* Re: [PATCH v3 2/2] virtio_net: notify MAC address change on device initialization
@ 2023-01-31 9:01 ` Paolo Abeni
0 siblings, 0 replies; 21+ messages in thread
From: Paolo Abeni @ 2023-01-31 9:01 UTC (permalink / raw)
To: Laurent Vivier, linux-kernel
Cc: David S. Miller, virtualization, Cindy Lu, Michael S. Tsirkin,
Eli Cohen, Gautam Dawar, Eugenio Pérez, Jason Wang, netdev,
Parav Pandit
On Fri, 2023-01-27 at 21:45 +0100, Laurent Vivier wrote:
> In virtnet_probe(), if the device doesn't provide a MAC address the
> driver assigns a random one.
> As we modify the MAC address we need to notify the device to allow it
> to update all the related information.
>
> The problem can be seen with vDPA and mlx5_vdpa driver as it doesn't
> assign a MAC address by default. The virtio_net device uses a random
> MAC address (we can see it with "ip link"), but we can't ping a net
> namespace from another one using the virtio-vdpa device because the
> new MAC address has not been provided to the hardware:
> RX packets are dropped since they don't go through the receive filters,
> TX packets go through unaffected.
>
> Signed-off-by: Laurent Vivier <lvivier@redhat.com>
> ---
> drivers/net/virtio_net.c | 20 ++++++++++++++++++++
> 1 file changed, 20 insertions(+)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 7d700f8e545a..704a05f1c279 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -3806,6 +3806,8 @@ static int virtnet_probe(struct virtio_device *vdev)
> eth_hw_addr_set(dev, addr);
> } else {
> eth_hw_addr_random(dev);
> + dev_info(&vdev->dev, "Assigned random MAC address %pM\n",
> + dev->dev_addr);
> }
>
> /* Set up our device-specific information */
> @@ -3933,6 +3935,24 @@ static int virtnet_probe(struct virtio_device *vdev)
>
> virtio_device_ready(vdev);
>
> + /* a random MAC address has been assigned, notify the device.
> + * We don't fail probe if VIRTIO_NET_F_CTRL_MAC_ADDR is not there
> + * because many devices work fine without getting MAC explicitly
> + */
> + if (!virtio_has_feature(vdev, VIRTIO_NET_F_MAC) &&
> + virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
> + struct scatterlist sg;
> +
> + sg_init_one(&sg, dev->dev_addr, dev->addr_len);
> + if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_MAC,
> + VIRTIO_NET_CTRL_MAC_ADDR_SET, &sg)) {
> + pr_debug("virtio_net: setting MAC address failed\n");
> + rtnl_unlock();
> + err = -EINVAL;
> + goto free_unregister_netdev;
Since the above is still dealing with device initialization, would it
make sense moving such init step before registering the netdevice?
Cheers,
Paolo
^ permalink raw reply [flat|nested] 21+ messages in thread* Re: [PATCH v3 2/2] virtio_net: notify MAC address change on device initialization
2023-01-31 9:01 ` Paolo Abeni
@ 2023-01-31 9:32 ` Laurent Vivier
-1 siblings, 0 replies; 21+ messages in thread
From: Laurent Vivier @ 2023-01-31 9:32 UTC (permalink / raw)
To: Paolo Abeni, linux-kernel
Cc: Cindy Lu, Michael S. Tsirkin, netdev, David S. Miller,
Eugenio Pérez, Eli Cohen, virtualization, Gautam Dawar
On 1/31/23 10:01, Paolo Abeni wrote:
> On Fri, 2023-01-27 at 21:45 +0100, Laurent Vivier wrote:
>> In virtnet_probe(), if the device doesn't provide a MAC address the
>> driver assigns a random one.
>> As we modify the MAC address we need to notify the device to allow it
>> to update all the related information.
