[net-next PATCH v2 00/10] Extend GbEth checksum offload support to VLAN/IPv6 packets

[net-next PATCH v2 00/10] Extend GbEth checksum offload support to VLAN/IPv6 packets

[Paul Barker]

The GbEth IP found in Renesas RZ/G2L, RZ/G3S and related SoCs supports
hardware checksum offload for packets in the following cases:

 - there are zero or one VLAN headers with TPID=0x8100
 - the network protocol is IPv4 or IPv6
 - the transport protocol is TCP, UDP or ICMP
 - the packet is not fragmented 

To complete the support for all these cases in the ravb driver, we need
to add handling for VLAN-tagged packets and IPv6 packets in both the TX
and RX code paths. Handling for ICMP packets in the TX path is not added
as it cannot currently be tested.

These patches also do some refactoring/tidy-up, drop unnecessary checks
from performance sensitive code paths and disable unnecessary IP header
checksum offloading.

Changes v1->v2:
  - Use get_unaligned() to access hw_csum in ravb_rx_csum_gbeth().
  - Dropped patch "Support ICMP TX checksum offload for GbEth" and
    simplified TX checksum offload path.
  - Moved last_frag into if condition in ravb_can_tx_csum_gbeth().
  - Dropped unnecessary check for skb->vlan_proto in
    ravb_can_tx_csum_gbeth().
  - s/HW/hardware/ in "Simplify types in RX csum validation" commit
    message.
  - s/null/zero/ in "Simplify UDP TX checksum offload" commit message.
  - Specified TX/RX direction in "Enable IPv6 TX/RX checksum offloading
    for GbEth" commit messages.
  - Dropped unnecessary whitespace.
  - Added Sergey's Reviewed-by tags.

Paul Barker (10):
  net: ravb: Factor out checksum offload enable bits
  net: ravb: Disable IP header RX checksum offloading
  net: ravb: Drop IP protocol check from RX csum verification
  net: ravb: Combine if conditions in RX csum validation
  net: ravb: Simplify types in RX csum validation
  net: ravb: Disable IP header TX checksum offloading
  net: ravb: Simplify UDP TX checksum offload
  net: ravb: Enable IPv6 RX checksum offloading for GbEth
  net: ravb: Enable IPv6 TX checksum offload for GbEth
  net: ravb: Add VLAN checksum support

 drivers/net/ethernet/renesas/ravb.h      |   6 ++
 drivers/net/ethernet/renesas/ravb_main.c | 101 ++++++++++++-----------
 2 files changed, 61 insertions(+), 46 deletions(-)

-- 
2.43.0

[net-next PATCH v2 01/10] net: ravb: Factor out checksum offload enable bits

[Paul Barker]

Introduce new constants for the CSR1 (TX) and CSR2 (RX) checksum enable
bits, removing the risk of inconsistency when we change which flags we
enable.

Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
---
 drivers/net/ethernet/renesas/ravb.h      | 4 ++++
 drivers/net/ethernet/renesas/ravb_main.c | 9 ++++-----
 2 files changed, 8 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/renesas/ravb.h b/drivers/net/ethernet/renesas/ravb.h
index a7de5cf6b317..4e1e0a754cd9 100644
--- a/drivers/net/ethernet/renesas/ravb.h
+++ b/drivers/net/ethernet/renesas/ravb.h
@@ -998,6 +998,8 @@ enum CSR1_BIT {
 	CSR1_TDHD	= 0x08000000,
 };
 
+#define CSR1_CSUM_ENABLE (CSR1_TIP4 | CSR1_TTCP4 | CSR1_TUDP4)
+
 enum CSR2_BIT {
 	CSR2_RIP4	= 0x00000001,
 	CSR2_RTCP4	= 0x00000010,
@@ -1012,6 +1014,8 @@ enum CSR2_BIT {
 	CSR2_RDHD	= 0x08000000,
 };
 
