[PATCH 2/4] virtio_net: Add a virtqueue for outbound control commands

[PATCH 2/4] virtio_net: Add a virtqueue for outbound control commands

[Alex Williamson]

This will be used for RX mode, MAC filter table, VLAN filtering, etc...

Signed-off-by: Alex Williamson <alex.williamson@hp.com>
---

 drivers/net/virtio_net.c   |   55 +++++++++++++++++++++++++++++++++++++++++++-
 include/linux/virtio_net.h |    3 ++
 2 files changed, 57 insertions(+), 1 deletions(-)


diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index e7700de..de348de 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -39,7 +39,7 @@ module_param(gso, bool, 0444);
 struct virtnet_info
 {
 	struct virtio_device *vdev;
-	struct virtqueue *rvq, *svq;
+	struct virtqueue *rvq, *svq, *cvq;
 	struct net_device *dev;
 	struct napi_struct napi;
 
@@ -603,6 +603,47 @@ static int virtnet_open(struct net_device *dev)
 	return 0;
 }
 
+static int virtnet_send_command(struct virtnet_info *vi, u8 class, u8 cmd,
+				void *data, unsigned int len)
+{
+	struct scatterlist sg[3];
+	struct {
+		u8 class;
+		u8 cmd;
+	} ctrl_cmd;
+	u8 ctrl_status;
+	unsigned int tmp;
+	int i = 0;
+
+	if (!vi->cvq)
+		return -EFAULT;
+
+	sg_init_table(sg, len ? 3 : 2);
+
+	sg_set_buf(&sg[i++], &ctrl_cmd, sizeof(ctrl_cmd));
+	if (len)
+		sg_set_buf(&sg[i++], data, len);
+	sg_set_buf(&sg[i], &ctrl_status, sizeof(ctrl_status));
+
+	ctrl_cmd.class = class;
+	ctrl_cmd.cmd = cmd;
+
+	ctrl_status = ~0;
+
+	if (vi->cvq->vq_ops->add_buf(vi->cvq, sg, i, 1, vi) != 0)
+		BUG();
+
+	vi->cvq->vq_ops->kick(vi->cvq);
+
+	while (!vi->cvq->vq_ops->get_buf(vi->cvq, &tmp))
+		cpu_relax();
+
+	if (ctrl_status == VIRTIO_NET_OK)
+		return 0;
+	else
+		return -EFAULT;
+}
+
 static int virtnet_close(struct net_device *dev)
 {
 	struct virtnet_info *vi = netdev_priv(dev);
@@ -733,6 +774,14 @@ static int virtnet_probe(struct virtio_device *vdev)
 		goto free_recv;
 	}
 
+	/*
+	 * Outbound control channel virtqueue.  We can live without it,
+	 * so don't go fatal if it's not there.
+	 */
+	vi->cvq = vdev->config->find_vq(vdev, 2, NULL);
+	if (IS_ERR(vi->cvq))
+		vi->cvq = NULL;
+
 	/* Initialize our empty receive and send queues. */
 	skb_queue_head_init(&vi->recv);
 	skb_queue_head_init(&vi->send);
@@ -763,6 +812,8 @@ static int virtnet_probe(struct virtio_device *vdev)
 unregister:
 	unregister_netdev(dev);
 free_send:
+	if (vi->cvq)
+		vdev->config->del_vq(vi->cvq);
 	vdev->config->del_vq(vi->svq);
 free_recv:
 	vdev->config->del_vq(vi->rvq);
@@ -793,6 +844,8 @@ static void virtnet_remove(struct virtio_device *vdev)
 
 	vdev->config->del_vq(vi->svq);
 	vdev->config->del_vq(vi->rvq);
+	if (vi->cvq)
+		vdev->config->del_vq(vi->cvq);
 	unregister_netdev(vi->dev);
 
 	while (vi->pages)
diff --git a/include/linux/virtio_net.h b/include/linux/virtio_net.h
index 5cdd0aa..1de7c86 100644
--- a/include/linux/virtio_net.h
+++ b/include/linux/virtio_net.h
@@ -53,4 +53,7 @@ struct virtio_net_hdr_mrg_rxbuf {
 	__u16 num_buffers;	/* Number of merged rx buffers */
 };
 
+#define VIRTIO_NET_OK     0
+#define VIRTIO_NET_ERR    1
+
 #endif /* _LINUX_VIRTIO_NET_H */

Re: [PATCH 2/4] virtio_net: Add a virtqueue for outbound control commands

[Mark McLoughlin]

On Tue, 2009-01-13 at 14:23 -0700, Alex Williamson wrote:
> This will be used for RX mode, MAC filter table, VLAN filtering, etc...

Looks very reasonable. I'm a bit wary of send_command() being
synchronous, but it probably makes sense.

Could do with some details in the commit log as to why this approach was
chosen over increasing virtio_net_config.

> Signed-off-by: Alex Williamson <alex.williamson@hp.com>
> ---
> 
>  drivers/net/virtio_net.c   |   55 +++++++++++++++++++++++++++++++++++++++++++-
>  include/linux/virtio_net.h |    3 ++
>  2 files changed, 57 insertions(+), 1 deletions(-)
> 
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index e7700de..de348de 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -39,7 +39,7 @@ module_param(gso, bool, 0444);
>  struct virtnet_info
>  {
>  	struct virtio_device *vdev;
> -	struct virtqueue *rvq, *svq;
> +	struct virtqueue *rvq, *svq, *cvq;
>  	struct net_device *dev;
>  	struct napi_struct napi;
>  
> @@ -603,6 +603,47 @@ static int virtnet_open(struct net_device *dev)
>  	return 0;
>  }
>  
> +static int virtnet_send_command(struct virtnet_info *vi, u8 class, u8 cmd,
> +				void *data, unsigned int len)
> +{
> +	struct scatterlist sg[3];
> +	struct {
> +		u8 class;
> +		u8 cmd;
> +	} ctrl_cmd;

I'd like to see this defined in virtio_net_hdr.

