WIP: net: bcmgenet: experimental 9000 byte MTU without status blocks

RBUF_PKT_RDY_THLD is where the status block gets written, not a limit on
the frame. With the status blocks off jumbo frames come through and the
threshold splits nothing.

Costs both checksum offloads and all but one transmit queue.

Signed-off-by: Nicolai Buchwitz <nb@tipi-net.de>

Author: nbuchwitz

drivers/net/ethernet/broadcom/genet/bcmgenet.c

@@ -10,6 +10,7 @@
 #include <linux/acpi.h>
 #include <linux/kernel.h>
 #include <linux/module.h>
+#include <linux/of.h>
 #include <linux/sched.h>
 #include <linux/types.h>
 #include <linux/fcntl.h>
@@ -51,6 +52,12 @@
 
 #define SKB_ALIGNMENT		32
 
+/* Test only, see the commit message */
+static bool experimental_jumbo;
+module_param(experimental_jumbo, bool, 0444);
+MODULE_PARM_DESC(experimental_jumbo,
+		 "9000 byte MTU, costs the status blocks and checksum offload");
+
 /* Tx/Rx DMA register offset, skip 256 descriptors */
 #define WORDS_PER_BD(p)		(p->hw_params->words_per_bd)
 #define DMA_DESC_SIZE		(WORDS_PER_BD(priv) * sizeof(u32))
@@ -2150,11 +2157,13 @@ static netdev_tx_t bcmgenet_xmit(struct sk_buff *skb, struct net_device *dev)
 	 */
 	GENET_CB(skb)->bytes_sent = skb->len;
 
-	/* add the Transmit Status Block */
-	skb = bcmgenet_add_tsb(dev, skb, ring);
-	if (!skb) {
-		ret = NETDEV_TX_OK;
-		goto out;
+	/* No TSB, probe cleared the checksum offloads */
+	if (!experimental_jumbo) {
+		skb = bcmgenet_add_tsb(dev, skb, ring);
+		if (!skb) {
+			ret = NETDEV_TX_OK;
+			goto out;
+		}
 	}
 
 	for (i = 0; i <= nr_frags; i++) {
@@ -2209,7 +2218,8 @@ static netdev_tx_t bcmgenet_xmit(struct sk_buff *skb, struct net_device *dev)
 
 	GENET_CB(skb)->last_cb = tx_cb_ptr;
 
-	bcmgenet_hide_tsb(skb);
+	if (!experimental_jumbo)
+		bcmgenet_hide_tsb(skb);
 	skb_tx_timestamp(skb);
 
 	/* Decrement total BD count and advance our write pointer */
@@ -2339,20 +2349,25 @@ static unsigned int bcmgenet_desc_rx(struct bcmgenet_rx_ring *ring,
 		__be16 rx_csum;
 
 		cb = &priv->rx_cbs[ring->read_ptr];
+		if (experimental_jumbo)
+			dma_length_status = bcmgenet_readl(cb->bd_addr +
+						 DMA_DESC_LENGTH_STATUS);
 		skb = bcmgenet_rx_refill(priv, cb);
 
 		if (unlikely(!skb)) {
 			BCMGENET_STATS64_INC(stats, dropped);
 			goto next;
 		}
 
-		status = (struct status_64 *)skb->data;
-		dma_length_status = status->length_status;
-		if (dev->features & NETIF_F_RXCSUM) {
-			rx_csum = (__force __be16)(status->rx_csum & 0xffff);
-			if (rx_csum) {
-				skb->csum = (__force __wsum)ntohs(rx_csum);
-				skb->ip_summed = CHECKSUM_COMPLETE;
+		if (!experimental_jumbo) {
+			status = (struct status_64 *)skb->data;
+			dma_length_status = status->length_status;
+			if (dev->features & NETIF_F_RXCSUM) {
+				rx_csum = (__force __be16)(status->rx_csum & 0xffff);
+				if (rx_csum) {
+					skb->csum = (__force __wsum)ntohs(rx_csum);
+					skb->ip_summed = CHECKSUM_COMPLETE;
+				}
 			}
 		}
 
@@ -2367,7 +2382,10 @@ static unsigned int bcmgenet_desc_rx(struct bcmgenet_rx_ring *ring,
 			  __func__, p_index, ring->c_index,
 			  ring->read_ptr, dma_length_status);
 
-		if (unlikely(len > priv->rx_buf_len)) {
+		if (unlikely(len > priv->rx_buf_len ||
+			     len < (experimental_jumbo ? ENET_RBUF_ALIGN :
+				    ENET_RX_OFFSET) + ETH_HLEN +
+				   (priv->crc_fwd_en ? ETH_FCS_LEN : 0))) {
 			netif_err(priv, rx_status, dev, "oversized packet\n");
 			BCMGENET_STATS64_INC(stats, length_errors);
 			dev_kfree_skb_any(skb);
@@ -2412,9 +2430,14 @@ static unsigned int bcmgenet_desc_rx(struct bcmgenet_rx_ring *ring,
 
 		skb_put(skb, len);
 
