[PATCH net-next 0/3] r8169: support IPv6

[PATCH net-next 0/3] r8169: support IPv6

[Hayes Wang]

The RTL8168C and the later chips support the hardware checksum
for IPv6. Adjust some code and add the relative code.

Hayes Wang (3):
  r8169: split rtl8169_tso_csum
  r8169: use Giant Send
  r8169: support IPv6

 drivers/net/ethernet/realtek/r8169.c | 220 ++++++++++++++++++++++++++++-------
 1 file changed, 181 insertions(+), 39 deletions(-)

-- 
1.9.3

[PATCH net-next 1/3] r8169: split rtl8169_tso_csum

[Hayes Wang]

According to the txd_version, split rtl8169_tso_csum() into
rtl8169_tso_csum_v1() and rtl8169_tso_csum_v2().

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
---
 drivers/net/ethernet/realtek/r8169.c | 71 +++++++++++++++++++-----------------
 1 file changed, 37 insertions(+), 34 deletions(-)

diff --git a/drivers/net/ethernet/realtek/r8169.c b/drivers/net/ethernet/realtek/r8169.c
index be425ad..6b7e21a 100644
--- a/drivers/net/ethernet/realtek/r8169.c
+++ b/drivers/net/ethernet/realtek/r8169.c
@@ -633,32 +633,6 @@ enum rtl_tx_desc_bit_1 {
 	TD1_UDP_CS	= (1 << 31),		/* Calculate UDP/IP checksum */
 };
 
-static const struct rtl_tx_desc_info {
-	struct {
-		u32 udp;
-		u32 tcp;
-	} checksum;
-	u16 mss_shift;
-	u16 opts_offset;
-} tx_desc_info [] = {
-	[RTL_TD_0] = {
-		.checksum = {
-			.udp	= TD0_IP_CS | TD0_UDP_CS,
-			.tcp	= TD0_IP_CS | TD0_TCP_CS
-		},
-		.mss_shift	= TD0_MSS_SHIFT,
-		.opts_offset	= 0
-	},
-	[RTL_TD_1] = {
-		.checksum = {
-			.udp	= TD1_IP_CS | TD1_UDP_CS,
-			.tcp	= TD1_IP_CS | TD1_TCP_CS
-		},
-		.mss_shift	= TD1_MSS_SHIFT,
-		.opts_offset	= 1
-	}
-};
-
 enum rtl_rx_desc_bit {
 	/* Rx private */
 	PID1		= (1 << 18), /* Protocol ID bit 1/2 */
@@ -782,6 +756,7 @@ struct rtl8169_private {
 	unsigned int (*phy_reset_pending)(struct rtl8169_private *tp);
 	unsigned int (*link_ok)(void __iomem *);
 	int (*do_ioctl)(struct rtl8169_private *tp, struct mii_ioctl_data *data, int cmd);
+	bool (*tso_csum)(struct rtl8169_private *, struct sk_buff *, u32 *);
 
 	struct {
 		DECLARE_BITMAP(flags, RTL_FLAG_MAX);
@@ -5941,16 +5916,36 @@ static bool rtl_test_hw_pad_bug(struct rtl8169_private *tp, struct sk_buff *skb)
 	return skb->len < ETH_ZLEN && tp->mac_version == RTL_GIGA_MAC_VER_34;
 }
 
-static inline bool rtl8169_tso_csum(struct rtl8169_private *tp,
-				    struct sk_buff *skb, u32 *opts)
+static bool rtl8169_tso_csum_v1(struct rtl8169_private *tp,
+				struct sk_buff *skb, u32 *opts)
 {
-	const struct rtl_tx_desc_info *info = tx_desc_info + tp->txd_version;
 	u32 mss = skb_shinfo(skb)->gso_size;
-	int offset = info->opts_offset;
 