>>
>> The problem can be seen with vDPA and mlx5_vdpa driver as it doesn't
>> assign a MAC address by default. The virtio_net device uses a random
>> MAC address (we can see it with "ip link"), but we can't ping a net
>> namespace from another one using the virtio-vdpa device because the
>> new MAC address has not been provided to the hardware:
>> RX packets are dropped since they don't go through the receive filters,
>> TX packets go through unaffected.
>>
>> Signed-off-by: Laurent Vivier <lvivier@redhat.com>
>> ---
>> drivers/net/virtio_net.c | 20 ++++++++++++++++++++
>> 1 file changed, 20 insertions(+)
>>
>> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
>> index 7d700f8e545a..704a05f1c279 100644
>> --- a/drivers/net/virtio_net.c
>> +++ b/drivers/net/virtio_net.c
>> @@ -3806,6 +3806,8 @@ static int virtnet_probe(struct virtio_device *vdev)
>> eth_hw_addr_set(dev, addr);
>> } else {
>> eth_hw_addr_random(dev);
>> + dev_info(&vdev->dev, "Assigned random MAC address %pM\n",
>> + dev->dev_addr);
>> }
>>
>> /* Set up our device-specific information */
>> @@ -3933,6 +3935,24 @@ static int virtnet_probe(struct virtio_device *vdev)
>>
>> virtio_device_ready(vdev);
>>
>> + /* a random MAC address has been assigned, notify the device.
>> + * We don't fail probe if VIRTIO_NET_F_CTRL_MAC_ADDR is not there
>> + * because many devices work fine without getting MAC explicitly
>> + */
>> + if (!virtio_has_feature(vdev, VIRTIO_NET_F_MAC) &&
>> + virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
>> + struct scatterlist sg;
>> +
>> + sg_init_one(&sg, dev->dev_addr, dev->addr_len);
>> + if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_MAC,
>> + VIRTIO_NET_CTRL_MAC_ADDR_SET, &sg)) {
>> + pr_debug("virtio_net: setting MAC address failed\n");
>> + rtnl_unlock();
>> + err = -EINVAL;
>> + goto free_unregister_netdev;
>
> Since the above is still dealing with device initialization, would it
> make sense moving such init step before registering the netdevice?
It depends if we can send the command using the control command queue or not.
I don't think we can use a vq before virtio_device_ready().
Thanks,
Laurent
_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization
^ permalink raw reply [flat|nested] 21+ messages in thread* Re: [PATCH v3 2/2] virtio_net: notify MAC address change on device initialization
@ 2023-01-31 9:32 ` Laurent Vivier
0 siblings, 0 replies; 21+ messages in thread
From: Laurent Vivier @ 2023-01-31 9:32 UTC (permalink / raw)
To: Paolo Abeni, linux-kernel
Cc: David S. Miller, virtualization, Cindy Lu, Michael S. Tsirkin,
Eli Cohen, Gautam Dawar, Eugenio Pérez, Jason Wang, netdev,
Parav Pandit
On 1/31/23 10:01, Paolo Abeni wrote:
> On Fri, 2023-01-27 at 21:45 +0100, Laurent Vivier wrote:
>> In virtnet_probe(), if the device doesn't provide a MAC address the
>> driver assigns a random one.
>> As we modify the MAC address we need to notify the device to allow it
>> to update all the related information.
>>
>> The problem can be seen with vDPA and mlx5_vdpa driver as it doesn't
>> assign a MAC address by default. The virtio_net device uses a random
>> MAC address (we can see it with "ip link"), but we can't ping a net
>> namespace from another one using the virtio-vdpa device because the
>> new MAC address has not been provided to the hardware:
>> RX packets are dropped since they don't go through the receive filters,
>> TX packets go through unaffected.