Why the need for class/cmd? Why not just a single 16 bit command field?

> +	u8 ctrl_status;
> +	unsigned int tmp;
> +	int i = 0;
> +
> +	if (!vi->cvq)
> +		return -EFAULT;

BUG_ON() probably makes more sense here.

> +
> +	sg_init_table(sg, len ? 3 : 2);
> +
> +	sg_set_buf(&sg[i++], &ctrl_cmd, sizeof(ctrl_cmd));
> +	if (len)
> +		sg_set_buf(&sg[i++], data, len);
> +	sg_set_buf(&sg[i], &ctrl_status, sizeof(ctrl_status));
> +
> +	ctrl_cmd.class = class;
> +	ctrl_cmd.cmd = cmd;
> +
> +	ctrl_status = ~0;
> +
> +	if (vi->cvq->vq_ops->add_buf(vi->cvq, sg, i, 1, vi) != 0)
> +		BUG();
> +
> +	vi->cvq->vq_ops->kick(vi->cvq);
> +
> +	while (!vi->cvq->vq_ops->get_buf(vi->cvq, &tmp))
> +		cpu_relax();
> +
> +	if (ctrl_status == VIRTIO_NET_OK)
> +		return 0;
> +	else
> +		return -EFAULT;

Could be all on one line:

        return ctrl_status ? -EFAULT : 0;

Cheers,
Mark.

Re: [PATCH 2/4] virtio_net: Add a virtqueue for outbound control commands

[Alex Williamson]

On Wed, 2009-01-14 at 10:15 +0000, Mark McLoughlin wrote:
> On Tue, 2009-01-13 at 14:23 -0700, Alex Williamson wrote:
> > This will be used for RX mode, MAC filter table, VLAN filtering, etc...
> 
> Looks very reasonable. I'm a bit wary of send_command() being
> synchronous, but it probably makes sense.

Yes, I tried to make it async, but quickly ran into problems sleeping
when I wasn't allowed to.  Using it in this way does pretty much limit
it to an outbound channel though... all worthy of commit log comments.

> > +static int virtnet_send_command(struct virtnet_info *vi, u8 class, u8 cmd,
> > +				void *data, unsigned int len)
> > +{
> > +	struct scatterlist sg[3];
> > +	struct {
> > +		u8 class;
> > +		u8 cmd;
> > +	} ctrl_cmd;
> 
> I'd like to see this defined in virtio_net_hdr.

As part of struct virtio_net_hdr?  I'm not sure what that'd buy us and
would likely break compatibility.  Or do you simply mean defined in
virtio-net.h?

> Why the need for class/cmd? Why not just a single 16 bit command field?

It seemed like a good way to logically divide up an address space and
makes it easy for the backend to break up the code so it doesn't become
a huge table.

> > +	u8 ctrl_status;
> > +	unsigned int tmp;
> > +	int i = 0;
> > +
> > +	if (!vi->cvq)
> > +		return -EFAULT;
> 
> BUG_ON() probably makes more sense here.

This is to allow a newer virtio_net guest driver to run on an old qemu.
That's why I don't generate a fatal error if we don't find the control
queue.  In that case the backend will be running in promiscuous mode and
I think all of these commands can safely fail.  Thanks for the comments.

Alex

-- 
Alex Williamson                             HP Open Source & Linux Org.

Re: [PATCH 2/4] virtio_net: Add a virtqueue for outbound control commands

[Mark McLoughlin]

On Wed, 2009-01-14 at 09:01 -0700, Alex Williamson wrote:
> On Wed, 2009-01-14 at 10:15 +0000, Mark McLoughlin wrote:
> > On Tue, 2009-01-13 at 14:23 -0700, Alex Williamson wrote:

> > > +static int virtnet_send_command(struct virtnet_info *vi, u8 class, u8 cmd,
> > > +				void *data, unsigned int len)
> > > +{
> > > +	struct scatterlist sg[3];
> > > +	struct {
> > > +		u8 class;
> > > +		u8 cmd;
> > > +	} ctrl_cmd;
> > 
> > I'd like to see this defined in virtio_net_hdr.
> 
> As part of struct virtio_net_hdr?  I'm not sure what that'd buy us and
> would likely break compatibility.  Or do you simply mean defined in
> virtio-net.h?

Yep, sorry, I meant virtio-net.h

> > > +	u8 ctrl_status;
> > > +	unsigned int tmp;
> > > +	int i = 0;
> > > +
> > > +	if (!vi->cvq)
> > > +		return -EFAULT;
> > 
> > BUG_ON() probably makes more sense here.
> 
> This is to allow a newer virtio_net guest driver to run on an old qemu.
> That's why I don't generate a fatal error if we don't find the control
> queue.  In that case the backend will be running in promiscuous mode and
> I think all of these commands can safely fail.  Thanks for the comments.

Yep, but what I mean is that virtio_net_send_command() should never be
called if the host doesn't have send queue support - i.e. force the
calling code to think about what should be done on older hosts.

Cheers,
Mark.