Jason Wang wrote:
> On Mon, Mar 9, 2026 at 10:37 AM Yui Washizu <yui.washidu@gmail.com> wrote:
>> Recent QEMU versions added support for virtio SR-IOV emulation,
>> allowing virtio devices to expose SR-IOV VFs to the guest.
>> However, virtio_bus does not implement the num_vf callback of bus_type,
>> causing dev_num_vf() to return 0 for virtio devices even when
>> SR-IOV VFs are active.
>>
>> net/core/rtnetlink.c calls dev_num_vf(dev->dev.parent) to populate
>> IFLA_NUM_VF and IFLA_VFINFO_LIST
> Patch looks good but,
>
> If I understand correctly, ndo_get_vf_config is needed for
> IFLA_VFINFO_LIST but not implemeted in this patch?

Thank you for your comment.

I submitted this patch because I confirmed
that "ip link show" returns the correct VF IDs with this implementation.

In the current implementation,
I believe there is no information that would need to be referenced 
through IFLA_VFINFO_LIST.
Therefore, I did not include ndo_get_vf_config in this patch,
as it did not seem essential at this stage.

Would it be better to implement IFLA_VFINFO_LIST support together with this?

Regards,


>> in RTM_GETLINK responses.  For a
>> virtio-net device, dev.parent points to the virtio_device, whose bus
>> is virtio_bus.  Without num_vf, SR-IOV VF information is silently
>> omitted from tools that rely on rtnetlink, such as 'ip link show'.
>>
>> Add a num_vf callback that delegates to dev_num_vf(dev->parent),
>> which in turn reaches the underlying transport (pci_bus_type for
>> virtio-pci) where the actual VF count is tracked.  Non-PCI transports
>> are unaffected as dev_num_vf() returns 0 when no num_vf callback is
>> present.
>>
>> Signed-off-by: Yui Washizu <yui.washidu@gmail.com>
>> ---
>>   drivers/virtio/virtio.c | 9 +++++++++
>>   1 file changed, 9 insertions(+)
>>
>> diff --git a/drivers/virtio/virtio.c b/drivers/virtio/virtio.c
>> index 5bdc6b82b30b..299fa83be1d5 100644
>> --- a/drivers/virtio/virtio.c
>> +++ b/drivers/virtio/virtio.c
>> @@ -435,6 +435,14 @@ static void virtio_dev_shutdown(struct device *_d)
>>          dev->config->reset(dev);
>>   }
>>
>> +static int virtio_dev_num_vf(struct device *dev)
>> +{
>> +       struct virtio_device *vdev = dev_to_virtio(dev);
>> +
>> +       return dev_num_vf(vdev->dev.parent);
>> +}
>> +
>> +
>>   static const struct bus_type virtio_bus = {
>>          .name  = "virtio",
>>          .match = virtio_dev_match,
>> @@ -444,6 +452,7 @@ static const struct bus_type virtio_bus = {
>>          .remove = virtio_dev_remove,
>>          .irq_get_affinity = virtio_irq_get_affinity,
>>          .shutdown = virtio_dev_shutdown,
>> +       .num_vf = virtio_dev_num_vf,
>>   };
>>
>>   int __register_virtio_driver(struct virtio_driver *driver, struct module *owner)
>> --
>> 2.47.3
>>
> Thanks
>

Re: [PATCH] virtio: add num_vf callback to virtio_bus

[Jason Wang]

On Mon, Mar 9, 2026 at 2:19 PM Yui Washizu <yui.washidu@gmail.com> wrote:
>
>
> On 3/9/2026 12:33 PM, Jason Wang wrote:
> > On Mon, Mar 9, 2026 at 10:37 AM Yui Washizu <yui.washidu@gmail.com> wrote:
> >> Recent QEMU versions added support for virtio SR-IOV emulation,
> >> allowing virtio devices to expose SR-IOV VFs to the guest.
> >> However, virtio_bus does not implement the num_vf callback of bus_type,
> >> causing dev_num_vf() to return 0 for virtio devices even when
> >> SR-IOV VFs are active.
> >>
> >> net/core/rtnetlink.c calls dev_num_vf(dev->dev.parent) to populate
> >> IFLA_NUM_VF and IFLA_VFINFO_LIST
> > Patch looks good but,
> >
> > If I understand correctly, ndo_get_vf_config is needed for
> > IFLA_VFINFO_LIST but not implemeted in this patch?
>
> Thank you for your comment.
>
> I submitted this patch because I confirmed
> that "ip link show" returns the correct VF IDs with this implementation.
>
> In the current implementation,
> I believe there is no information that would need to be referenced
> through IFLA_VFINFO_LIST.
> Therefore, I did not include ndo_get_vf_config in this patch,
> as it did not seem essential at this stage.
>
> Would it be better to implement IFLA_VFINFO_LIST support together with this?

Then it's not, I think it's better to not mention IFLA_VFINFO_LIST in
the commit log as it may confuse people.