>>
>> Signed-off-by: Laurent Vivier <lvivier@redhat.com>
>> ---
>> drivers/net/virtio_net.c | 20 ++++++++++++++++++++
>> 1 file changed, 20 insertions(+)
>>
>> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
>> index 7d700f8e545a..704a05f1c279 100644
>> --- a/drivers/net/virtio_net.c
>> +++ b/drivers/net/virtio_net.c
>> @@ -3806,6 +3806,8 @@ static int virtnet_probe(struct virtio_device *vdev)
>> eth_hw_addr_set(dev, addr);
>> } else {
>> eth_hw_addr_random(dev);
>> + dev_info(&vdev->dev, "Assigned random MAC address %pM\n",
>> + dev->dev_addr);
>> }
>>
>> /* Set up our device-specific information */
>> @@ -3933,6 +3935,24 @@ static int virtnet_probe(struct virtio_device *vdev)
>>
>> virtio_device_ready(vdev);
>>
>> + /* a random MAC address has been assigned, notify the device.
>> + * We don't fail probe if VIRTIO_NET_F_CTRL_MAC_ADDR is not there
>> + * because many devices work fine without getting MAC explicitly
>> + */
>> + if (!virtio_has_feature(vdev, VIRTIO_NET_F_MAC) &&
>> + virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
>> + struct scatterlist sg;
>> +
>> + sg_init_one(&sg, dev->dev_addr, dev->addr_len);
>> + if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_MAC,
>> + VIRTIO_NET_CTRL_MAC_ADDR_SET, &sg)) {
>> + pr_debug("virtio_net: setting MAC address failed\n");
>> + rtnl_unlock();
>> + err = -EINVAL;
>> + goto free_unregister_netdev;
>
> Since the above is still dealing with device initialization, would it
> make sense moving such init step before registering the netdevice?
It depends if we can send the command using the control command queue or not.
I don't think we can use a vq before virtio_device_ready().
Thanks,
Laurent
^ permalink raw reply [flat|nested] 21+ messages in thread* Re: [PATCH v3 2/2] virtio_net: notify MAC address change on device initialization
2023-01-31 9:32 ` Laurent Vivier
@ 2023-01-31 11:41 ` Paolo Abeni
-1 siblings, 0 replies; 21+ messages in thread
From: Paolo Abeni @ 2023-01-31 11:41 UTC (permalink / raw)
To: Laurent Vivier, linux-kernel
Cc: Cindy Lu, Michael S. Tsirkin, netdev, David S. Miller,
Eugenio Pérez, Eli Cohen, virtualization, Gautam Dawar
On Tue, 2023-01-31 at 10:32 +0100, Laurent Vivier wrote:
> On 1/31/23 10:01, Paolo Abeni wrote:
> > On Fri, 2023-01-27 at 21:45 +0100, Laurent Vivier wrote:
> > > In virtnet_probe(), if the device doesn't provide a MAC address the
> > > driver assigns a random one.
> > > As we modify the MAC address we need to notify the device to allow it
> > > to update all the related information.
> > >
> > > The problem can be seen with vDPA and mlx5_vdpa driver as it doesn't
> > > assign a MAC address by default. The virtio_net device uses a random
> > > MAC address (we can see it with "ip link"), but we can't ping a net
> > > namespace from another one using the virtio-vdpa device because the
> > > new MAC address has not been provided to the hardware:
> > > RX packets are dropped since they don't go through the receive filters,
> > > TX packets go through unaffected.
> > >
> > > Signed-off-by: Laurent Vivier <lvivier@redhat.com>
> > > ---
> > > drivers/net/virtio_net.c | 20 ++++++++++++++++++++
> > > 1 file changed, 20 insertions(+)
> > >
> > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > > index 7d700f8e545a..704a05f1c279 100644
> > > --- a/drivers/net/virtio_net.c
> > > +++ b/drivers/net/virtio_net.c
> > > @@ -3806,6 +3806,8 @@ static int virtnet_probe(struct virtio_device *vdev)
> > > eth_hw_addr_set(dev, addr);
> > > } else {
> > > eth_hw_addr_random(dev);
> > > + dev_info(&vdev->dev, "Assigned random MAC address %pM\n",
> > > + dev->dev_addr);
> > > }
> > >
> > > /* Set up our device-specific information */
> > > @@ -3933,6 +3935,24 @@ static int virtnet_probe(struct virtio_device *vdev)
> > >
> > > virtio_device_ready(vdev);
> > >
> > > + /* a random MAC address has been assigned, notify the device.