+#define CSR2_CSUM_ENABLE (CSR2_RIP4 | CSR2_RTCP4 | CSR2_RUDP4 | CSR2_RICMP4)
+
 #define DBAT_ENTRY_NUM	22
 #define RX_QUEUE_OFFSET	4
 #define NUM_RX_QUEUE	2
diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
index 41f88f8836f8..c8988c0c85a1 100644
--- a/drivers/net/ethernet/renesas/ravb_main.c
+++ b/drivers/net/ethernet/renesas/ravb_main.c
@@ -504,11 +504,10 @@ static void ravb_csum_init_gbeth(struct net_device *ndev)
 			ndev->features &= ~NETIF_F_RXCSUM;
 	} else {
 		if (tx_enable)
-			ravb_write(ndev, CSR1_TIP4 | CSR1_TTCP4 | CSR1_TUDP4, CSR1);
+			ravb_write(ndev, CSR1_CSUM_ENABLE, CSR1);
 
 		if (rx_enable)
-			ravb_write(ndev, CSR2_RIP4 | CSR2_RTCP4 | CSR2_RUDP4 | CSR2_RICMP4,
-				   CSR2);
+			ravb_write(ndev, CSR2_CSUM_ENABLE, CSR2);
 	}
 
 done:
@@ -2531,7 +2530,7 @@ static int ravb_set_features_gbeth(struct net_device *ndev,
 	spin_lock_irqsave(&priv->lock, flags);
 	if (changed & NETIF_F_RXCSUM) {
 		if (features & NETIF_F_RXCSUM)
-			val = CSR2_RIP4 | CSR2_RTCP4 | CSR2_RUDP4 | CSR2_RICMP4;
+			val = CSR2_CSUM_ENABLE;
 		else
 			val = 0;
 
@@ -2542,7 +2541,7 @@ static int ravb_set_features_gbeth(struct net_device *ndev,
 
 	if (changed & NETIF_F_HW_CSUM) {
 		if (features & NETIF_F_HW_CSUM)
-			val = CSR1_TIP4 | CSR1_TTCP4 | CSR1_TUDP4;
+			val = CSR1_CSUM_ENABLE;
 		else
 			val = 0;
 
-- 
2.43.0

[net-next PATCH v2 02/10] net: ravb: Disable IP header RX checksum offloading

[Paul Barker]

For IPv4 packets, the header checksum will always be checked in software
in the RX path (inet_gro_receive() calls ip_fast_csum() unconditionally)
so there is no advantage in asking the hardware to also calculate this
checksum.

Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
---
 drivers/net/ethernet/renesas/ravb.h      |  2 +-
 drivers/net/ethernet/renesas/ravb_main.c | 16 +++++++++-------
 2 files changed, 10 insertions(+), 8 deletions(-)

diff --git a/drivers/net/ethernet/renesas/ravb.h b/drivers/net/ethernet/renesas/ravb.h
index 4e1e0a754cd9..98496aa39f3d 100644
--- a/drivers/net/ethernet/renesas/ravb.h
+++ b/drivers/net/ethernet/renesas/ravb.h
@@ -1014,7 +1014,7 @@ enum CSR2_BIT {
 	CSR2_RDHD	= 0x08000000,
 };
 
-#define CSR2_CSUM_ENABLE (CSR2_RIP4 | CSR2_RTCP4 | CSR2_RUDP4 | CSR2_RICMP4)
+#define CSR2_CSUM_ENABLE (CSR2_RTCP4 | CSR2_RUDP4 | CSR2_RICMP4)
 
 #define DBAT_ENTRY_NUM	22
 #define RX_QUEUE_OFFSET	4
diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
index c8988c0c85a1..43db69d03684 100644
--- a/drivers/net/ethernet/renesas/ravb_main.c
+++ b/drivers/net/ethernet/renesas/ravb_main.c
@@ -749,13 +749,18 @@ static void ravb_get_tx_tstamp(struct net_device *ndev)
 static void ravb_rx_csum_gbeth(struct sk_buff *skb)
 {
 	struct skb_shared_info *shinfo = skb_shinfo(skb);
-	__wsum csum_ip_hdr, csum_proto;
+	__wsum csum_proto;
 	skb_frag_t *last_frag;
 	u8 *hw_csum;
 
