[dpdk-dev] [PATCH 4/6] net/hns3: fix Tx checksum for UDP packets with special port

["Min Hu (Connor)" <humin29@huawei.com>]

From: Chengchang Tang <tangchengchang@huawei.com>

For Kunpeng920 network engine, UDP packets with destination port 6081,
4789 or 4790 will be identified as tunnel packets. If the UDP CKSUM
offload is set in the mbuf, and the TX tunnel mask is not set, the
CKSUM of these packets will be wrong. In this case, the upper layer
user may not identify the packet as a tunnel packet, and processes it
as non-tunnel packet, and expect to offload the outer UDP CKSUM, so
they may not fill the outer L2/L3 length to mbuf. However, the HW
identifies these packet as tunnel packets and therefore offload the
inner UDP CKSUM. As a result, the inner and outer UDP CKSUM are
incorrect. And for non-tunnel UDP packets with preceding special
destination port will also exist similar checksum error.

For the new generation Kunpeng930 network engine, the above errata
have been fixed. Therefore, the concept of udp_cksum_mode is
introduced. There are two udp_cksum_mode for hns3 PMD,
HNS3_SPECIAL_PORT_HW_CKSUM_MODE means HW could solve the above
problem. And in HNS3_SPECIAL_PORT_SW_CKSUM_MODE, hns3 PMD will check
packets in the Tx prepare and perform the UDP CKSUM for such packets
to avoid a checksum error.

Fixes: bba636698316 ("net/hns3: support Rx/Tx and related operations")
Cc: stable@dpdk.org

Signed-off-by: Chengchang Tang <tangchengchang@huawei.com>
Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
---
 drivers/net/hns3/hns3_ethdev.c |  2 ++
 drivers/net/hns3/hns3_ethdev.h | 19 ++++++++++++
 drivers/net/hns3/hns3_rxtx.c   | 68 ++++++++++++++++++++++++++++++++++++++++++
 drivers/net/hns3/hns3_rxtx.h   | 16 ++++++++++
 4 files changed, 105 insertions(+)

diff --git a/drivers/net/hns3/hns3_ethdev.c b/drivers/net/hns3/hns3_ethdev.c
index 410b241..c318350 100644
--- a/drivers/net/hns3/hns3_ethdev.c
+++ b/drivers/net/hns3/hns3_ethdev.c
@@ -3129,6 +3129,7 @@ hns3_get_capability(struct hns3_hw *hw)
 		hw->min_tx_pkt_len = HNS3_HIP08_MIN_TX_PKT_LEN;
 		pf->tqp_config_mode = HNS3_FIXED_MAX_TQP_NUM_MODE;
 		hw->rss_info.ipv6_sctp_offload_supported = false;
+		hw->udp_cksum_mode = HNS3_SPECIAL_PORT_SW_CKSUM_MODE;
 		return 0;
 	}
 
@@ -3148,6 +3149,7 @@ hns3_get_capability(struct hns3_hw *hw)
 	hw->min_tx_pkt_len = HNS3_HIP09_MIN_TX_PKT_LEN;
 	pf->tqp_config_mode = HNS3_FLEX_MAX_TQP_NUM_MODE;
 	hw->rss_info.ipv6_sctp_offload_supported = true;
+	hw->udp_cksum_mode = HNS3_SPECIAL_PORT_HW_CKSUM_MODE;
 
 	return 0;
 }
diff --git a/drivers/net/hns3/hns3_ethdev.h b/drivers/net/hns3/hns3_ethdev.h
index 586979b..d35ca98 100644
--- a/drivers/net/hns3/hns3_ethdev.h
+++ b/drivers/net/hns3/hns3_ethdev.h
@@ -47,6 +47,9 @@
 #define HNS3_UNLIMIT_PROMISC_MODE       0
 #define HNS3_LIMIT_PROMISC_MODE         1
 