> > > + * We don't fail probe if VIRTIO_NET_F_CTRL_MAC_ADDR is not there
> > > + * because many devices work fine without getting MAC explicitly
> > > + */
> > > + if (!virtio_has_feature(vdev, VIRTIO_NET_F_MAC) &&
> > > + virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
> > > + struct scatterlist sg;
> > > +
> > > + sg_init_one(&sg, dev->dev_addr, dev->addr_len);
> > > + if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_MAC,
> > > + VIRTIO_NET_CTRL_MAC_ADDR_SET, &sg)) {
> > > + pr_debug("virtio_net: setting MAC address failed\n");
> > > + rtnl_unlock();
> > > + err = -EINVAL;
> > > + goto free_unregister_netdev;
> >
> > Since the above is still dealing with device initialization, would it
> > make sense moving such init step before registering the netdevice?
>
> It depends if we can send the command using the control command queue or not.
> I don't think we can use a vq before virtio_device_ready().
Sounds reasonable. @Michael: do you have any additional comments?
Thanks!
Paolo
_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization
^ permalink raw reply [flat|nested] 21+ messages in thread* Re: [PATCH v3 2/2] virtio_net: notify MAC address change on device initialization
@ 2023-01-31 11:41 ` Paolo Abeni
0 siblings, 0 replies; 21+ messages in thread
From: Paolo Abeni @ 2023-01-31 11:41 UTC (permalink / raw)
To: Laurent Vivier, linux-kernel
Cc: David S. Miller, virtualization, Cindy Lu, Michael S. Tsirkin,
Eli Cohen, Gautam Dawar, Eugenio Pérez, Jason Wang, netdev,
Parav Pandit
On Tue, 2023-01-31 at 10:32 +0100, Laurent Vivier wrote:
> On 1/31/23 10:01, Paolo Abeni wrote:
> > On Fri, 2023-01-27 at 21:45 +0100, Laurent Vivier wrote:
> > > In virtnet_probe(), if the device doesn't provide a MAC address the
> > > driver assigns a random one.
> > > As we modify the MAC address we need to notify the device to allow it
> > > to update all the related information.
> > >
> > > The problem can be seen with vDPA and mlx5_vdpa driver as it doesn't
> > > assign a MAC address by default. The virtio_net device uses a random
> > > MAC address (we can see it with "ip link"), but we can't ping a net
> > > namespace from another one using the virtio-vdpa device because the
> > > new MAC address has not been provided to the hardware:
> > > RX packets are dropped since they don't go through the receive filters,
> > > TX packets go through unaffected.
> > >
> > > Signed-off-by: Laurent Vivier <lvivier@redhat.com>
> > > ---
> > > drivers/net/virtio_net.c | 20 ++++++++++++++++++++
> > > 1 file changed, 20 insertions(+)
> > >
> > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > > index 7d700f8e545a..704a05f1c279 100644
> > > --- a/drivers/net/virtio_net.c
> > > +++ b/drivers/net/virtio_net.c
> > > @@ -3806,6 +3806,8 @@ static int virtnet_probe(struct virtio_device *vdev)
> > > eth_hw_addr_set(dev, addr);
> > > } else {
> > > eth_hw_addr_random(dev);
> > > + dev_info(&vdev->dev, "Assigned random MAC address %pM\n",
> > > + dev->dev_addr);
> > > }
> > >
> > > /* Set up our device-specific information */
> > > @@ -3933,6 +3935,24 @@ static int virtnet_probe(struct virtio_device *vdev)
> > >
> > > virtio_device_ready(vdev);
> > >
> > > + /* a random MAC address has been assigned, notify the device.
