Re: [PATCH v7] can: virtio: Add virtio CAN driver

["Michael S. Tsirkin" <mst@redhat.com>]

On Tue, Feb 03, 2026 at 04:18:09PM +0100, Harald Mommer wrote:
> 
> 
> On 2/3/26 13:05, Michael S. Tsirkin wrote:
> > On Tue, Feb 03, 2026 at 12:55:07PM +0100, Harald Mommer wrote:
> >>
> >>
> >> On 1/9/26 18:23, Francesco Valla wrote:
> >>>> +static u8 virtio_can_send_ctrl_msg(struct net_device *ndev, u16 msg_type)
> >>>> +{
> >>>> +	struct scatterlist sg_out, sg_in, *sgs[2] = { &sg_out, &sg_in };
> >>>> +	struct virtio_can_priv *priv = netdev_priv(ndev);
> >>>> +	struct device *dev = &priv->vdev->dev;
> >>>> +	struct virtqueue *vq;
> >>>> +	unsigned int len;
> >>>> +	int err;
> >>>> +
> >>>> +	vq = priv->vqs[VIRTIO_CAN_QUEUE_CONTROL];
> >>> Nit: consider initializing this above, while declaring it.
> >>
> >> All those "Nit" regarding initialization cause problems. There is a reason why it was done the way it is.
> >>
> >> The network people require that the declaration lines are ordered by line length. longest line first. This is called "Reverse Christmas tree". Don't ask me why, this formatting style is what the network people require. Their subsystem, their rules.
> >>
> >> To initialize the vq you need now already the priv initialized. If now the vq line becomes longer than the priv line you will violate the special formatting requirements of the network subsystem.
> >>
> >> Solution was: What you see above.
> >>
> >> Regards
> >> Harald
> > 
> > So you reorder it then:
> > 
> > 	struct scatterlist sg_out, sg_in, *sgs[2] = { &sg_out, &sg_in };
> > 	struct virtqueue *vq = priv->vqs[VIRTIO_CAN_QUEUE_CONTROL]; // priv not initialized, will be done too late in the next line
> > 	struct virtio_can_priv *priv = netdev_priv(ndev); // you see it?
> > 	struct device *dev = &priv->vdev->dev;
> > 	unsigned int len;
> > 	int err;
> > 
> > 
> > and where is the problem?
> 
> The problem is that you use priv here to initialize vq in the line before priv is initialized.


Got it. Ignore the tree thing then. It's a guideline.


> > 
> > On the flip size, this guarantees we will not forget to initialize.
> 
> Static analysis is your friend.

And then someone monkey patches it to = NULL or something else silly.
I prefer correct by construction.