 	/* The hardware checksum status is contained in sizeof(__sum16) * 2 = 4
-	 * bytes appended to packet data. First 2 bytes is ip header checksum
-	 * and last 2 bytes is protocol checksum.
+	 * bytes appended to packet data.
+	 *
+	 * For ipv4, the first 2 bytes are the ip header checksum status. We can
+	 * ignore this as it will always be re-checked in inet_gro_receive().
+	 *
+	 * The last 2 bytes are the protocol checksum status which will be zero
+	 * if the checksum has been validated.
 	 */
 	if (unlikely(skb->len < sizeof(__sum16) * 2))
 		return;
@@ -771,16 +776,13 @@ static void ravb_rx_csum_gbeth(struct sk_buff *skb)
 	hw_csum -= sizeof(__sum16);
 	csum_proto = csum_unfold((__force __sum16)get_unaligned_le16(hw_csum));
 
-	hw_csum -= sizeof(__sum16);
-	csum_ip_hdr = csum_unfold((__force __sum16)get_unaligned_le16(hw_csum));
-
 	if (skb_is_nonlinear(skb))
 		skb_frag_size_sub(last_frag, 2 * sizeof(__sum16));
 	else
 		skb_trim(skb, skb->len - 2 * sizeof(__sum16));
 
 	/* TODO: IPV6 Rx checksum */
-	if (skb->protocol == htons(ETH_P_IP) && !csum_ip_hdr && !csum_proto)
+	if (skb->protocol == htons(ETH_P_IP) && !csum_proto)
 		skb->ip_summed = CHECKSUM_UNNECESSARY;
 }
 
-- 
2.43.0

[net-next PATCH v2 03/10] net: ravb: Drop IP protocol check from RX csum verification

[Paul Barker]

We do not need to confirm that the protocol is IPv4. If the hardware
encounters an unsupported protocol, it will set the checksum value to
0xFFFF.

Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
---
 drivers/net/ethernet/renesas/ravb_main.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
index 43db69d03684..4bc2532706c2 100644
--- a/drivers/net/ethernet/renesas/ravb_main.c
+++ b/drivers/net/ethernet/renesas/ravb_main.c
@@ -781,8 +781,7 @@ static void ravb_rx_csum_gbeth(struct sk_buff *skb)
 	else
 		skb_trim(skb, skb->len - 2 * sizeof(__sum16));
 
-	/* TODO: IPV6 Rx checksum */
-	if (skb->protocol == htons(ETH_P_IP) && !csum_proto)
+	if (!csum_proto)
 		skb->ip_summed = CHECKSUM_UNNECESSARY;
 }
 
-- 
2.43.0

[net-next PATCH v2 04/10] net: ravb: Combine if conditions in RX csum validation

[Paul Barker]

We can merge the two if conditions on skb_is_nonlinear(). Since
skb_frag_size_sub() and skb_trim() do not free memory, it is still safe
to access the trimmed bytes at the end of the packet after these calls.

Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
---
 drivers/net/ethernet/renesas/ravb_main.c | 11 ++++-------
 1 file changed, 4 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
index 4bc2532706c2..2f584c353c78 100644
--- a/drivers/net/ethernet/renesas/ravb_main.c
+++ b/drivers/net/ethernet/renesas/ravb_main.c
@@ -750,7 +750,6 @@ static void ravb_rx_csum_gbeth(struct sk_buff *skb)
 {
 	struct skb_shared_info *shinfo = skb_shinfo(skb);
 	__wsum csum_proto;
-	skb_frag_t *last_frag;
 	u8 *hw_csum;
 
 	/* The hardware checksum status is contained in sizeof(__sum16) * 2 = 4
@@ -766,21 +765,19 @@ static void ravb_rx_csum_gbeth(struct sk_buff *skb)
 		return;
 
 	if (skb_is_nonlinear(skb)) {
-		last_frag = &shinfo->frags[shinfo->nr_frags - 1];
+		skb_frag_t *last_frag = &shinfo->frags[shinfo->nr_frags - 1];
+
 		hw_csum = skb_frag_address(last_frag) +
 			  skb_frag_size(last_frag);
+		skb_frag_size_sub(last_frag, 2 * sizeof(__sum16));
 	} else {
 		hw_csum = skb_tail_pointer(skb);
+		skb_trim(skb, skb->len - 2 * sizeof(__sum16));
 	}
 
 	hw_csum -= sizeof(__sum16);
 	csum_proto = csum_unfold((__force __sum16)get_unaligned_le16(hw_csum));
 