> > > + * We don't fail probe if VIRTIO_NET_F_CTRL_MAC_ADDR is not there
> > > + * because many devices work fine without getting MAC explicitly
> > > + */
> > > + if (!virtio_has_feature(vdev, VIRTIO_NET_F_MAC) &&
> > > + virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
> > > + struct scatterlist sg;
> > > +
> > > + sg_init_one(&sg, dev->dev_addr, dev->addr_len);
> > > + if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_MAC,
> > > + VIRTIO_NET_CTRL_MAC_ADDR_SET, &sg)) {
> > > + pr_debug("virtio_net: setting MAC address failed\n");
> > > + rtnl_unlock();
> > > + err = -EINVAL;
> > > + goto free_unregister_netdev;
> >
> > Since the above is still dealing with device initialization, would it
> > make sense moving such init step before registering the netdevice?
>
> It depends if we can send the command using the control command queue or not.
> I don't think we can use a vq before virtio_device_ready().
Sounds reasonable. @Michael: do you have any additional comments?
Thanks!
Paolo
^ permalink raw reply [flat|nested] 21+ messages in thread
* Re: [PATCH v3 2/2] virtio_net: notify MAC address change on device initialization
2023-01-31 9:01 ` Paolo Abeni
@ 2023-01-31 13:43 ` Michael S. Tsirkin
-1 siblings, 0 replies; 21+ messages in thread
From: Michael S. Tsirkin @ 2023-01-31 13:43 UTC (permalink / raw)
To: Paolo Abeni
Cc: Laurent Vivier, Cindy Lu, netdev, Gautam Dawar, linux-kernel,
virtualization, Eugenio Pérez, Eli Cohen, David S. Miller
On Tue, Jan 31, 2023 at 10:01:53AM +0100, Paolo Abeni wrote:
> On Fri, 2023-01-27 at 21:45 +0100, Laurent Vivier wrote:
> > In virtnet_probe(), if the device doesn't provide a MAC address the
> > driver assigns a random one.
> > As we modify the MAC address we need to notify the device to allow it
> > to update all the related information.
> >
> > The problem can be seen with vDPA and mlx5_vdpa driver as it doesn't
> > assign a MAC address by default. The virtio_net device uses a random
> > MAC address (we can see it with "ip link"), but we can't ping a net
> > namespace from another one using the virtio-vdpa device because the
> > new MAC address has not been provided to the hardware:
> > RX packets are dropped since they don't go through the receive filters,
> > TX packets go through unaffected.
> >
> > Signed-off-by: Laurent Vivier <lvivier@redhat.com>
> > ---
> > drivers/net/virtio_net.c | 20 ++++++++++++++++++++
> > 1 file changed, 20 insertions(+)
> >
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index 7d700f8e545a..704a05f1c279 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> > @@ -3806,6 +3806,8 @@ static int virtnet_probe(struct virtio_device *vdev)
> > eth_hw_addr_set(dev, addr);
> > } else {
> > eth_hw_addr_random(dev);
> > + dev_info(&vdev->dev, "Assigned random MAC address %pM\n",
> > + dev->dev_addr);
> > }
> >
> > /* Set up our device-specific information */
> > @@ -3933,6 +3935,24 @@ static int virtnet_probe(struct virtio_device *vdev)
> >
> > virtio_device_ready(vdev);
> >
> > + /* a random MAC address has been assigned, notify the device.