 	if (mss) {
 		opts[0] |= TD_LSO;
-		opts[offset] |= min(mss, TD_MSS_MAX) << info->mss_shift;
+		opts[0] |= min(mss, TD_MSS_MAX) << TD0_MSS_SHIFT;
+	} else if (skb->ip_summed == CHECKSUM_PARTIAL) {
+		const struct iphdr *ip = ip_hdr(skb);
+
+		if (ip->protocol == IPPROTO_TCP)
+			opts[0] |= TD0_IP_CS | TD0_TCP_CS;
+		else if (ip->protocol == IPPROTO_UDP)
+			opts[0] |= TD0_IP_CS | TD0_UDP_CS;
+		else
+			WARN_ON_ONCE(1);
+	}
+
+	return true;
+}
+
+static bool rtl8169_tso_csum_v2(struct rtl8169_private *tp,
+				struct sk_buff *skb, u32 *opts)
+{
+	u32 mss = skb_shinfo(skb)->gso_size;
+
+	if (mss) {
+		opts[0] |= TD_LSO;
+		opts[1] |= min(mss, TD_MSS_MAX) << TD1_MSS_SHIFT;
 	} else if (skb->ip_summed == CHECKSUM_PARTIAL) {
 		const struct iphdr *ip = ip_hdr(skb);
 
@@ -5958,15 +5953,16 @@ static inline bool rtl8169_tso_csum(struct rtl8169_private *tp,
 			return skb_checksum_help(skb) == 0 && rtl_skb_pad(skb);
 
 		if (ip->protocol == IPPROTO_TCP)
-			opts[offset] |= info->checksum.tcp;
+			opts[1] |= TD1_IP_CS | TD1_TCP_CS;
 		else if (ip->protocol == IPPROTO_UDP)
-			opts[offset] |= info->checksum.udp;
+			opts[1] |= TD1_IP_CS | TD1_UDP_CS;
 		else
 			WARN_ON_ONCE(1);
 	} else {
 		if (unlikely(rtl_test_hw_pad_bug(tp, skb)))
 			return rtl_skb_pad(skb);
 	}
+
 	return true;
 }
 
@@ -5994,7 +5990,7 @@ static netdev_tx_t rtl8169_start_xmit(struct sk_buff *skb,
 	opts[1] = cpu_to_le32(rtl8169_tx_vlan_tag(skb));
 	opts[0] = DescOwn;
 
-	if (!rtl8169_tso_csum(tp, skb, opts))
+	if (!tp->tso_csum(tp, skb, opts))
 		goto err_update_stats;
 
 	len = skb_headlen(skb);
@@ -7145,6 +7141,13 @@ rtl_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 		/* 8110SCd requires hardware Rx VLAN - disallow toggling */
 		dev->hw_features &= ~NETIF_F_HW_VLAN_CTAG_RX;
 
+	if (tp->txd_version == RTL_TD_0)
+		tp->tso_csum = rtl8169_tso_csum_v1;
+	else if (tp->txd_version == RTL_TD_1)
+		tp->tso_csum = rtl8169_tso_csum_v2;
+	else
+		WARN_ON_ONCE(1);
+
 	dev->hw_features |= NETIF_F_RXALL;
 	dev->hw_features |= NETIF_F_RXFCS;
 
-- 
1.9.3

[PATCH net-next 2/3] r8169: use Giant Send

[Hayes Wang]

Replace large send with giant send for TSO for RTL8111C and later ICs.
The large send setting of the RTL8111DP is different from the other
chips. However, the giant send setting is the same for all the chips
which support it. Use the giant send to synchronize the settings.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
---
 drivers/net/ethernet/realtek/r8169.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/realtek/r8169.c b/drivers/net/ethernet/realtek/r8169.c
index 6b7e21a..e5a0bdb 100644
--- a/drivers/net/ethernet/realtek/r8169.c
+++ b/drivers/net/ethernet/realtek/r8169.c
@@ -626,6 +626,10 @@ enum rtl_tx_desc_bit_0 {
 
