Re: [RFC 1/2] net: xdp: introduce bulking for xdp tx return path

[Jesper Dangaard Brouer <brouer@redhat.com>]

On Tue, 20 Oct 2020 11:33:37 +0200
Lorenzo Bianconi <lorenzo@kernel.org> wrote:

> diff --git a/drivers/net/ethernet/marvell/mvneta.c b/drivers/net/ethernet/marvell/mvneta.c
> index 54b0bf574c05..af33cc62ed4c 100644
> --- a/drivers/net/ethernet/marvell/mvneta.c
> +++ b/drivers/net/ethernet/marvell/mvneta.c
> @@ -663,6 +663,8 @@ struct mvneta_tx_queue {
>  
>  	/* Affinity mask for CPUs*/
>  	cpumask_t affinity_mask;
> +
> +	struct xdp_frame_bulk bq;
>  };
>  
>  struct mvneta_rx_queue {
> @@ -1854,12 +1856,10 @@ static void mvneta_txq_bufs_free(struct mvneta_port *pp,
>  			dev_kfree_skb_any(buf->skb);
>  		} else if (buf->type == MVNETA_TYPE_XDP_TX ||
>  			   buf->type == MVNETA_TYPE_XDP_NDO) {
> -			if (napi && buf->type == MVNETA_TYPE_XDP_TX)
> -				xdp_return_frame_rx_napi(buf->xdpf);
> -			else
> -				xdp_return_frame(buf->xdpf);
> +			xdp_return_frame_bulk(buf->xdpf, &txq->bq, napi);

Hmm, I don't think you can use 'napi' directly here.

You are circumventing check (buf->type == MVNETA_TYPE_XDP_TX), and will
now also allow XDP_NDO (XDP_REDIRECT) to basically use xdp_return_frame_rx_napi().


>  		}
>  	}
> +	xdp_flush_frame_bulk(&txq->bq, napi);
>  
>  	netdev_tx_completed_queue(nq, pkts_compl, bytes_compl);
>  }



-- 
Best regards,
  Jesper Dangaard Brouer
  MSc.CS, Principal Kernel Engineer at Red Hat
  LinkedIn: http://www.linkedin.com/in/brouer