Received-SPF: pass (google.com: domain of linux-kernel+bounces-55144-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
From: Stefano Garzarella <sgarzare@redhat.com>
To: virtualization@lists.linux.dev
Cc: Shannon Nelson <shannon.nelson@amd.com>,
	=?UTF-8?q?Eugenio=20P=C3=A9rez?= <eperezma@redhat.com>,
	"Michael S. Tsirkin" <mst@redhat.com>,
	kvm@vger.kernel.org,
	Kevin Wolf <kwolf@redhat.com>,
	Jason Wang <jasowang@redhat.com>,
	netdev@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	Stefano Garzarella <sgarzare@redhat.com>
Subject: [PATCH] vhost-vdpa: fail enabling virtqueue in certain conditions
Date: Tue,  6 Feb 2024 15:51:54 +0100
Message-ID: <20240206145154.118044-1-sgarzare@redhat.com>
Precedence: bulk
MIME-Version: 1.0
Content-type: text/plain
Content-Transfer-Encoding: 8bit

If VHOST_BACKEND_F_ENABLE_AFTER_DRIVER_OK is not negotiated, we expect
the driver to enable virtqueue before setting DRIVER_OK. If the driver
tries anyway, better to fail right away as soon as we get the ioctl.
Let's also update the documentation to make it clearer.

We had a problem in QEMU for not meeting this requirement, see
https://lore.kernel.org/qemu-devel/20240202132521.32714-1-kwolf@redhat.com/

Fixes: 9f09fd6171fe ("vdpa: accept VHOST_BACKEND_F_ENABLE_AFTER_DRIVER_OK backend feature")
Cc: eperezma@redhat.com
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
---
 include/uapi/linux/vhost_types.h | 3 ++-
 drivers/vhost/vdpa.c             | 4 ++++
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/include/uapi/linux/vhost_types.h b/include/uapi/linux/vhost_types.h
index d7656908f730..5df49b6021a7 100644
--- a/include/uapi/linux/vhost_types.h
+++ b/include/uapi/linux/vhost_types.h
@@ -182,7 +182,8 @@ struct vhost_vdpa_iova_range {
 /* Device can be resumed */
 #define VHOST_BACKEND_F_RESUME  0x5
 /* Device supports the driver enabling virtqueues both before and after
- * DRIVER_OK
+ * DRIVER_OK. If this feature is not negotiated, the virtqueues must be
+ * enabled before setting DRIVER_OK.
  */
 #define VHOST_BACKEND_F_ENABLE_AFTER_DRIVER_OK  0x6
 /* Device may expose the virtqueue's descriptor area, driver area and
diff --git a/drivers/vhost/vdpa.c b/drivers/vhost/vdpa.c
index bc4a51e4638b..1fba305ba8c1 100644
--- a/drivers/vhost/vdpa.c
+++ b/drivers/vhost/vdpa.c
@@ -651,6 +651,10 @@ static long vhost_vdpa_vring_ioctl(struct vhost_vdpa *v, unsigned int cmd,
 	case VHOST_VDPA_SET_VRING_ENABLE:
 		if (copy_from_user(&s, argp, sizeof(s)))
 			return -EFAULT;
+		if (!vhost_backend_has_feature(vq,
+			VHOST_BACKEND_F_ENABLE_AFTER_DRIVER_OK) &&
+		    (ops->get_status(vdpa) & VIRTIO_CONFIG_S_DRIVER_OK))
+			return -EINVAL;
 		ops->set_vq_ready(vdpa, idx, s.num);
 		return 0;
 	case VHOST_VDPA_GET_VRING_GROUP:
-- 
2.43.0