If mlx5_vdpa gets unloaded while a VM is running, the workqueue will be
destroyed. However, vhost might still have reference to the kick
function and might attempt to push new works. This could lead to null
pointer dereference.
To fix this, set mvdev->wq to NULL just before destroying and verify
that the workqueue is not NULL in mlx5_vdpa_kick_vq before attempting to
push a new work.
Fixes: 5262912ef3cf ("vdpa/mlx5: Add support for control VQ and MAC setting")
Signed-off-by: Eli Cohen <elic@xxxxxxxxxx>
Link: https://lore.kernel.org/r/20220321141303.9586-1-elic@xxxxxxxxxx
Signed-off-by: Michael S. Tsirkin <mst@xxxxxxxxxx>
diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
index c0f0ecb82c6f..2f4fb09f1e89 100644
--- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
+++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
@@ -1712,7 +1712,7 @@ static void mlx5_vdpa_kick_vq(struct vdpa_device *vdev, u16 idx)
return;
if (unlikely(is_ctrl_vq_idx(mvdev, idx))) {
- if (!mvdev->cvq.ready)
+ if (!mvdev->wq || !mvdev->cvq.ready)
return;
wqent = kzalloc(sizeof(*wqent), GFP_ATOMIC);
@@ -2779,9 +2779,12 @@ static void mlx5_vdpa_dev_del(struct vdpa_mgmt_dev *v_mdev, struct vdpa_device *
struct mlx5_vdpa_mgmtdev *mgtdev = container_of(v_mdev, struct mlx5_vdpa_mgmtdev, mgtdev);
struct mlx5_vdpa_dev *mvdev = to_mvdev(dev);
struct mlx5_vdpa_net *ndev = to_mlx5_vdpa_ndev(mvdev);
+ struct workqueue_struct *wq;
mlx5_notifier_unregister(mvdev->mdev, &ndev->nb);
- destroy_workqueue(mvdev->wq);
+ wq = mvdev->wq;
+ mvdev->wq = NULL;
+ destroy_workqueue(wq);
_vdpa_unregister_device(dev);
mgtdev->ndev = NULL;
}