 /* 8102e, 8168c and beyond. */
 enum rtl_tx_desc_bit_1 {
+	/* First doubleword. */
+	TD1_GTSENV4	= (1 << 26),		/* Giant Send for IPv4 */
+#define GTTCPHO_SHIFT			18
+
 	/* Second doubleword. */
 #define TD1_MSS_SHIFT			18	/* MSS position (11 bits) */
 	TD1_IP_CS	= (1 << 29),		/* Calculate IP checksum */
@@ -5941,10 +5945,12 @@ static bool rtl8169_tso_csum_v1(struct rtl8169_private *tp,
 static bool rtl8169_tso_csum_v2(struct rtl8169_private *tp,
 				struct sk_buff *skb, u32 *opts)
 {
+	u32 transport_offset = (u32)skb_transport_offset(skb);
 	u32 mss = skb_shinfo(skb)->gso_size;
 
 	if (mss) {
-		opts[0] |= TD_LSO;
+		opts[0] |= TD1_GTSENV4;
+		opts[0] |= transport_offset << GTTCPHO_SHIFT;
 		opts[1] |= min(mss, TD_MSS_MAX) << TD1_MSS_SHIFT;
 	} else if (skb->ip_summed == CHECKSUM_PARTIAL) {
 		const struct iphdr *ip = ip_hdr(skb);
-- 
1.9.3

[PATCH net-next 3/3] r8169: support IPv6

[Hayes Wang]

Support the IPv6 hw checksum for RTL8111C and later chips. Note
that the hw has the limitation for the transport offset. The
checksum must be calculated by sw, when the transport offset is
out of the range which the hw accepts.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
---
 drivers/net/ethernet/realtek/r8169.c | 157 ++++++++++++++++++++++++++++++++---
 1 file changed, 145 insertions(+), 12 deletions(-)

diff --git a/drivers/net/ethernet/realtek/r8169.c b/drivers/net/ethernet/realtek/r8169.c
index e5a0bdb..51c78ce 100644
--- a/drivers/net/ethernet/realtek/r8169.c
+++ b/drivers/net/ethernet/realtek/r8169.c
@@ -27,6 +27,8 @@
 #include <linux/firmware.h>
 #include <linux/pci-aspm.h>
 #include <linux/prefetch.h>
+#include <linux/ipv6.h>
+#include <net/ip6_checksum.h>
 
 #include <asm/io.h>
 #include <asm/irq.h>
@@ -628,11 +630,16 @@ enum rtl_tx_desc_bit_0 {
 enum rtl_tx_desc_bit_1 {
 	/* First doubleword. */
 	TD1_GTSENV4	= (1 << 26),		/* Giant Send for IPv4 */
+	TD1_GTSENV6	= (1 << 25),		/* Giant Send for IPv6 */
 #define GTTCPHO_SHIFT			18
+#define GTTCPHO_MAX			0x7fU
 
 	/* Second doubleword. */
+#define TCPHO_SHIFT			18
+#define TCPHO_MAX			0x3ffU
 #define TD1_MSS_SHIFT			18	/* MSS position (11 bits) */
-	TD1_IP_CS	= (1 << 29),		/* Calculate IP checksum */
+	TD1_IPv6_CS	= (1 << 28),		/* Calculate IPv6 checksum */
+	TD1_IPv4_CS	= (1 << 29),		/* Calculate IPv4 checksum */
 	TD1_TCP_CS	= (1 << 30),		/* Calculate TCP/IP checksum */
 	TD1_UDP_CS	= (1 << 31),		/* Calculate UDP/IP checksum */
 };
@@ -5920,6 +5927,82 @@ static bool rtl_test_hw_pad_bug(struct rtl8169_private *tp, struct sk_buff *skb)
 	return skb->len < ETH_ZLEN && tp->mac_version == RTL_GIGA_MAC_VER_34;
 }
 