Thanks

>
> Regards,
>
>
> >> in RTM_GETLINK responses.  For a
> >> virtio-net device, dev.parent points to the virtio_device, whose bus
> >> is virtio_bus.  Without num_vf, SR-IOV VF information is silently
> >> omitted from tools that rely on rtnetlink, such as 'ip link show'.
> >>
> >> Add a num_vf callback that delegates to dev_num_vf(dev->parent),
> >> which in turn reaches the underlying transport (pci_bus_type for
> >> virtio-pci) where the actual VF count is tracked.  Non-PCI transports
> >> are unaffected as dev_num_vf() returns 0 when no num_vf callback is
> >> present.
> >>
> >> Signed-off-by: Yui Washizu <yui.washidu@gmail.com>
> >> ---
> >>   drivers/virtio/virtio.c | 9 +++++++++
> >>   1 file changed, 9 insertions(+)
> >>
> >> diff --git a/drivers/virtio/virtio.c b/drivers/virtio/virtio.c
> >> index 5bdc6b82b30b..299fa83be1d5 100644
> >> --- a/drivers/virtio/virtio.c
> >> +++ b/drivers/virtio/virtio.c
> >> @@ -435,6 +435,14 @@ static void virtio_dev_shutdown(struct device *_d)
> >>          dev->config->reset(dev);
> >>   }
> >>
> >> +static int virtio_dev_num_vf(struct device *dev)
> >> +{
> >> +       struct virtio_device *vdev = dev_to_virtio(dev);
> >> +
> >> +       return dev_num_vf(vdev->dev.parent);
> >> +}
> >> +
> >> +
> >>   static const struct bus_type virtio_bus = {
> >>          .name  = "virtio",
> >>          .match = virtio_dev_match,
> >> @@ -444,6 +452,7 @@ static const struct bus_type virtio_bus = {
> >>          .remove = virtio_dev_remove,
> >>          .irq_get_affinity = virtio_irq_get_affinity,
> >>          .shutdown = virtio_dev_shutdown,
> >> +       .num_vf = virtio_dev_num_vf,
> >>   };
> >>
> >>   int __register_virtio_driver(struct virtio_driver *driver, struct module *owner)
> >> --
> >> 2.47.3
> >>
> > Thanks
> >
>

Re: [PATCH] virtio: add num_vf callback to virtio_bus

[Washizu Yui]

2026年3月10日(火) 12:02 Jason Wang <jasowang@redhat.com>:
>
> On Mon, Mar 9, 2026 at 2:19 PM Yui Washizu <yui.washidu@gmail.com> wrote:
> >
> >
> > On 3/9/2026 12:33 PM, Jason Wang wrote:
> > > On Mon, Mar 9, 2026 at 10:37 AM Yui Washizu <yui.washidu@gmail.com> wrote:
> > >> Recent QEMU versions added support for virtio SR-IOV emulation,
> > >> allowing virtio devices to expose SR-IOV VFs to the guest.
> > >> However, virtio_bus does not implement the num_vf callback of bus_type,
> > >> causing dev_num_vf() to return 0 for virtio devices even when
> > >> SR-IOV VFs are active.
> > >>
> > >> net/core/rtnetlink.c calls dev_num_vf(dev->dev.parent) to populate
> > >> IFLA_NUM_VF and IFLA_VFINFO_LIST
> > > Patch looks good but,
> > >
> > > If I understand correctly, ndo_get_vf_config is needed for
> > > IFLA_VFINFO_LIST but not implemeted in this patch?
> >
> > Thank you for your comment.
> >
> > I submitted this patch because I confirmed
> > that "ip link show" returns the correct VF IDs with this implementation.
> >
> > In the current implementation,
> > I believe there is no information that would need to be referenced
> > through IFLA_VFINFO_LIST.
> > Therefore, I did not include ndo_get_vf_config in this patch,
> > as it did not seem essential at this stage.
> >
> > Would it be better to implement IFLA_VFINFO_LIST support together with this?
>
> Then it's not, I think it's better to not mention IFLA_VFINFO_LIST in
> the commit log as it may confuse people.
>
> Thanks
>

Thank you.
I'll update the commit message to remove the reference to IFLA_VFINFO_LIST
and send a v2.
Regards,

>
> >
> > Regards,
> >
> >
> > >> in RTM_GETLINK responses.  For a
> > >> virtio-net device, dev.parent points to the virtio_device, whose bus
> > >> is virtio_bus.  Without num_vf, SR-IOV VF information is silently
> > >> omitted from tools that rely on rtnetlink, such as 'ip link show'.
> > >>
> > >> Add a num_vf callback that delegates to dev_num_vf(dev->parent),
> > >> which in turn reaches the underlying transport (pci_bus_type for
> > >> virtio-pci) where the actual VF count is tracked.  Non-PCI transports
> > >> are unaffected as dev_num_vf() returns 0 when no num_vf callback is
> > >> present.
> > >>
> > >> Signed-off-by: Yui Washizu <yui.washidu@gmail.com>
> > >> ---
> > >>   drivers/virtio/virtio.c | 9 +++++++++
> > >>   1 file changed, 9 insertions(+)
> > >>
> > >> diff --git a/drivers/virtio/virtio.c b/drivers/virtio/virtio.c
> > >> index 5bdc6b82b30b..299fa83be1d5 100644
> > >> --- a/drivers/virtio/virtio.c
> > >> +++ b/drivers/virtio/virtio.c
> > >> @@ -435,6 +435,14 @@ static void virtio_dev_shutdown(struct device *_d)
> > >>          dev->config->reset(dev);
> > >>   }
> > >>
> > >> +static int virtio_dev_num_vf(struct device *dev)
> > >> +{
> > >> +       struct virtio_device *vdev = dev_to_virtio(dev);
> > >> +
> > >> +       return dev_num_vf(vdev->dev.parent);
> > >> +}
> > >> +
> > >> +
> > >>   static const struct bus_type virtio_bus = {
> > >>          .name  = "virtio",
> > >>          .match = virtio_dev_match,
> > >> @@ -444,6 +452,7 @@ static const struct bus_type virtio_bus = {
> > >>          .remove = virtio_dev_remove,
> > >>          .irq_get_affinity = virtio_irq_get_affinity,
> > >>          .shutdown = virtio_dev_shutdown,
> > >> +       .num_vf = virtio_dev_num_vf,
> > >>   };
> > >>
> > >>   int __register_virtio_driver(struct virtio_driver *driver, struct module *owner)
> > >> --
> > >> 2.47.3
> > >>
> > > Thanks
> > >
> >
>