Acked-by: Viacheslav Ovsiienko <viachesl...@nvidia.com>
> -----Original Message-----
> From: Gavin Hu <g...@nvidia.com>
> Sent: Friday, December 6, 2024 2:58 AM
> To: dev@dpdk.org
> Cc: sta...@dpdk.org; Dariusz Sosnowski <dsosnow...@nvidia.com>; Slava
> Ovsiienko <viachesl...@nvidia.com>; Bing Zhao <bi...@nvidia.com>; Ori
> Kam <or...@nvidia.com>; Suanming Mou <suanmi...@nvidia.com>; Matan
> Azrad <ma...@nvidia.com>; Alexander Kozyrev <akozy...@nvidia.com>
> Subject: [PATCH] net/mlx5: do not poll CQEs when no available elts
>
> In certain situations, the receive queue (rxq) fails to replenish its
> internal ring
> with memory buffers (mbufs) from the pool. This can happen when the pool
> has a limited number of mbufs allocated, and the user application holds
> incoming packets for an extended period, resulting in a delayed release of
> mbufs. Consequently, the pool becomes depleted, preventing the rxq from
> replenishing from it.
>
> There was a bug in the behavior of the vectorized rxq_cq_process_v routine,
> which handled completion queue entries (CQEs) in batches of four. This
> routine consistently accessed four mbufs from the internal queue ring,
> regardless of whether they had been replenished. As a result, it could access
> mbufs that no longer belonged to the poll mode driver (PMD).
>
> The fix involves checking if there are four replenished mbufs available before
> allowing rxq_cq_process_v to handle the batch. Once replenishment succeeds
> during the polling process, the routine will resume its operation.
>
> Fixes: 1ded26239aa0 ("net/mlx5: refactor vectorized Rx")
> Cc: sta...@dpdk.org
>
> Reported-by: Changqi Dingluo <dingluochangqi...@bytedance.com>
> Signed-off-by: Gavin Hu <g...@nvidia.com>
> ---
> drivers/net/mlx5/mlx5_rxtx_vec.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/drivers/net/mlx5/mlx5_rxtx_vec.c
> b/drivers/net/mlx5/mlx5_rxtx_vec.c
> index 1872bf310c..1b701801c5 100644
> --- a/drivers/net/mlx5/mlx5_rxtx_vec.c
> +++ b/drivers/net/mlx5/mlx5_rxtx_vec.c
> @@ -325,6 +325,9 @@ rxq_burst_v(struct mlx5_rxq_data *rxq, struct
> rte_mbuf **pkts,
> /* Not to cross queue end. */
> pkts_n = RTE_MIN(pkts_n, q_n - elts_idx);
> pkts_n = RTE_MIN(pkts_n, q_n - cq_idx);
> + /* Not to move past the allocated mbufs. */
> + pkts_n = RTE_MIN(pkts_n, RTE_ALIGN_FLOOR(rxq->rq_ci - rxq-
> >rq_pi,
> +
> MLX5_VPMD_DESCS_PER_LOOP));
> if (!pkts_n) {
> *no_cq = !rcvd_pkt;
> return rcvd_pkt;
> --
> 2.18.2
