Message-ID: <98ae4f14-397b-49b7-a0a9-cb316f2594f6@bp.renesas.com>
Date: Mon, 15 Apr 2024 08:34:09 +0100
From: Paul Barker <paul.barker.ct@...renesas.com>
To: Niklas Söderlund
 <niklas.soderlund+renesas@...natech.se>,
 Yoshihiro Shimoda <yoshihiro.shimoda.uh@...esas.com>,
 "David S. Miller" <davem@...emloft.net>, Eric Dumazet <edumazet@...gle.com>,
 Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
 netdev@...r.kernel.org
Cc: linux-renesas-soc@...r.kernel.org
Subject: Re: [net-next] net: ethernet: rtsn: Add support for Renesas
 Ethernet-TSN

On 14/04/2024 14:59, Niklas Söderlund wrote:
> Add initial support for Renesas Ethernet-TSN End-station device of R-Car
> V4H. The Ethernet End-station can connect to an Ethernet network using a
> 10 Mbps, 100 Mbps, or 1 Gbps full-duplex link via MII/GMII/RMII/RGMII.
> Depending on the connected PHY.
> 
> Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@...natech.se>
> ---
> * Changes since RFC
> - Fix issues in MDIO communication.
> - Use a dedicated OF node for the MDIO bus.
> ---
>  MAINTAINERS                           |    8 +
>  drivers/net/ethernet/renesas/Kconfig  |   11 +
>  drivers/net/ethernet/renesas/Makefile |    2 +
>  drivers/net/ethernet/renesas/rtsn.c   | 1421 +++++++++++++++++++++++++
>  drivers/net/ethernet/renesas/rtsn.h   |  464 ++++++++
>  5 files changed, 1906 insertions(+)
>  create mode 100644 drivers/net/ethernet/renesas/rtsn.c
>  create mode 100644 drivers/net/ethernet/renesas/rtsn.h

<snip>

> diff --git a/drivers/net/ethernet/renesas/rtsn.c b/drivers/net/ethernet/renesas/rtsn.c
> new file mode 100644
> index 000000000000..291ab421d68f
> --- /dev/null
> +++ b/drivers/net/ethernet/renesas/rtsn.c

<snip>

> +static bool rtsn_rx(struct net_device *ndev, int *quota)
> +{
> +	struct rtsn_ext_ts_desc *desc;
> +	struct rtsn_private *priv;
> +	struct sk_buff *skb;
> +	dma_addr_t dma_addr;
> +	int boguscnt;

I find the variable name `boguscnt` very unclear, I'm not sure if it
means the count is bogus, or it is counting bogus items?

I don't think you need to match what I've done in ravb_main.c exactly,
but I'd prefer to see a better variable name here.

> +	u16 pkt_len;
> +	u32 get_ts;
> +	int entry;
> +	int limit;
> +
> +	priv = netdev_priv(ndev);
> +
> +	entry = priv->cur_rx % priv->num_rx_ring;
> +	desc = &priv->rx_ring[entry];
> +
> +	boguscnt = priv->dirty_rx + priv->num_rx_ring - priv->cur_rx;
> +	boguscnt = min(boguscnt, *quota);
> +	limit = boguscnt;
> +
> +	while ((desc->die_dt & DT_MASK) != DT_FEMPTY) {
> +		dma_rmb();
> +		pkt_len = le16_to_cpu(desc->info_ds) & RX_DS;
> +		if (--boguscnt < 0)
> +			break;
> +
> +		skb = priv->rx_skb[entry];
> +		priv->rx_skb[entry] = NULL;
> +		dma_addr = le32_to_cpu(desc->dptr);
> +		dma_unmap_single(ndev->dev.parent, dma_addr, PKT_BUF_SZ,
> +				 DMA_FROM_DEVICE);
> +
> +		get_ts = priv->ptp_priv->tstamp_rx_ctrl &
> +			RCAR_GEN4_RXTSTAMP_TYPE_V2_L2_EVENT;
> +		if (get_ts) {
> +			struct skb_shared_hwtstamps *shhwtstamps;
> +			struct timespec64 ts;
> +
> +			shhwtstamps = skb_hwtstamps(skb);
> +			memset(shhwtstamps, 0, sizeof(*shhwtstamps));
> +
> +			ts.tv_sec = (u64)le32_to_cpu(desc->ts_sec);
> +			ts.tv_nsec = le32_to_cpu(desc->ts_nsec & cpu_to_le32(0x3fffffff));
> +
> +			shhwtstamps->hwtstamp = timespec64_to_ktime(ts);
> +		}
> +
> +		skb_put(skb, pkt_len);
> +		skb->protocol = eth_type_trans(skb, ndev);
> +		netif_receive_skb(skb);
> +		ndev->stats.rx_packets++;
> +		ndev->stats.rx_bytes += pkt_len;
> +
> +		entry = (++priv->cur_rx) % priv->num_rx_ring;
> +		desc = &priv->rx_ring[entry];
> +	}
> +
> +	/* Refill the RX ring buffers */
> +	for (; priv->cur_rx - priv->dirty_rx > 0; priv->dirty_rx++) {
> +		entry = priv->dirty_rx % priv->num_rx_ring;
> +		desc = &priv->rx_ring[entry];
> +		desc->info_ds = cpu_to_le16(PKT_BUF_SZ);
> +
> +		if (!priv->rx_skb[entry]) {
> +			skb = netdev_alloc_skb(ndev,
> +					       PKT_BUF_SZ + RTSN_ALIGN - 1);

I'll send my work using a page pool today as an RFC so you can see if it
would be beneficial to use that here as well. I was going to hold off
until the bugfix patches have merged so that I don't need to go through
another RFC round, but it will be good to get some more review on the
series anyway.

