On Mon, Mar 28, 2022 at 4:08 AM Jiayu Hu <jiayu...@intel.com> wrote:
>
> NULL check for vq->async must be protected by lock. Otherwise, it is
> possible that the data plane thread dereferences vq->async with NULL
> value, since the control plane thread is freeing vq->async.
>
> Fixes: ee8024b3d4ad (vhost: move async data in dedicated structure)
> Cc: sta...@dpdk.org
>
> Signed-off-by: Jiayu Hu <jiayu...@intel.com>
> ---
> lib/vhost/vhost.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/lib/vhost/vhost.c b/lib/vhost/vhost.c
> index bc88148347..7f60c2824f 100644
> --- a/lib/vhost/vhost.c
> +++ b/lib/vhost/vhost.c
> @@ -1887,9 +1887,6 @@ rte_vhost_async_get_inflight(int vid, uint16_t queue_id)
> if (vq == NULL)
> return ret;
>
> - if (!vq->async)
> - return ret;
> -
> if (!rte_spinlock_trylock(&vq->access_lock)) {
> VHOST_LOG_CONFIG(DEBUG,
> "(%s) failed to check in-flight packets. virtqueue
> busy.\n",
> @@ -1897,6 +1894,9 @@ rte_vhost_async_get_inflight(int vid, uint16_t queue_id)
> return ret;
> }
>
> + if (!vq->async)
> + return ret;
Lock is still taken at this point.
FYI, I'll post a series to instrument locks in vhost, soon.
> +
> ret = vq->async->pkts_inflight_n;
> rte_spinlock_unlock(&vq->access_lock);
>
--
David Marchand