> > + * We don't fail probe if VIRTIO_NET_F_CTRL_MAC_ADDR is not there
> > + * because many devices work fine without getting MAC explicitly
> > + */
> > + if (!virtio_has_feature(vdev, VIRTIO_NET_F_MAC) &&
> > + virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
> > + struct scatterlist sg;
> > +
> > + sg_init_one(&sg, dev->dev_addr, dev->addr_len);
> > + if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_MAC,
> > + VIRTIO_NET_CTRL_MAC_ADDR_SET, &sg)) {
> > + pr_debug("virtio_net: setting MAC address failed\n");
> > + rtnl_unlock();
> > + err = -EINVAL;
> > + goto free_unregister_netdev;
>
> Since the above is still dealing with device initialization, would it
> make sense moving such init step before registering the netdevice?
>
> Cheers,
>
> Paolo
We can't really, device has to be ready otherwise
we can't send commands to it.
--
MST
_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization
^ permalink raw reply [flat|nested] 21+ messages in thread* Re: [PATCH v3 2/2] virtio_net: notify MAC address change on device initialization
@ 2023-01-31 13:43 ` Michael S. Tsirkin
0 siblings, 0 replies; 21+ messages in thread
From: Michael S. Tsirkin @ 2023-01-31 13:43 UTC (permalink / raw)
To: Paolo Abeni
Cc: Laurent Vivier, linux-kernel, David S. Miller, virtualization,
Cindy Lu, Eli Cohen, Gautam Dawar, Eugenio Pérez, Jason Wang,
netdev, Parav Pandit
On Tue, Jan 31, 2023 at 10:01:53AM +0100, Paolo Abeni wrote:
> On Fri, 2023-01-27 at 21:45 +0100, Laurent Vivier wrote:
> > In virtnet_probe(), if the device doesn't provide a MAC address the
> > driver assigns a random one.
> > As we modify the MAC address we need to notify the device to allow it
> > to update all the related information.
> >
> > The problem can be seen with vDPA and mlx5_vdpa driver as it doesn't
> > assign a MAC address by default. The virtio_net device uses a random
> > MAC address (we can see it with "ip link"), but we can't ping a net
> > namespace from another one using the virtio-vdpa device because the
> > new MAC address has not been provided to the hardware:
> > RX packets are dropped since they don't go through the receive filters,
> > TX packets go through unaffected.
> >
> > Signed-off-by: Laurent Vivier <lvivier@redhat.com>
> > ---
> > drivers/net/virtio_net.c | 20 ++++++++++++++++++++
> > 1 file changed, 20 insertions(+)
> >
> > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > index 7d700f8e545a..704a05f1c279 100644
> > --- a/drivers/net/virtio_net.c
> > +++ b/drivers/net/virtio_net.c
> > @@ -3806,6 +3806,8 @@ static int virtnet_probe(struct virtio_device *vdev)
> > eth_hw_addr_set(dev, addr);
> > } else {
> > eth_hw_addr_random(dev);
> > + dev_info(&vdev->dev, "Assigned random MAC address %pM\n",
> > + dev->dev_addr);
> > }
> >
> > /* Set up our device-specific information */
> > @@ -3933,6 +3935,24 @@ static int virtnet_probe(struct virtio_device *vdev)
> >
> > virtio_device_ready(vdev);
> >
> > + /* a random MAC address has been assigned, notify the device.
> > + * We don't fail probe if VIRTIO_NET_F_CTRL_MAC_ADDR is not there
> > + * because many devices work fine without getting MAC explicitly
> > + */
> > + if (!virtio_has_feature(vdev, VIRTIO_NET_F_MAC) &&
> > + virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
> > + struct scatterlist sg;
> > +
> > + sg_init_one(&sg, dev->dev_addr, dev->addr_len);
> > + if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_MAC,
> > + VIRTIO_NET_CTRL_MAC_ADDR_SET, &sg)) {
> > + pr_debug("virtio_net: setting MAC address failed\n");
> > + rtnl_unlock();
> > + err = -EINVAL;
> > + goto free_unregister_netdev;
>
> Since the above is still dealing with device initialization, would it
> make sense moving such init step before registering the netdevice?