-	if (skb_is_nonlinear(skb))
-		skb_frag_size_sub(last_frag, 2 * sizeof(__sum16));
-	else
-		skb_trim(skb, skb->len - 2 * sizeof(__sum16));
-
 	if (!csum_proto)
 		skb->ip_summed = CHECKSUM_UNNECESSARY;
 }
-- 
2.43.0

[net-next PATCH v2 05/10] net: ravb: Simplify types in RX csum validation

[Paul Barker]

The hardware checksum value is used as a 16-bit flag, it is zero when
the checksum has been validated and non-zero otherwise. Therefore we
don't need to treat this as an actual __wsum type or call csum_unfold(),
we can just use a u16 pointer.

Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
---
 drivers/net/ethernet/renesas/ravb_main.c | 26 +++++++++++-------------
 1 file changed, 12 insertions(+), 14 deletions(-)

diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
index 2f584c353c78..ca8f785b96b4 100644
--- a/drivers/net/ethernet/renesas/ravb_main.c
+++ b/drivers/net/ethernet/renesas/ravb_main.c
@@ -749,11 +749,11 @@ static void ravb_get_tx_tstamp(struct net_device *ndev)
 static void ravb_rx_csum_gbeth(struct sk_buff *skb)
 {
 	struct skb_shared_info *shinfo = skb_shinfo(skb);
-	__wsum csum_proto;
-	u8 *hw_csum;
+	size_t csum_len;
+	u16 *hw_csum;
 
-	/* The hardware checksum status is contained in sizeof(__sum16) * 2 = 4
-	 * bytes appended to packet data.
+	/* The hardware checksum status is contained in 4 bytes appended to
+	 * packet data.
 	 *
 	 * For ipv4, the first 2 bytes are the ip header checksum status. We can
 	 * ignore this as it will always be re-checked in inet_gro_receive().
@@ -761,24 +761,22 @@ static void ravb_rx_csum_gbeth(struct sk_buff *skb)
 	 * The last 2 bytes are the protocol checksum status which will be zero
 	 * if the checksum has been validated.
 	 */
-	if (unlikely(skb->len < sizeof(__sum16) * 2))
+	csum_len = sizeof(*hw_csum) * 2;
+	if (unlikely(skb->len < csum_len))
 		return;
 
 	if (skb_is_nonlinear(skb)) {
 		skb_frag_t *last_frag = &shinfo->frags[shinfo->nr_frags - 1];
 
-		hw_csum = skb_frag_address(last_frag) +
-			  skb_frag_size(last_frag);
-		skb_frag_size_sub(last_frag, 2 * sizeof(__sum16));
+		hw_csum = (u16 *)(skb_frag_address(last_frag) +
+				  skb_frag_size(last_frag));
+		skb_frag_size_sub(last_frag, csum_len);
 	} else {
-		hw_csum = skb_tail_pointer(skb);
-		skb_trim(skb, skb->len - 2 * sizeof(__sum16));
+		hw_csum = (u16 *)skb_tail_pointer(skb);
+		skb_trim(skb, skb->len - csum_len);
 	}
 
-	hw_csum -= sizeof(__sum16);
-	csum_proto = csum_unfold((__force __sum16)get_unaligned_le16(hw_csum));
-
-	if (!csum_proto)
+	if (!get_unaligned(--hw_csum))
 		skb->ip_summed = CHECKSUM_UNNECESSARY;
 }
 
-- 
2.43.0

Re: [net-next PATCH v2 05/10] net: ravb: Simplify types in RX csum validation

[Sergey Shtylyov]

On 10/15/24 4:36 PM, Paul Barker wrote:

> The hardware checksum value is used as a 16-bit flag, it is zero when
> the checksum has been validated and non-zero otherwise. Therefore we
> don't need to treat this as an actual __wsum type or call csum_unfold(),
> we can just use a u16 pointer.
> 
> Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
[...]

Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>

MBR, Sergey

[net-next PATCH v2 06/10] net: ravb: Disable IP header TX checksum offloading

[Paul Barker]

For IPv4 packets, the header checksum will always be calculated in software
in the TX path (Documentation/networking/checksum-offloads.rst says "No
offloading of the IP header checksum is performed; it is always done in
software.") so there is no advantage in asking the hardware to also
calculate this checksum.

Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
---
 drivers/net/ethernet/renesas/ravb.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/renesas/ravb.h b/drivers/net/ethernet/renesas/ravb.h
index 98496aa39f3d..a5b4f4fe77b1 100644
--- a/drivers/net/ethernet/renesas/ravb.h
+++ b/drivers/net/ethernet/renesas/ravb.h
@@ -998,7 +998,7 @@ enum CSR1_BIT {
 	CSR1_TDHD	= 0x08000000,
 };
 
-#define CSR1_CSUM_ENABLE (CSR1_TIP4 | CSR1_TTCP4 | CSR1_TUDP4)
+#define CSR1_CSUM_ENABLE (CSR1_TTCP4 | CSR1_TUDP4)
 
 enum CSR2_BIT {
 	CSR2_RIP4	= 0x00000001,
-- 
2.43.0

[net-next PATCH v2 07/10] net: ravb: Simplify UDP TX checksum offload

[Paul Barker]

The GbEth IP will pass through a zero UDP checksum without asserting any
error flags so we do not need to resort to software checksum calculation
in this case.

Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
---
 drivers/net/ethernet/renesas/ravb_main.c | 11 +----------
 1 file changed, 1 insertion(+), 10 deletions(-)

diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
index ca8f785b96b4..80c0d36bffcb 100644
--- a/drivers/net/ethernet/renesas/ravb_main.c
+++ b/drivers/net/ethernet/renesas/ravb_main.c
@@ -2075,20 +2075,11 @@ static bool ravb_can_tx_csum_gbeth(struct sk_buff *skb)
 
 	switch (ip->protocol) {
 	case IPPROTO_TCP:
-		break;
 	case IPPROTO_UDP:
-		/* If the checksum value in the UDP header field is 0, TOE does
-		 * not calculate checksum for UDP part of this frame as it is
-		 * optional function as per standards.
-		 */
-		if (udp_hdr(skb)->check == 0)
-			return false;
-		break;
+		return true;
 	default:
 		return false;
 	}
-
-	return true;
 }
 
 /* Packet transmit function for Ethernet AVB */
-- 
2.43.0

[net-next PATCH v2 08/10] net: ravb: Enable IPv6 RX checksum offloading for GbEth

[Paul Barker]

The GbEth IP supports offloading IPv6 TCP, UDP & ICMPv6 checksums in the
RX path.

Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
---
 drivers/net/ethernet/renesas/ravb.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/renesas/ravb.h b/drivers/net/ethernet/renesas/ravb.h
index a5b4f4fe77b1..e1e55e677215 100644
--- a/drivers/net/ethernet/renesas/ravb.h
+++ b/drivers/net/ethernet/renesas/ravb.h
@@ -1014,7 +1014,8 @@ enum CSR2_BIT {
 	CSR2_RDHD	= 0x08000000,
 };
 
-#define CSR2_CSUM_ENABLE (CSR2_RTCP4 | CSR2_RUDP4 | CSR2_RICMP4)
+#define CSR2_CSUM_ENABLE (CSR2_RTCP4 | CSR2_RUDP4 | CSR2_RICMP4 | \
+			  CSR2_RTCP6 | CSR2_RUDP6 | CSR2_RICMP6)
 
 #define DBAT_ENTRY_NUM	22
 #define RX_QUEUE_OFFSET	4
-- 
2.43.0

[net-next PATCH v2 09/10] net: ravb: Enable IPv6 TX checksum offload for GbEth

[Paul Barker]

The GbEth IP supports offloading IPv6 TCP, UDP & ICMPv6 checksums in the
TX path.

Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
---
 drivers/net/ethernet/renesas/ravb.h      |  2 +-
 drivers/net/ethernet/renesas/ravb_main.c | 15 +++++++++++----
 2 files changed, 12 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/renesas/ravb.h b/drivers/net/ethernet/renesas/ravb.h
index e1e55e677215..d7b3810ce21b 100644
--- a/drivers/net/ethernet/renesas/ravb.h
+++ b/drivers/net/ethernet/renesas/ravb.h
@@ -998,7 +998,7 @@ enum CSR1_BIT {
 	CSR1_TDHD	= 0x08000000,
 };
 
-#define CSR1_CSUM_ENABLE (CSR1_TTCP4 | CSR1_TUDP4)
+#define CSR1_CSUM_ENABLE (CSR1_TTCP4 | CSR1_TUDP4 | CSR1_TTCP6 | CSR1_TUDP6)
 
 enum CSR2_BIT {
 	CSR2_RIP4	= 0x00000001,
diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
index 80c0d36bffcb..14b4462331b0 100644
--- a/drivers/net/ethernet/renesas/ravb_main.c
+++ b/drivers/net/ethernet/renesas/ravb_main.c
@@ -2063,17 +2063,24 @@ static void ravb_tx_timeout_work(struct work_struct *work)
 
 static bool ravb_can_tx_csum_gbeth(struct sk_buff *skb)
 {
-	struct iphdr *ip = ip_hdr(skb);
+	u8 inner_protocol;
 
 	/* TODO: Need to add support for VLAN tag 802.1Q */
 	if (skb_vlan_tag_present(skb))
 		return false;
 
-	/* TODO: Need to add hardware checksum for IPv6 */
-	if (skb->protocol != htons(ETH_P_IP))
+	switch (ntohs(skb->protocol)) {
+	case ETH_P_IP:
+		inner_protocol = ip_hdr(skb)->protocol;
+		break;
+	case ETH_P_IPV6:
+		inner_protocol = ipv6_hdr(skb)->nexthdr;
+		break;
+	default:
 		return false;
+	}
 
-	switch (ip->protocol) {
+	switch (inner_protocol) {
 	case IPPROTO_TCP:
 	case IPPROTO_UDP:
 		return true;
-- 
2.43.0

Re: [net-next PATCH v2 09/10] net: ravb: Enable IPv6 TX checksum offload for GbEth

[Sergey Shtylyov]

On 10/15/24 4:36 PM, Paul Barker wrote:

> The GbEth IP supports offloading IPv6 TCP, UDP & ICMPv6 checksums in the
> TX path.
> 
> Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
[...]

Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>

[...]

MBR, Sergey

[net-next PATCH v2 10/10] net: ravb: Add VLAN checksum support

[Paul Barker]

The GbEth IP supports offloading checksum calculation for VLAN-tagged
packets, provided that the EtherType is 0x8100 and only one VLAN tag is
present.

Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>
---
 drivers/net/ethernet/renesas/ravb.h      |  1 +
 drivers/net/ethernet/renesas/ravb_main.c | 24 ++++++++++++++++++++----
 2 files changed, 21 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/renesas/ravb.h b/drivers/net/ethernet/renesas/ravb.h
index d7b3810ce21b..7b48060c250b 100644
--- a/drivers/net/ethernet/renesas/ravb.h
+++ b/drivers/net/ethernet/renesas/ravb.h
@@ -1055,6 +1055,7 @@ struct ravb_hw_info {
 	size_t gstrings_size;
 	netdev_features_t net_hw_features;
 	netdev_features_t net_features;
+	netdev_features_t vlan_features;
 	int stats_len;
 	u32 tccr_mask;
 	u32 tx_max_frame_size;
diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
index 14b4462331b0..bc56f1f4bec9 100644
--- a/drivers/net/ethernet/renesas/ravb_main.c
+++ b/drivers/net/ethernet/renesas/ravb_main.c
@@ -2063,13 +2063,27 @@ static void ravb_tx_timeout_work(struct work_struct *work)
 
 static bool ravb_can_tx_csum_gbeth(struct sk_buff *skb)
 {
+	u16 net_protocol = ntohs(skb->protocol);
 	u8 inner_protocol;
 
-	/* TODO: Need to add support for VLAN tag 802.1Q */
-	if (skb_vlan_tag_present(skb))
-		return false;
+	/* GbEth IP can calculate the checksum if:
+	 * - there are zero or one VLAN headers with TPID=0x8100
+	 * - the network protocol is IPv4 or IPv6
+	 * - the transport protocol is TCP, UDP or ICMP
+	 * - the packet is not fragmented
+	 */
+
+	if (net_protocol == ETH_P_8021Q) {
+		struct vlan_hdr vhdr, *vh;
+
+		vh = skb_header_pointer(skb, ETH_HLEN, sizeof(vhdr), &vhdr);
+		if (!vh)
+			return false;
+
+		net_protocol = ntohs(vh->h_vlan_encapsulated_proto);
+	}
 
