Re: [PATCH] virtio_ring: add an error code check in virtqueue_resize

[Xuan Zhuo]

On Fri, 20 Oct 2023 06:08:06 -0400, "Michael S. Tsirkin" <mst@xxxxxxxxxx> wrote:
> On Fri, Oct 20, 2023 at 05:50:22PM +0800, Xuan Zhuo wrote:
> > On Fri, 20 Oct 2023 05:42:14 -0400, "Michael S. Tsirkin" <mst@xxxxxxxxxx> wrote:
> > > On Fri, Oct 20, 2023 at 05:36:41PM +0800, Xuan Zhuo wrote:
> > > > On Fri, 20 Oct 2023 05:34:32 -0400, "Michael S. Tsirkin" <mst@xxxxxxxxxx> wrote:
> > > > > On Fri, Oct 20, 2023 at 05:23:21PM +0800, Su Hui wrote:
> > > > > > virtqueue_resize_packed() or virtqueue_resize_split() can return
> > > > > > error code if failed, so add a check for this.
> > > > > >
> > > > > > Signed-off-by: Su Hui <suhui@xxxxxxxxxxxx>
> > > > > > ---
> > > > > >
> > > > > > I'm not sure that return directly is right or not,
> > > > > > maybe there are some process should do before return.
> > > > >
> > > > > yes - presizely what virtqueue_enable_after_reset does.
> > > > >
> > > > > Error handling in virtqueue_enable_after_reset is really weird BTW.
> > > > > For some reason it overrides the error code returned.
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > >  drivers/virtio/virtio_ring.c | 3 +++
> > > > > >  1 file changed, 3 insertions(+)
> > > > > >
> > > > > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
> > > > > > index 51d8f3299c10..cf662c3a755b 100644
> > > > > > --- a/drivers/virtio/virtio_ring.c
> > > > > > +++ b/drivers/virtio/virtio_ring.c
> > > > > > @@ -2759,6 +2759,9 @@ int virtqueue_resize(struct virtqueue *_vq, u32 num,
> > > > > >  	else
> > > > > >  		err = virtqueue_resize_split(_vq, num);
> > > > > >
> > > > > > +	if (err)
> > > > > > +		return err;
> > > > > > +
> > > > > >  	return virtqueue_enable_after_reset(_vq);
> > > > >
> > > > > So I think it should be something like:
> > > > >
> > > > > 	int err_reset = virtqueue_enable_after_reset(_vq);
> > > > > 	BUG_ON(err_reset);
> > > > >
> > > > > 	return err;
> > > > >
> > > >
> > > > How about WARN and vq->broken?
> > > >
> > > > Thanks.
> > >
> > > Well, what are the cases where it can happen practically?
> >
> > Device error. Such as vp_active_vq()
> >
> > Thanks.
>
> Hmm interesting. OK. But do callers know to recover?


No.

So I think WARN + broken is suitable.

Thanks.


>
> >
> > >
> > > >
> > > > >
> > > > >
> > > > > >  }
> > > > > >  EXPORT_SYMBOL_GPL(virtqueue_resize);
> > > > > > --
> > > > > > 2.30.2
> > > > >
> > >
>
> _______________________________________________
> Virtualization mailing list
> Virtualization@xxxxxxxxxxxxxxxxxxxxxxxxxx
> https://lists.linuxfoundation.org/mailman/listinfo/virtualization