> +			if (!skb)
> +				break;
> +			skb_reserve(skb, NET_IP_ALIGN);
> +			dma_addr = dma_map_single(ndev->dev.parent, skb->data,
> +						  le16_to_cpu(desc->info_ds),
> +						  DMA_FROM_DEVICE);
> +			if (dma_mapping_error(ndev->dev.parent, dma_addr))
> +				desc->info_ds = cpu_to_le16(0);
> +			desc->dptr = cpu_to_le32(dma_addr);
> +			skb_checksum_none_assert(skb);
> +			priv->rx_skb[entry] = skb;
> +		}
> +		dma_wmb();
> +		desc->die_dt = DT_FEMPTY | D_DIE;
> +	}
> +
> +	desc = &priv->rx_ring[priv->num_rx_ring];
> +	desc->die_dt = DT_LINK;
> +
> +	*quota -= limit - (++boguscnt);
> +
> +	return boguscnt <= 0;
> +}
> +
> +static int rtsn_poll(struct napi_struct *napi, int budget)
> +{
> +	struct rtsn_private *priv;
> +	struct net_device *ndev;
> +	unsigned long flags;
> +	int quota = budget;
> +
> +	ndev = napi->dev;
> +	priv = netdev_priv(ndev);
> +
> +	/* Processing RX Descriptor Ring */
> +	if (rtsn_rx(ndev, &quota))
> +		goto out;
> +
> +	/* Processing TX Descriptor Ring */
> +	spin_lock_irqsave(&priv->lock, flags);
> +	rtsn_tx_free(ndev, true);
> +	netif_wake_subqueue(ndev, 0);
> +	spin_unlock_irqrestore(&priv->lock, flags);
> +
> +	napi_complete(napi);

We should use napi_complete_done() here as described in
Documentation/networking/napi.rst. That will require rtsn_rx() to return
the number of packets received so that it can be passed as the work_done
argument to napi_complete_done().

> +
> +	/* Re-enable TX/RX interrupts */
> +	spin_lock_irqsave(&priv->lock, flags);
> +	rtsn_ctrl_data_irq(priv, true);
> +	__iowmb();
> +	spin_unlock_irqrestore(&priv->lock, flags);
> +out:
> +	return budget - quota;
> +}

<snip>

> +static int rtsn_probe(struct platform_device *pdev)
> +{
> +	struct rtsn_private *priv;
> +	struct net_device *ndev;
> +	struct resource *res;
> +	int ret;
> +
> +	ndev = alloc_etherdev_mqs(sizeof(struct rtsn_private), TX_NUM_CHAINS,
> +				  RX_NUM_CHAINS);
> +	if (!ndev)
> +		return -ENOMEM;
> +
> +	priv = netdev_priv(ndev);
> +	priv->pdev = pdev;
> +	priv->ndev = ndev;
> +	priv->ptp_priv = rcar_gen4_ptp_alloc(pdev);
> +
> +	spin_lock_init(&priv->lock);
> +	platform_set_drvdata(pdev, priv);
> +
> +	priv->clk = devm_clk_get(&pdev->dev, NULL);
> +	if (IS_ERR(priv->clk)) {
> +		ret = -PTR_ERR(priv->clk);
> +		goto error_alloc;
> +	}
> +
> +	priv->reset = devm_reset_control_get(&pdev->dev, NULL);
> +	if (IS_ERR(priv->reset)) {
> +		ret = -PTR_ERR(priv->reset);
> +		goto error_alloc;
> +	}
> +
> +	res = platform_get_resource_byname(pdev, IORESOURCE_MEM, "tsnes");
> +	if (!res) {
> +		dev_err(&pdev->dev, "Can't find tsnes resource\n");
> +		ret = -EINVAL;
> +		goto error_alloc;
> +	}
> +
> +	priv->base = devm_ioremap_resource(&pdev->dev, res);
> +	if (IS_ERR(priv->base)) {
> +		ret = PTR_ERR(priv->base);
> +		goto error_alloc;
> +	}
> +
> +	SET_NETDEV_DEV(ndev, &pdev->dev);
> +	ether_setup(ndev);
> +
> +	ndev->features = NETIF_F_RXCSUM;
> +	ndev->hw_features = NETIF_F_RXCSUM;

A quick skim of the datasheet suggests that TX checksum calculation is
also supported. It's probably worth listing which hardware features this
driver supports/does not support in the commit message.

Thanks,

-- 
Paul Barker
Download attachment "OpenPGP_0x27F4B3459F002257.asc" of type "application/pgp-keys" (3521 bytes)

Download attachment "OpenPGP_signature.asc" of type "application/pgp-signature" (237 bytes)

Powered by blists - more mailing lists