This patch enables i40e TSO feature for both non-tunneling packet and UDP
tunneling packet.
Signed-off-by: Jijiang Liu <jijiang.liu at intel.com>
Signed-off-by: Miroslaw Walukiewicz <miroslaw.walukiewicz at intel.com>
---
lib/librte_pmd_i40e/i40e_ethdev.c | 3 +-
lib/librte_pmd_i40e/i40e_rxtx.c | 111 +++++++++++++++++++++++++++---------
lib/librte_pmd_i40e/i40e_rxtx.h | 13 ++++
3 files changed, 98 insertions(+), 29 deletions(-)
diff --git a/lib/librte_pmd_i40e/i40e_ethdev.c
b/lib/librte_pmd_i40e/i40e_ethdev.c
index b47a3d2..af95296 100644
--- a/lib/librte_pmd_i40e/i40e_ethdev.c
+++ b/lib/librte_pmd_i40e/i40e_ethdev.c
@@ -1516,7 +1516,8 @@ i40e_dev_info_get(struct rte_eth_dev *dev, struct
rte_eth_dev_info *dev_info)
DEV_TX_OFFLOAD_IPV4_CKSUM |
DEV_TX_OFFLOAD_UDP_CKSUM |
DEV_TX_OFFLOAD_TCP_CKSUM |
- DEV_TX_OFFLOAD_SCTP_CKSUM;
+ DEV_TX_OFFLOAD_SCTP_CKSUM |
+ DEV_TX_OFFLOAD_TCP_TSO;
dev_info->reta_size = pf->hash_lut_size;
dev_info->default_rxconf = (struct rte_eth_rxconf) {
diff --git a/lib/librte_pmd_i40e/i40e_rxtx.c b/lib/librte_pmd_i40e/i40e_rxtx.c
index 2beae3c..529ffb2 100644
--- a/lib/librte_pmd_i40e/i40e_rxtx.c
+++ b/lib/librte_pmd_i40e/i40e_rxtx.c
@@ -460,18 +460,15 @@ static inline void
i40e_txd_enable_checksum(uint64_t ol_flags,
uint32_t *td_cmd,
uint32_t *td_offset,
- uint8_t l2_len,
- uint16_t l3_len,
- uint8_t outer_l2_len,
- uint16_t outer_l3_len,
+ union i40e_tx_offload tx_offload,
uint32_t *cd_tunneling)
{
- if (!l2_len) {
+ if (!tx_offload.l2_len) {
PMD_DRV_LOG(DEBUG, "L2 length set to 0");
return;
}
- if (!l3_len) {
+ if (!tx_offload.l3_len) {
PMD_DRV_LOG(DEBUG, "L3 length set to 0");
return;
}
@@ -479,7 +476,7 @@ i40e_txd_enable_checksum(uint64_t ol_flags,
/* UDP tunneling packet TX checksum offload */
if (unlikely(ol_flags & PKT_TX_UDP_TUNNEL_PKT)) {
- *td_offset |= (outer_l2_len >> 1)
+ *td_offset |= (tx_offload.outer_l2_len >> 1)
<< I40E_TX_DESC_LENGTH_MACLEN_SHIFT;
if (ol_flags & PKT_TX_OUTER_IP_CKSUM)
@@ -490,26 +487,36 @@ i40e_txd_enable_checksum(uint64_t ol_flags,
*cd_tunneling |= I40E_TX_CTX_EXT_IP_IPV6;
/* Now set the ctx descriptor fields */
- *cd_tunneling |= (outer_l3_len >> 2) <<