-	switch (ntohs(skb->protocol)) {
+	switch (net_protocol) {
 	case ETH_P_IP:
 		inner_protocol = ip_hdr(skb)->protocol;
 		break;
@@ -2772,6 +2786,7 @@ static const struct ravb_hw_info gbeth_hw_info = {
 	.gstrings_size = sizeof(ravb_gstrings_stats_gbeth),
 	.net_hw_features = NETIF_F_RXCSUM | NETIF_F_HW_CSUM,
 	.net_features = NETIF_F_RXCSUM | NETIF_F_HW_CSUM,
+	.vlan_features = NETIF_F_RXCSUM | NETIF_F_HW_CSUM,
 	.stats_len = ARRAY_SIZE(ravb_gstrings_stats_gbeth),
 	.tccr_mask = TCCR_TSRQ0,
 	.tx_max_frame_size = 1522,
@@ -2914,6 +2929,7 @@ static int ravb_probe(struct platform_device *pdev)
 
 	ndev->features = info->net_features;
 	ndev->hw_features = info->net_hw_features;
+	ndev->vlan_features = info->vlan_features;
 
 	error = reset_control_deassert(rstc);
 	if (error)
-- 
2.43.0

Re: [net-next PATCH v2 10/10] net: ravb: Add VLAN checksum support

[Sergey Shtylyov]

On 10/15/24 4:36 PM, Paul Barker wrote:

> The GbEth IP supports offloading checksum calculation for VLAN-tagged
> packets, provided that the EtherType is 0x8100 and only one VLAN tag is
> present.
> 
> Signed-off-by: Paul Barker <paul.barker.ct@bp.renesas.com>

Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>

[...]
MBR, Sergey

Re: [net-next PATCH v2 00/10] Extend GbEth checksum offload support to VLAN/IPv6 packets

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net-next.git (main)
by Andrew Lunn <andrew@lunn.ch>:

On Tue, 15 Oct 2024 14:36:24 +0100 you wrote:
> The GbEth IP found in Renesas RZ/G2L, RZ/G3S and related SoCs supports
> hardware checksum offload for packets in the following cases:
> 
>  - there are zero or one VLAN headers with TPID=0x8100
>  - the network protocol is IPv4 or IPv6
>  - the transport protocol is TCP, UDP or ICMP
>  - the packet is not fragmented
> 
> [...]

Here is the summary with links:
  - [net-next,v2,01/10] net: ravb: Factor out checksum offload enable bits
    https://git.kernel.org/netdev/net-next/c/8e3037924a36
  - [net-next,v2,02/10] net: ravb: Disable IP header RX checksum offloading
    https://git.kernel.org/netdev/net-next/c/c4e347a02b14
  - [net-next,v2,03/10] net: ravb: Drop IP protocol check from RX csum verification
    https://git.kernel.org/netdev/net-next/c/8d2109c1a515
  - [net-next,v2,04/10] net: ravb: Combine if conditions in RX csum validation
    https://git.kernel.org/netdev/net-next/c/5a2d973e3606
  - [net-next,v2,05/10] net: ravb: Simplify types in RX csum validation
    https://git.kernel.org/netdev/net-next/c/faacdbba0180
  - [net-next,v2,06/10] net: ravb: Disable IP header TX checksum offloading
    https://git.kernel.org/netdev/net-next/c/4574ba5b711d
  - [net-next,v2,07/10] net: ravb: Simplify UDP TX checksum offload
    https://git.kernel.org/netdev/net-next/c/e63b5fd02a00
  - [net-next,v2,08/10] net: ravb: Enable IPv6 RX checksum offloading for GbEth
    https://git.kernel.org/netdev/net-next/c/59cceae40c67
  - [net-next,v2,09/10] net: ravb: Enable IPv6 TX checksum offload for GbEth
    https://git.kernel.org/netdev/net-next/c/85c171509821
  - [net-next,v2,10/10] net: ravb: Add VLAN checksum support
    https://git.kernel.org/netdev/net-next/c/546875ccba93

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html