+#define HNS3_SPECIAL_PORT_SW_CKSUM_MODE         0
+#define HNS3_SPECIAL_PORT_HW_CKSUM_MODE         1
+
 #define HNS3_UC_MACADDR_NUM		128
 #define HNS3_VF_UC_MACADDR_NUM		48
 #define HNS3_MC_MACADDR_NUM		128
@@ -567,6 +570,22 @@ struct hns3_hw {
 	uint8_t drop_stats_mode;
 
 	uint8_t max_non_tso_bd_num; /* max BD number of one non-TSO packet */
+	/*
+	 * udp checksum mode.
+	 * value range:
+	 *      HNS3_SPECIAL_PORT_HW_CKSUM_MODE/HNS3_SPECIAL_PORT_SW_CKSUM_MODE
+	 *
+	 *  - HNS3_SPECIAL_PORT_SW_CKSUM_MODE
+	 *     In this mode, HW can not do checksum for special UDP port like
+	 *     4789, 4790, 6081 for non-tunnel UDP packets and UDP tunnel
+	 *     packets without the PKT_TX_TUNEL_MASK in the mbuf. So, PMD need
+	 *     do the checksum for these packets to avoid a checksum error.
+	 *
+	 *  - HNS3_SPECIAL_PORT_HW_CKSUM_MODE
+	 *     In this mode, HW does not have the preceding problems and can
+	 *     directly calculate the checksum of these UDP packets.
+	 */
+	uint8_t udp_cksum_mode;
 
 	struct hns3_port_base_vlan_config port_base_vlan_cfg;
 	/*
diff --git a/drivers/net/hns3/hns3_rxtx.c b/drivers/net/hns3/hns3_rxtx.c
index a76dfa5..3d7e831 100644
--- a/drivers/net/hns3/hns3_rxtx.c
+++ b/drivers/net/hns3/hns3_rxtx.c
@@ -5,6 +5,7 @@
 #include <rte_bus_pci.h>
 #include <rte_common.h>
 #include <rte_cycles.h>
+#include <rte_geneve.h>
 #include <rte_vxlan.h>
 #include <ethdev_driver.h>
 #include <rte_io.h>
@@ -2920,6 +2921,7 @@ hns3_tx_queue_setup(struct rte_eth_dev *dev, uint16_t idx, uint16_t nb_desc,
 					     HNS3_RING_TX_TAIL_REG);
 	txq->min_tx_pkt_len = hw->min_tx_pkt_len;
 	txq->tso_mode = hw->tso_mode;
+	txq->udp_cksum_mode = hw->udp_cksum_mode;
 	memset(&txq->basic_stats, 0, sizeof(struct hns3_tx_basic_stats));
 	memset(&txq->dfx_stats, 0, sizeof(struct hns3_tx_dfx_stats));
 
@@ -3629,6 +3631,69 @@ hns3_vld_vlan_chk(struct hns3_tx_queue *txq, struct rte_mbuf *m)
 }
 #endif
 
