[dpdk-stable] [PATCH v1] net/ice: revert the TSO fake fix

[dpdk-stable] [PATCH v1] net/ice: revert the TSO fake fix

[Haiyue Wang]

The two fixes are not the real root cause for MDD event, it mitigates
the failure rate when different test mode, so revert them.

Fixes: 2a0c9ae4f646 ("net/ice: fix TCP checksum offload")
Fixes: 7365a3cee51f ("net/ice: calculate TCP header size for offload")
Cc: stable@dpdk.org

Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
---
 drivers/net/ice/ice_rxtx.c | 32 ++------------------------------
 1 file changed, 2 insertions(+), 30 deletions(-)

diff --git a/drivers/net/ice/ice_rxtx.c b/drivers/net/ice/ice_rxtx.c
index b83c5c8629..2e1f06d2c0 100644
--- a/drivers/net/ice/ice_rxtx.c
+++ b/drivers/net/ice/ice_rxtx.c
@@ -2234,7 +2234,7 @@ ice_txd_enable_checksum(uint64_t ol_flags,
 	switch (ol_flags & PKT_TX_L4_MASK) {
 	case PKT_TX_TCP_CKSUM:
 		*td_cmd |= ICE_TX_DESC_CMD_L4T_EOFT_TCP;
-		*td_offset |= (tx_offload.l4_len >> 2) <<
+		*td_offset |= (sizeof(struct rte_tcp_hdr) >> 2) <<
 			      ICE_TX_DESC_LEN_L4_LEN_S;
 		break;
 	case PKT_TX_SCTP_CKSUM:
@@ -2371,28 +2371,6 @@ ice_calc_pkt_desc(struct rte_mbuf *tx_pkt)
 	return count;
 }
 
-/* Calculate TCP header length for PKT_TX_TCP_CKSUM if not provided */
-static inline uint16_t
-ice_calc_pkt_tcp_hdr(struct rte_mbuf *tx_pkt, union ice_tx_offload tx_offload)
-{
-	uint16_t tcpoff = tx_offload.l2_len + tx_offload.l3_len;
-	const struct rte_tcp_hdr *tcp_hdr;
-	struct rte_tcp_hdr _tcp_hdr;
-
-	if (tcpoff + sizeof(struct rte_tcp_hdr) < tx_pkt->data_len) {
-		tcp_hdr = rte_pktmbuf_mtod_offset(tx_pkt, struct rte_tcp_hdr *,
-						  tcpoff);
-
-		return (tcp_hdr->data_off & 0xf0) >> 2;
-	}
-
-	tcp_hdr = rte_pktmbuf_read(tx_pkt, tcpoff, sizeof(_tcp_hdr), &_tcp_hdr);
-	if (tcp_hdr)
-		return (tcp_hdr->data_off & 0xf0) >> 2;
-	else
-		return 0;
-}
-
 uint16_t
 ice_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
 {
@@ -2491,15 +2469,9 @@ ice_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
 						   &cd_tunneling_params);
 
 		/* Enable checksum offloading */
-		if (ol_flags & ICE_TX_CKSUM_OFFLOAD_MASK) {
-			if ((ol_flags & PKT_TX_L4_MASK) == PKT_TX_TCP_CKSUM &&
-			    !tx_offload.l4_len)
-				tx_offload.l4_len =
-				     ice_calc_pkt_tcp_hdr(tx_pkt, tx_offload);
-
+		if (ol_flags & ICE_TX_CKSUM_OFFLOAD_MASK)
 			ice_txd_enable_checksum(ol_flags, &td_cmd,
 						&td_offset, tx_offload);
-		}
 
 		if (nb_ctx) {
 			/* Setup TX context descriptor if required */
-- 
2.28.0

Re: [dpdk-stable] [PATCH v1] net/ice: revert the TSO fake fix

[Zhang, Qi Z]

> -----Original Message-----
> From: Wang, Haiyue <haiyue.wang@intel.com>
> Sent: Friday, July 31, 2020 2:00 PM
> To: dev@dpdk.org; Yang, Qiming <qiming.yang@intel.com>; Zhang, Qi Z
> <qi.z.zhang@intel.com>
> Cc: Xu, Ting <ting.xu@intel.com>; Sun, Chenmin <chenmin.sun@intel.com>;
> Wang, Yinan <yinan.wang@intel.com>; Wang, Haiyue
> <haiyue.wang@intel.com>; stable@dpdk.org
> Subject: [PATCH v1] net/ice: revert the TSO fake fix
> 
> The two fixes are not the real root cause for MDD event, it mitigates the failure
> rate when different test mode, so revert them.
> 
> Fixes: 2a0c9ae4f646 ("net/ice: fix TCP checksum offload")
> Fixes: 7365a3cee51f ("net/ice: calculate TCP header size for offload")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
Acked-by: Qi zhang <qi.z.zhang@intel.com>

Applied to dpdk-next-net-intel.

Thanks
Qi