>
> Cheers,
>
> Paolo
We can't really, device has to be ready otherwise
we can't send commands to it.
--
MST
^ permalink raw reply [flat|nested] 21+ messages in thread* Re: [PATCH v3 2/2] virtio_net: notify MAC address change on device initialization
2023-01-31 13:43 ` Michael S. Tsirkin
@ 2023-01-31 14:40 ` Paolo Abeni
-1 siblings, 0 replies; 21+ messages in thread
From: Paolo Abeni @ 2023-01-31 14:40 UTC (permalink / raw)
To: Michael S. Tsirkin
Cc: Laurent Vivier, Cindy Lu, David S. Miller, netdev, linux-kernel,
virtualization, Eugenio Pérez, Gautam Dawar, Eli Cohen
On Tue, 2023-01-31 at 08:43 -0500, Michael S. Tsirkin wrote:
> On Tue, Jan 31, 2023 at 10:01:53AM +0100, Paolo Abeni wrote:
> > On Fri, 2023-01-27 at 21:45 +0100, Laurent Vivier wrote:
> > > In virtnet_probe(), if the device doesn't provide a MAC address the
> > > driver assigns a random one.
> > > As we modify the MAC address we need to notify the device to allow it
> > > to update all the related information.
> > >
> > > The problem can be seen with vDPA and mlx5_vdpa driver as it doesn't
> > > assign a MAC address by default. The virtio_net device uses a random
> > > MAC address (we can see it with "ip link"), but we can't ping a net
> > > namespace from another one using the virtio-vdpa device because the
> > > new MAC address has not been provided to the hardware:
> > > RX packets are dropped since they don't go through the receive filters,
> > > TX packets go through unaffected.
> > >
> > > Signed-off-by: Laurent Vivier <lvivier@redhat.com>
> > > ---
> > > drivers/net/virtio_net.c | 20 ++++++++++++++++++++
> > > 1 file changed, 20 insertions(+)
> > >
> > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > > index 7d700f8e545a..704a05f1c279 100644
> > > --- a/drivers/net/virtio_net.c
> > > +++ b/drivers/net/virtio_net.c
> > > @@ -3806,6 +3806,8 @@ static int virtnet_probe(struct virtio_device *vdev)
> > > eth_hw_addr_set(dev, addr);
> > > } else {
> > > eth_hw_addr_random(dev);
> > > + dev_info(&vdev->dev, "Assigned random MAC address %pM\n",
> > > + dev->dev_addr);
> > > }
> > >
> > > /* Set up our device-specific information */
> > > @@ -3933,6 +3935,24 @@ static int virtnet_probe(struct virtio_device *vdev)
> > >
> > > virtio_device_ready(vdev);
> > >
> > > + /* a random MAC address has been assigned, notify the device.
> > > + * We don't fail probe if VIRTIO_NET_F_CTRL_MAC_ADDR is not there
> > > + * because many devices work fine without getting MAC explicitly
> > > + */
> > > + if (!virtio_has_feature(vdev, VIRTIO_NET_F_MAC) &&
> > > + virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
> > > + struct scatterlist sg;
> > > +
> > > + sg_init_one(&sg, dev->dev_addr, dev->addr_len);
> > > + if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_MAC,
> > > + VIRTIO_NET_CTRL_MAC_ADDR_SET, &sg)) {
> > > + pr_debug("virtio_net: setting MAC address failed\n");
> > > + rtnl_unlock();
> > > + err = -EINVAL;
> > > + goto free_unregister_netdev;
> >
> > Since the above is still dealing with device initialization, would it
> > make sense moving such init step before registering the netdevice?
> >
> > Cheers,
> >
> > Paolo
>
> We can't really, device has to be ready otherwise
> we can't send commands to it.
Thanks, I see. Also, Laurent already clarified the above.
Should I read your comment as you are ok with the patches in the
current form?