+static uint16_t
+hns3_udp_cksum_help(struct rte_mbuf *m)
+{
+	uint64_t ol_flags = m->ol_flags;
+	uint16_t cksum = 0;
+	uint32_t l4_len;
+
+	if (ol_flags & PKT_TX_IPV4) {
+		struct rte_ipv4_hdr *ipv4_hdr = rte_pktmbuf_mtod_offset(m,
+				struct rte_ipv4_hdr *, m->l2_len);
+		l4_len = rte_be_to_cpu_16(ipv4_hdr->total_length) - m->l3_len;
+	} else {
+		struct rte_ipv6_hdr *ipv6_hdr = rte_pktmbuf_mtod_offset(m,
+				struct rte_ipv6_hdr *, m->l2_len);
+		l4_len = rte_be_to_cpu_16(ipv6_hdr->payload_len);
+	}
+
+	rte_raw_cksum_mbuf(m, m->l2_len + m->l3_len, l4_len, &cksum);
+
+	cksum = ~cksum;
+	/*
+	 * RFC 768:If the computed checksum is zero for UDP, it is transmitted
+	 * as all ones
+	 */
+	if (cksum == 0)
+		cksum = 0xffff;
+
+	return (uint16_t)cksum;
+}
+
+static bool
+hns3_validate_tunnel_cksum(struct hns3_tx_queue *tx_queue, struct rte_mbuf *m)
+{
+	uint64_t ol_flags = m->ol_flags;
+	struct rte_udp_hdr *udp_hdr;
+	uint16_t dst_port;
+
+	if (tx_queue->udp_cksum_mode == HNS3_SPECIAL_PORT_HW_CKSUM_MODE ||
+	    ol_flags & PKT_TX_TUNNEL_MASK ||
+	    (ol_flags & PKT_TX_L4_MASK) != PKT_TX_UDP_CKSUM)
+		return true;
+	/*
+	 * A UDP packet with the same dst_port as VXLAN\VXLAN_GPE\GENEVE will
+	 * be recognized as a tunnel packet in HW. In this case, if UDP CKSUM
+	 * offload is set and the tunnel mask has not been set, the CKSUM will
+	 * be wrong since the header length is wrong and driver should complete
+	 * the CKSUM to avoid CKSUM error.
+	 */
+	udp_hdr = rte_pktmbuf_mtod_offset(m, struct rte_udp_hdr *,
+						m->l2_len + m->l3_len);
+	dst_port = rte_be_to_cpu_16(udp_hdr->dst_port);
+	switch (dst_port) {
+	case RTE_VXLAN_DEFAULT_PORT:
+	case RTE_VXLAN_GPE_DEFAULT_PORT:
+	case RTE_GENEVE_DEFAULT_PORT:
+		udp_hdr->dgram_cksum = hns3_udp_cksum_help(m);
+		m->ol_flags = ol_flags & ~PKT_TX_L4_MASK;
+		return false;
+	default:
+		return true;
+	}
+}
+
 static int
 hns3_prep_pkt_proc(struct hns3_tx_queue *tx_queue, struct rte_mbuf *m)
 {
@@ -3673,6 +3738,9 @@ hns3_prep_pkt_proc(struct hns3_tx_queue *tx_queue, struct rte_mbuf *m)
 		return ret;
 	}
 
+	if (!hns3_validate_tunnel_cksum(tx_queue, m))
+		return 0;
+
 	hns3_outer_header_cksum_prepare(m);
 
 	return 0;
diff --git a/drivers/net/hns3/hns3_rxtx.h b/drivers/net/hns3/hns3_rxtx.h
index ab2b05a..2a51c71 100644
--- a/drivers/net/hns3/hns3_rxtx.h
+++ b/drivers/net/hns3/hns3_rxtx.h
@@ -466,6 +466,22 @@ struct hns3_tx_queue {
 	 */
 	uint8_t tso_mode;
 	/*
+	 * udp checksum mode.
+	 * value range:
+	 *      HNS3_SPECIAL_PORT_HW_CKSUM_MODE/HNS3_SPECIAL_PORT_SW_CKSUM_MODE
+	 *
+	 *  - HNS3_SPECIAL_PORT_SW_CKSUM_MODE
+	 *     In this mode, HW can not do checksum for special UDP port like
+	 *     4789, 4790, 6081 for non-tunnel UDP packets and UDP tunnel
+	 *     packets without the PKT_TX_TUNEL_MASK in the mbuf. So, PMD need
+	 *     do the checksum for these packets to avoid a checksum error.
+	 *
+	 *  - HNS3_SPECIAL_PORT_HW_CKSUM_MODE
+	 *     In this mode, HW does not have the preceding problems and can
+	 *     directly calculate the checksum of these UDP packets.
+	 */
+	uint8_t udp_cksum_mode;
+	/*
 	 * The minimum length of the packet supported by hardware in the Tx
 	 * direction.
 	 */
-- 
2.7.4