Re: [RFC PATCH net-next 07/10] net: hibmcge: Implement rx_poll function to receive packets

[Joe Damato <jdamato@fastly.com>]

On Wed, Jul 31, 2024 at 05:42:42PM +0800, Jijie Shao wrote:
> Implement rx_poll function to read the rx descriptor after
> receiving the rx interrupt. Adjust the skb based on the
> descriptor to complete the reception of the packet.
> 
> Signed-off-by: Jijie Shao <shaojijie@huawei.com>
> ---
>  .../ethernet/hisilicon/hibmcge/hbg_common.h   |   5 +
>  .../net/ethernet/hisilicon/hibmcge/hbg_hw.c   |  10 ++
>  .../net/ethernet/hisilicon/hibmcge/hbg_hw.h   |   1 +
>  .../net/ethernet/hisilicon/hibmcge/hbg_irq.c  |   9 +-
>  .../net/ethernet/hisilicon/hibmcge/hbg_main.c |   2 +
>  .../net/ethernet/hisilicon/hibmcge/hbg_reg.h  |   2 +
>  .../hisilicon/hibmcge/hbg_reg_union.h         |  65 ++++++++
>  .../net/ethernet/hisilicon/hibmcge/hbg_txrx.c | 157 +++++++++++++++++-
>  8 files changed, 248 insertions(+), 3 deletions(-)
 
> diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_main.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_main.c
> index 8efeea9b0c26..bb5f8321da8a 100644
> --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_main.c
> +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_main.c
> @@ -36,6 +36,7 @@ static int hbg_net_open(struct net_device *dev)
>  		return 0;
>  
>  	netif_carrier_off(dev);
> +	napi_enable(&priv->rx_ring.napi);
>  	napi_enable(&priv->tx_ring.napi);
>  	hbg_enable_intr(priv, true);
>  	hbg_hw_mac_enable(priv, HBG_STATUS_ENABLE);

In the future, it might be good to consider using:
   - netif_napi_set_irq
   - netif_queue_set_napi
 
to link NAPIs with IRQs and queues.

For an example, see 64b62146ba9e ("net/mlx4: link NAPI instances to
queues and IRQs).

[...]

> +static int hbg_rx_fill_buffers(struct hbg_priv *priv)
> +{
> +	struct hbg_ring *ring = &priv->rx_ring;
> +	int ret;
> +
> +	while (!(hbg_fifo_is_full(priv, ring->dir) ||
> +		 hbg_queue_is_full(ring->ntc, ring->ntu, ring))) {
> +		ret = hbg_rx_fill_one_buffer(priv);
> +		if (ret)
> +			return ret;
> +	}
> +
> +	return 0;
> +}
> +
> +static bool hbg_sync_data_from_hw(struct hbg_priv *priv,
> +				  struct hbg_buffer *buffer)
> +{
> +	struct hbg_rx_desc *rx_desc;
> +
> +	/* make sure HW write desc complete */
> +	dma_rmb();
> +
> +	dma_sync_single_for_cpu(&priv->pdev->dev, buffer->skb_dma,
> +				buffer->skb_len, DMA_FROM_DEVICE);
> +
> +	rx_desc = (struct hbg_rx_desc *)buffer->skb->data;
> +	return rx_desc->len != 0;
> +}

Have you looked into using the page pool to simplify some of the
logic above?

> +static int hbg_napi_rx_poll(struct napi_struct *napi, int budget)
> +{
> +	struct hbg_ring *ring = container_of(napi, struct hbg_ring, napi);
> +	struct hbg_priv *priv = ring->priv;
> +	struct hbg_rx_desc *rx_desc;
> +	struct hbg_buffer *buffer;
> +	u32 packet_done = 0;
> +
> +	if (unlikely(!hbg_nic_is_open(priv))) {
> +		napi_complete(napi);
> +		return 0;
> +	}
> +
> +	while (packet_done < budget) {
> +		if (unlikely(hbg_queue_is_empty(ring->ntc, ring->ntu)))
> +			break;
> +
> +		buffer = &ring->queue[ring->ntc];
> +		if (unlikely(!buffer->skb))
> +			goto next_buffer;
> +
> +		if (unlikely(!hbg_sync_data_from_hw(priv, buffer)))
> +			break;
> +
> +		hbg_dma_unmap(buffer);
> +
> +		rx_desc = (struct hbg_rx_desc *)buffer->skb->data;
> +		skb_reserve(buffer->skb, HBG_PACKET_HEAD_SIZE + NET_IP_ALIGN);
> +		skb_put(buffer->skb, rx_desc->len);
> +		buffer->skb->protocol = eth_type_trans(buffer->skb, priv->netdev);
> +
> +		priv->netdev->stats.rx_bytes += rx_desc->len;
> +		priv->netdev->stats.rx_packets++;
> +		netif_receive_skb(buffer->skb);

Any reason why not napi_gro_receive ?

> +		buffer->skb = NULL;
> +		hbg_rx_fill_one_buffer(priv);
> +
> +next_buffer:
> +		hbg_queue_move_next(ntc, ring);
> +		packet_done++;
> +	}
> +
> +	hbg_rx_fill_buffers(priv);
> +	if (packet_done >= budget)
> +		return packet_done;
> +
> +	napi_complete(napi);

Maybe:

   if (napi_complete_done(napi))
     hbg_irq_enable(priv, HBG_IRQ_RX, true);

> +	hbg_irq_enable(priv, HBG_IRQ_RX, true);
> +
> +	return packet_done;
> +}

[...]