+static netdev_tx_t rtl8169_start_xmit(struct sk_buff *skb,
+				      struct net_device *dev);
+/* r8169_csum_workaround()
+ * The hw limites the value the transport offset. When the offset is out of the
+ * range, calculate the checksum by sw.
+ */
+static void r8169_csum_workaround(struct rtl8169_private *tp,
+				  struct sk_buff *skb)
+{
+	if (skb_shinfo(skb)->gso_size) {
+		netdev_features_t features = tp->dev->features;
+		struct sk_buff *segs, *nskb;
+
+		features &= ~(NETIF_F_SG | NETIF_F_IPV6_CSUM | NETIF_F_TSO6);
+		segs = skb_gso_segment(skb, features);
+		if (IS_ERR(segs) || !segs)
+			goto drop;
+
+		do {
+			nskb = segs;
+			segs = segs->next;
+			nskb->next = NULL;
+			rtl8169_start_xmit(nskb, tp->dev);
+		} while (segs);
+
+		dev_kfree_skb(skb);
+	} else if (skb->ip_summed == CHECKSUM_PARTIAL) {
+		if (skb_checksum_help(skb) < 0)
+			goto drop;
+
+		rtl8169_start_xmit(skb, tp->dev);
+	} else {
+		struct net_device_stats *stats;
+
+drop:
+		stats = &tp->dev->stats;
+		stats->tx_dropped++;
+		dev_kfree_skb(skb);
+	}
+}
+
+/* msdn_giant_send_check()
+ * According to the document of microsoft, the TCP Pseudo Header excludes the
+ * packet length for IPv6 TCP large packets.
+ */
+static int msdn_giant_send_check(struct sk_buff *skb)
+{
+	const struct ipv6hdr *ipv6h;
+	struct tcphdr *th;
+	int ret;
+
+	ret = skb_cow_head(skb, 0);
+	if (ret)
+		return ret;
+
+	ipv6h = ipv6_hdr(skb);
+	th = tcp_hdr(skb);
+
+	th->check = 0;
+	th->check = ~tcp_v6_check(0, &ipv6h->saddr, &ipv6h->daddr, 0);
+
+	return ret;
+}
+
+static inline __be16 get_protocol(struct sk_buff *skb)
+{
+	__be16 protocol;
+
+	if (skb->protocol == htons(ETH_P_8021Q))
+		protocol = vlan_eth_hdr(skb)->h_vlan_encapsulated_proto;
+	else
+		protocol = skb->protocol;
+
+	return protocol;
+}
+
 static bool rtl8169_tso_csum_v1(struct rtl8169_private *tp,
 				struct sk_buff *skb, u32 *opts)
 {
@@ -5949,21 +6032,69 @@ static bool rtl8169_tso_csum_v2(struct rtl8169_private *tp,
 	u32 mss = skb_shinfo(skb)->gso_size;
 
 	if (mss) {
-		opts[0] |= TD1_GTSENV4;
+		if (transport_offset > GTTCPHO_MAX) {
+			netif_warn(tp, tx_err, tp->dev,
+				   "Invalid transport offset 0x%x for TSO\n",
+				   transport_offset);
+			return false;
+		}
+
+		switch (get_protocol(skb)) {
+		case htons(ETH_P_IP):
+			opts[0] |= TD1_GTSENV4;
+			break;
+
+		case htons(ETH_P_IPV6):
+			if (msdn_giant_send_check(skb))
+				return false;
+
+			opts[0] |= TD1_GTSENV6;
+			break;
+
+		default:
+			WARN_ON_ONCE(1);
+			break;
+		}
+
 		opts[0] |= transport_offset << GTTCPHO_SHIFT;
 		opts[1] |= min(mss, TD_MSS_MAX) << TD1_MSS_SHIFT;
 	} else if (skb->ip_summed == CHECKSUM_PARTIAL) {
-		const struct iphdr *ip = ip_hdr(skb);
+		u8 ip_protocol;
 
 		if (unlikely(rtl_test_hw_pad_bug(tp, skb)))
 			return skb_checksum_help(skb) == 0 && rtl_skb_pad(skb);
 
