| Message ID | 20210216110645.1087321-1-cohuck@redhat.com |
|---|---|
| State | Accepted |
| Commit | 182f709c5cff683e6732d04c78e328de0532284f |
| Headers | show |
| Series | [v2] virtio/s390: implement virtio-ccw revision 2 correctly | expand |
I was thinking of queuing this, but maybe it is quicker to pick it into the s390 tree directly and save us the extra pull request dance? Especially as this is a stable-worthy bugfix. On Tue, 16 Feb 2021 12:06:45 +0100 Cornelia Huck <cohuck@redhat.com> wrote: > CCW_CMD_READ_STATUS was introduced with revision 2 of virtio-ccw, > and drivers should only rely on it being implemented when they > negotiated at least that revision with the device. > > However, virtio_ccw_get_status() issued READ_STATUS for any > device operating at least at revision 1. If the device accepts > READ_STATUS regardless of the negotiated revision (which some > implementations like QEMU do, even though the spec currently does > not allow it), everything works as intended. While a device > rejecting the command should also be handled gracefully, we will > not be able to see any changes the device makes to the status, > such as setting NEEDS_RESET or setting the status to zero after > a completed reset. > > We negotiated the revision to at most 1, as we never bumped the > maximum revision; let's do that now and properly send READ_STATUS > only if we are operating at least at revision 2. > > Cc: stable@vger.kernel.org > Fixes: 7d3ce5ab9430 ("virtio/s390: support READ_STATUS command for virtio-ccw") > Reviewed-by: Halil Pasic <pasic@linux.ibm.com> > Signed-off-by: Cornelia Huck <cohuck@redhat.com> > --- > > v1->v2: > tweak patch description and cc:stable > > --- > drivers/s390/virtio/virtio_ccw.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/drivers/s390/virtio/virtio_ccw.c b/drivers/s390/virtio/virtio_ccw.c > index 5730572b52cd..54e686dca6de 100644 > --- a/drivers/s390/virtio/virtio_ccw.c > +++ b/drivers/s390/virtio/virtio_ccw.c > @@ -117,7 +117,7 @@ struct virtio_rev_info { > }; > > /* the highest virtio-ccw revision we support */ > -#define VIRTIO_CCW_REV_MAX 1 > +#define VIRTIO_CCW_REV_MAX 2 > > struct virtio_ccw_vq_info { > struct virtqueue *vq; > @@ -952,7 +952,7 @@ static u8 virtio_ccw_get_status(struct virtio_device *vdev) > u8 old_status = vcdev->dma_area->status; > struct ccw1 *ccw; > > - if (vcdev->revision < 1) > + if (vcdev->revision < 2) > return vcdev->dma_area->status; > > ccw = ccw_device_dma_zalloc(vcdev->cdev, sizeof(*ccw));
On Fri, Feb 19, 2021 at 05:38:28PM +0100, Cornelia Huck wrote: > I was thinking of queuing this, but maybe it is quicker to pick it into > the s390 tree directly and save us the extra pull request dance? > Especially as this is a stable-worthy bugfix. Yes, sure. I'll pick it up. Thanks.
diff --git a/drivers/s390/virtio/virtio_ccw.c b/drivers/s390/virtio/virtio_ccw.c index 5730572b52cd..54e686dca6de 100644 --- a/drivers/s390/virtio/virtio_ccw.c +++ b/drivers/s390/virtio/virtio_ccw.c @@ -117,7 +117,7 @@ struct virtio_rev_info { }; /* the highest virtio-ccw revision we support */ -#define VIRTIO_CCW_REV_MAX 1 +#define VIRTIO_CCW_REV_MAX 2 struct virtio_ccw_vq_info { struct virtqueue *vq; @@ -952,7 +952,7 @@ static u8 virtio_ccw_get_status(struct virtio_device *vdev) u8 old_status = vcdev->dma_area->status; struct ccw1 *ccw; - if (vcdev->revision < 1) + if (vcdev->revision < 2) return vcdev->dma_area->status; ccw = ccw_device_dma_zalloc(vcdev->cdev, sizeof(*ccw));
CCW_CMD_READ_STATUS was introduced with revision 2 of virtio-ccw, and drivers should only rely on it being implemented when they negotiated at least that revision with the device. However, virtio_ccw_get_status() issued READ_STATUS for any device operating at least at revision 1. If the device accepts READ_STATUS regardless of the negotiated revision (which some implementations like QEMU do, even though the spec currently does not allow it), everything works as intended. While a device rejecting the command should also be handled gracefully, we will not be able to see any changes the device makes to the status, such as setting NEEDS_RESET or setting the status to zero after a completed reset. We negotiated the revision to at most 1, as we never bumped the maximum revision; let's do that now and properly send READ_STATUS only if we are operating at least at revision 2. Cc: stable@vger.kernel.org Fixes: 7d3ce5ab9430 ("virtio/s390: support READ_STATUS command for virtio-ccw") Reviewed-by: Halil Pasic <pasic@linux.ibm.com> Signed-off-by: Cornelia Huck <cohuck@redhat.com> --- v1->v2: tweak patch description and cc:stable --- drivers/s390/virtio/virtio_ccw.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)