-		/* remove RSB and hardware 2bytes added for IP alignment */
-		skb_pull(skb, ENET_RX_OFFSET);
-		len -= ENET_RX_OFFSET;
+		/* Remove the alignment bytes, and the RSB if there is one */
+		if (experimental_jumbo) {
+			skb_pull(skb, ENET_RBUF_ALIGN);
+			len -= ENET_RBUF_ALIGN;
+		} else {
+			skb_pull(skb, ENET_RX_OFFSET);
+			len -= ENET_RX_OFFSET;
+		}
 
 		if (priv->crc_fwd_en) {
 			skb_trim(skb, len - ETH_FCS_LEN);
@@ -2612,6 +2635,11 @@ static unsigned int bcmgenet_pkt_rdy_thld(unsigned int mtu)
 /* A buffer has to hold everything the threshold lets the hardware deliver */
 static unsigned int bcmgenet_rx_buf_len(unsigned int mtu)
 {
+	if (experimental_jumbo)
+		return max_t(unsigned int, ENET_THLD_DEFAULT * ENET_THLD_UNIT,
+			     round_up(ENET_MAX_FRAME_LEN(mtu) + ENET_RBUF_ALIGN,
+				      ENET_THLD_BURST));
+
 	return ENET_RSB_LEN + bcmgenet_pkt_rdy_thld(mtu) * ENET_THLD_UNIT;
 }
 
@@ -2652,17 +2680,28 @@ static void init_umac(struct bcmgenet_priv *priv)
 
 	/* init tx registers, enable TSB */
 	reg = bcmgenet_tbuf_ctrl_get(priv);
-	reg |= TBUF_64B_EN;
+	if (experimental_jumbo)
+		reg &= ~TBUF_64B_EN;
+	else
+		reg |= TBUF_64B_EN;
 	bcmgenet_tbuf_ctrl_set(priv, reg);
 
 	/* init rx registers, enable ip header optimization and RSB */
 	reg = bcmgenet_rbuf_readl(priv, RBUF_CTRL);
-	reg |= RBUF_ALIGN_2B | RBUF_64B_EN;
+	reg |= RBUF_ALIGN_2B;
+	if (experimental_jumbo)
+		reg &= ~RBUF_64B_EN;
+	else
+		reg |= RBUF_64B_EN;
 	bcmgenet_rbuf_writel(priv, reg, RBUF_CTRL);
 
 	/* enable rx checksumming */
 	reg = bcmgenet_rbuf_readl(priv, RBUF_CHK_CTRL);
-	reg |= RBUF_RXCHK_EN | RBUF_L3_PARSE_DIS;
+	reg |= RBUF_L3_PARSE_DIS;
+	if (experimental_jumbo)
+		reg &= ~RBUF_RXCHK_EN;
+	else
+		reg |= RBUF_RXCHK_EN;
 	/* If UniMAC forwards CRC, we need to skip over it to get
 	 * a valid CHK bit to be set in the per-packet status word
 	 */
@@ -4132,6 +4171,18 @@ static int bcmgenet_probe(struct platform_device *pdev)
 
 	bcmgenet_set_hw_params(priv);
 
+	if (experimental_jumbo) {
+		if (!of_device_is_compatible(pdev->dev.of_node,
+					     "brcm,bcm2711-genet-v5")) {
+			err = -EOPNOTSUPP;
+			goto err_clk_disable;
+		}
+		dev->features &= ~(NETIF_F_CSUM_MASK | NETIF_F_RXCSUM);
+		dev->hw_features &= ~(NETIF_F_CSUM_MASK | NETIF_F_RXCSUM);
+		dev->vlan_features &= ~(NETIF_F_CSUM_MASK | NETIF_F_RXCSUM);
+		dev_warn(&pdev->dev, "experimental jumbo mode enabled\n");
+	}
+
 	err = -EIO;
 	if (bcmgenet_has_40bits(priv))
 		err = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(40));
@@ -4147,7 +4198,7 @@ static int bcmgenet_probe(struct platform_device *pdev)
 	/* v1 cannot program the thresholds, so it stays at the default MTU */
 	priv->rx_buf_len = bcmgenet_rx_buf_len(dev->mtu);
 	if (!GENET_IS_V1(priv))
-		dev->max_mtu = ENET_MAX_MTU;
+		dev->max_mtu = experimental_jumbo ? 9000 : ENET_MAX_MTU;
 	INIT_WORK(&priv->bcmgenet_irq_work, bcmgenet_irq_task);
 
 	priv->clk_wol = devm_clk_get_optional(&priv->pdev->dev, "enet-wol");
@@ -4193,7 +4244,8 @@ static int bcmgenet_probe(struct platform_device *pdev)
 		goto err_clk_disable;
 
 	/* setup number of real queues + 1 */
-	netif_set_real_num_tx_queues(priv->dev, priv->hw_params->tx_queues + 1);
+	netif_set_real_num_tx_queues(priv->dev, experimental_jumbo ? 1 :
+				   priv->hw_params->tx_queues + 1);
 	netif_set_real_num_rx_queues(priv->dev, priv->hw_params->rx_queues + 1);
 
 	/* Set default coalescing parameters */