-		if (ip->protocol == IPPROTO_TCP)
-			opts[1] |= TD1_IP_CS | TD1_TCP_CS;
-		else if (ip->protocol == IPPROTO_UDP)
-			opts[1] |= TD1_IP_CS | TD1_UDP_CS;
+		if (transport_offset > TCPHO_MAX) {
+			netif_warn(tp, tx_err, tp->dev,
+				   "Invalid transport offset 0x%x\n",
+				   transport_offset);
+			return false;
+		}
+
+		switch (get_protocol(skb)) {
+		case htons(ETH_P_IP):
+			opts[1] |= TD1_IPv4_CS;
+			ip_protocol = ip_hdr(skb)->protocol;
+			break;
+
+		case htons(ETH_P_IPV6):
+			opts[1] |= TD1_IPv6_CS;
+			ip_protocol = ipv6_hdr(skb)->nexthdr;
+			break;
+
+		default:
+			ip_protocol = IPPROTO_RAW;
+			break;
+		}
+
+		if (ip_protocol == IPPROTO_TCP)
+			opts[1] |= TD1_TCP_CS;
+		else if (ip_protocol == IPPROTO_UDP)
+			opts[1] |= TD1_UDP_CS;
 		else
 			WARN_ON_ONCE(1);
+
+		opts[1] |= transport_offset << TCPHO_SHIFT;
 	} else {
 		if (unlikely(rtl_test_hw_pad_bug(tp, skb)))
 			return rtl_skb_pad(skb);
@@ -5996,8 +6127,10 @@ static netdev_tx_t rtl8169_start_xmit(struct sk_buff *skb,
 	opts[1] = cpu_to_le32(rtl8169_tx_vlan_tag(skb));
 	opts[0] = DescOwn;
 
-	if (!tp->tso_csum(tp, skb, opts))
-		goto err_update_stats;
+	if (!tp->tso_csum(tp, skb, opts)) {
+		r8169_csum_workaround(tp, skb);
+		return NETDEV_TX_OK;
+	}
 
 	len = skb_headlen(skb);
 	mapping = dma_map_single(d, skb->data, len, DMA_TO_DEVICE);
@@ -6062,7 +6195,6 @@ err_dma_1:
 	rtl8169_unmap_tx_skb(d, tp->tx_skb + entry, txd);
 err_dma_0:
 	dev_kfree_skb_any(skb);
-err_update_stats:
 	dev->stats.tx_dropped++;
 	return NETDEV_TX_OK;
 
@@ -7149,9 +7281,10 @@ rtl_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 
 	if (tp->txd_version == RTL_TD_0)
 		tp->tso_csum = rtl8169_tso_csum_v1;
-	else if (tp->txd_version == RTL_TD_1)
+	else if (tp->txd_version == RTL_TD_1) {
 		tp->tso_csum = rtl8169_tso_csum_v2;
-	else
+		dev->hw_features |= NETIF_F_IPV6_CSUM | NETIF_F_TSO6;
+	} else
 		WARN_ON_ONCE(1);
 
 	dev->hw_features |= NETIF_F_RXALL;
-- 
1.9.3

Re: [PATCH net-next 3/3] r8169: support IPv6

[David Woodhouse]

> Support the IPv6 hw checksum for RTL8111C and later chips. Note
> that the hw has the limitation for the transport offset. The
> checksum must be calculated by sw, when the transport offset is
> out of the range which the hw accepts.


It would be better to implement this check in a .ndo_features_check
method, just clearing the appropriate CSUM/TSO features for the skbs for
which the hardware cannot cope. That way the stack fixes them up for you.


> +	if (skb_shinfo(skb)->gso_size) {
> +		netdev_features_t features = tp->dev->features;
> +		struct sk_buff *segs, *nskb;
> +
> +		features &= ~(NETIF_F_SG | NETIF_F_IPV6_CSUM | NETIF_F_TSO6);
> +		segs = skb_gso_segment(skb, features);
> +		if (IS_ERR(segs) || !segs)
> +			goto drop;
> +
> +		do {
> +			nskb = segs;
> +			segs = segs->next;
> +			nskb->next = NULL;
> +			rtl8169_start_xmit(nskb, tp->dev);
> +		} while (segs);
> +
> +		dev_kfree_skb(skb);

This loop in particular makes no attempt to avoid exceeding the available
space in your descriptor ring, and can drop packets and trigger the
warning at the start of your hard_start_xmit function.

-- 
dwmw2

Re: [PATCH net-next 0/3] r8169: support IPv6

[David Miller]

From: Hayes Wang <hayeswang@realtek.com>
Date: Fri, 11 Jul 2014 16:25:55 +0800

> The RTL8168C and the later chips support the hardware checksum
> for IPv6. Adjust some code and add the relative code.

Series applied, thanks.