+ *cd_tunneling |= (tx_offload.outer_l3_len >> 2) <<
I40E_TXD_CTX_QW0_EXT_IPLEN_SHIFT |
I40E_TXD_CTX_UDP_TUNNELING |
- (l2_len >> 1) <<
+ (tx_offload.l2_len >> 1) <<
I40E_TXD_CTX_QW0_NATLEN_SHIFT;
} else
- *td_offset |= (l2_len >> 1)
+ *td_offset |= (tx_offload.l2_len >> 1)
<< I40E_TX_DESC_LENGTH_MACLEN_SHIFT;
/* Enable L3 checksum offloads */
if (ol_flags & PKT_TX_IPV4_CSUM) {
*td_cmd |= I40E_TX_DESC_CMD_IIPT_IPV4_CSUM;
- *td_offset |= (l3_len >> 2) << I40E_TX_DESC_LENGTH_IPLEN_SHIFT;
+ *td_offset |= (tx_offload.l3_len >> 2)
+ << I40E_TX_DESC_LENGTH_IPLEN_SHIFT;
} else if (ol_flags & PKT_TX_IPV4) {
*td_cmd |= I40E_TX_DESC_CMD_IIPT_IPV4;
- *td_offset |= (l3_len >> 2) << I40E_TX_DESC_LENGTH_IPLEN_SHIFT;
+ *td_offset |= (tx_offload.l3_len >> 2)
+ << I40E_TX_DESC_LENGTH_IPLEN_SHIFT;
} else if (ol_flags & PKT_TX_IPV6) {
*td_cmd |= I40E_TX_DESC_CMD_IIPT_IPV6;
- *td_offset |= (l3_len >> 2) << I40E_TX_DESC_LENGTH_IPLEN_SHIFT;
+ *td_offset |= (tx_offload.l3_len >> 2)
+ << I40E_TX_DESC_LENGTH_IPLEN_SHIFT;
+ }
+
+ if (ol_flags & PKT_TX_TCP_SEG) {
+ *td_cmd |= I40E_TX_DESC_CMD_L4T_EOFT_TCP;
+ *td_offset |= (tx_offload.l4_len >> 2)
+ << I40E_TX_DESC_LENGTH_L4_FC_LEN_SHIFT;
+ return;
}
/* Enable L4 checksum offloads */
@@ -1160,8 +1167,11 @@ i40e_calc_context_desc(uint64_t flags)
{
uint64_t mask = 0ULL;
- if (flags | PKT_TX_UDP_TUNNEL_PKT)
+ if (flags & PKT_TX_UDP_TUNNEL_PKT)
mask |= PKT_TX_UDP_TUNNEL_PKT;
+ if (flags & PKT_TX_TCP_SEG)
+ /* need for context descriptor when TSO enabled */
+ mask |= PKT_TX_TCP_SEG;
#ifdef RTE_LIBRTE_IEEE1588
mask |= PKT_TX_IEEE1588_TMST;
@@ -1172,6 +1182,47 @@ i40e_calc_context_desc(uint64_t flags)
return 0;
}
+/* set i40e TSO context descriptor */
+static inline uint64_t
+i40e_set_tso_ctx(struct rte_mbuf *mbuf, union i40e_tx_offload tx_offload)
+{
+
+ uint64_t ctx_desc = 0;
+ uint32_t cd_cmd, hdr_len, cd_tso_len;
+
+
+ if (!tx_offload.l4_len) {
+ PMD_DRV_LOG(DEBUG, "L4 length set to 0");
+ return ctx_desc;
+ }
+
+ if (unlikely(mbuf->ol_flags & PKT_TX_UDP_TUNNEL_PKT)) {
+
+ /**
+ * Caculate total header length of UDP tunneling packet.
+ * the l2_len is outer UDP header length plus tunnel
+ * header length plus inner L2 header length.