Cheers,
Paolo
_______________________________________________
Virtualization mailing list
Virtualization@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/virtualization
^ permalink raw reply [flat|nested] 21+ messages in thread* Re: [PATCH v3 2/2] virtio_net: notify MAC address change on device initialization
@ 2023-01-31 14:40 ` Paolo Abeni
0 siblings, 0 replies; 21+ messages in thread
From: Paolo Abeni @ 2023-01-31 14:40 UTC (permalink / raw)
To: Michael S. Tsirkin
Cc: Laurent Vivier, Cindy Lu, netdev, Gautam Dawar, linux-kernel,
virtualization, Eugenio Pérez, Eli Cohen, David S. Miller
On Tue, 2023-01-31 at 08:43 -0500, Michael S. Tsirkin wrote:
> On Tue, Jan 31, 2023 at 10:01:53AM +0100, Paolo Abeni wrote:
> > On Fri, 2023-01-27 at 21:45 +0100, Laurent Vivier wrote:
> > > In virtnet_probe(), if the device doesn't provide a MAC address the
> > > driver assigns a random one.
> > > As we modify the MAC address we need to notify the device to allow it
> > > to update all the related information.
> > >
> > > The problem can be seen with vDPA and mlx5_vdpa driver as it doesn't
> > > assign a MAC address by default. The virtio_net device uses a random
> > > MAC address (we can see it with "ip link"), but we can't ping a net
> > > namespace from another one using the virtio-vdpa device because the
> > > new MAC address has not been provided to the hardware:
> > > RX packets are dropped since they don't go through the receive filters,
> > > TX packets go through unaffected.
> > >
> > > Signed-off-by: Laurent Vivier <lvivier@redhat.com>
> > > ---
> > > drivers/net/virtio_net.c | 20 ++++++++++++++++++++
> > > 1 file changed, 20 insertions(+)
> > >
> > > diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> > > index 7d700f8e545a..704a05f1c279 100644
> > > --- a/drivers/net/virtio_net.c
> > > +++ b/drivers/net/virtio_net.c
> > > @@ -3806,6 +3806,8 @@ static int virtnet_probe(struct virtio_device *vdev)
> > > eth_hw_addr_set(dev, addr);
> > > } else {
> > > eth_hw_addr_random(dev);
> > > + dev_info(&vdev->dev, "Assigned random MAC address %pM\n",
> > > + dev->dev_addr);
> > > }
> > >
> > > /* Set up our device-specific information */
> > > @@ -3933,6 +3935,24 @@ static int virtnet_probe(struct virtio_device *vdev)
> > >
> > > virtio_device_ready(vdev);
> > >
> > > + /* a random MAC address has been assigned, notify the device.
> > > + * We don't fail probe if VIRTIO_NET_F_CTRL_MAC_ADDR is not there
> > > + * because many devices work fine without getting MAC explicitly
> > > + */
> > > + if (!virtio_has_feature(vdev, VIRTIO_NET_F_MAC) &&
> > > + virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_MAC_ADDR)) {
> > > + struct scatterlist sg;
> > > +
> > > + sg_init_one(&sg, dev->dev_addr, dev->addr_len);
> > > + if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_MAC,
> > > + VIRTIO_NET_CTRL_MAC_ADDR_SET, &sg)) {
> > > + pr_debug("virtio_net: setting MAC address failed\n");
> > > + rtnl_unlock();
> > > + err = -EINVAL;
> > > + goto free_unregister_netdev;
> >
> > Since the above is still dealing with device initialization, would it
> > make sense moving such init step before registering the netdevice?
> >
> > Cheers,
> >
> > Paolo
>
> We can't really, device has to be ready otherwise
> we can't send commands to it.
Thanks, I see. Also, Laurent already clarified the above.
Should I read your comment as you are ok with the patches in the
current form?
Cheers,
Paolo
^ permalink raw reply [flat|nested] 21+ messages in thread