+ */
+ hdr_len = tx_offload.outer_l2_len +
+ tx_offload.outer_l3_len +
+ tx_offload.l2_len +
+ tx_offload.l3_len +
+ tx_offload.l4_len;
+ } else
+ hdr_len = tx_offload.l2_len + tx_offload.l3_len +
+ tx_offload.l4_len;
+
+ cd_cmd = I40E_TX_CTX_DESC_TSO;
+ cd_tso_len = mbuf->pkt_len - hdr_len;
+ ctx_desc |= ((uint64_t)cd_cmd << I40E_TXD_CTX_QW1_CMD_SHIFT) |
+ ((uint64_t)cd_tso_len <<
+ I40E_TXD_CTX_QW1_TSO_LEN_SHIFT) |
+ ((uint64_t)mbuf->tso_segsz <<
+ I40E_TXD_CTX_QW1_MSS_SHIFT);
+
+ return ctx_desc;
+}
+
uint16_t
i40e_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts, uint16_t nb_pkts)
{
@@ -1190,15 +1241,12 @@ i40e_xmit_pkts(void *tx_queue, struct rte_mbuf
**tx_pkts, uint16_t nb_pkts)
uint32_t tx_flags;
uint32_t td_tag;
uint64_t ol_flags;
- uint8_t l2_len;
- uint16_t l3_len;
- uint8_t outer_l2_len;
- uint16_t outer_l3_len;
uint16_t nb_used;
uint16_t nb_ctx;
uint16_t tx_last;
uint16_t slen;
uint64_t buf_dma_addr;
+ union i40e_tx_offload tx_offload = { .data = 0 };
txq = tx_queue;
sw_ring = txq->sw_ring;
@@ -1220,10 +1268,12 @@ i40e_xmit_pkts(void *tx_queue, struct rte_mbuf
**tx_pkts, uint16_t nb_pkts)
RTE_MBUF_PREFETCH_TO_FREE(txe->mbuf);
ol_flags = tx_pkt->ol_flags;
- l2_len = tx_pkt->l2_len;
- l3_len = tx_pkt->l3_len;
- outer_l2_len = tx_pkt->outer_l2_len;
- outer_l3_len = tx_pkt->outer_l3_len;
+ tx_offload.l2_len = tx_pkt->l2_len;
+ tx_offload.l3_len = tx_pkt->l3_len;
+ tx_offload.l4_len = tx_pkt->l4_len;
+ tx_offload.tso_segsz = tx_pkt->tso_segsz;
+ tx_offload.outer_l2_len = tx_pkt->outer_l2_len;
+ tx_offload.outer_l3_len = tx_pkt->outer_l3_len;
/* Calculate the number of context descriptors needed. */
nb_ctx = i40e_calc_context_desc(ol_flags);
@@ -1273,8 +1323,7 @@ i40e_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts,
uint16_t nb_pkts)
/* Enable checksum offloading */
cd_tunneling_params = 0;
i40e_txd_enable_checksum(ol_flags, &td_cmd, &td_offset,
- l2_len, l3_len, outer_l2_len,
- outer_l3_len,
+ tx_offload,
&cd_tunneling_params);
if (unlikely(nb_ctx)) {
@@ -1292,12 +1341,18 @@ i40e_xmit_pkts(void *tx_queue, struct rte_mbuf
**tx_pkts, uint16_t nb_pkts)
rte_pktmbuf_free_seg(txe->mbuf);
txe->mbuf = NULL;
}
-#ifdef RTE_LIBRTE_IEEE1588
- if (ol_flags & PKT_TX_IEEE1588_TMST)
+ /* TSO enabled means no timestamp */
+ if (ol_flags & PKT_TX_TCP_SEG) {
cd_type_cmd_tso_mss |=
- ((uint64_t)I40E_TX_CTX_DESC_TSYN <<
- I40E_TXD_CTX_QW1_CMD_SHIFT);
+ i40e_set_tso_ctx(tx_pkt, tx_offload);
+ } else {
+#ifdef RTE_LIBRTE_IEEE1588
+ if (ol_flags & PKT_TX_IEEE1588_TMST)
+ cd_type_cmd_tso_mss |=
+
((uint64_t)I40E_TX_CTX_DESC_TSYN <<
+
I40E_TXD_CTX_QW1_CMD_SHIFT);
#endif
+ }
ctx_txd->tunneling_params =
rte_cpu_to_le_32(cd_tunneling_params);
ctx_txd->l2tag2 = rte_cpu_to_le_16(cd_l2tag2);
diff --git a/lib/librte_pmd_i40e/i40e_rxtx.h b/lib/librte_pmd_i40e/i40e_rxtx.h
index af932e3..4c81a24 100644
--- a/lib/librte_pmd_i40e/i40e_rxtx.h
+++ b/lib/librte_pmd_i40e/i40e_rxtx.h
@@ -154,6 +154,19 @@ struct i40e_tx_queue {
bool tx_deferred_start; /**< don't start this queue in dev start */
};
+/** Offload features */
+union i40e_tx_offload {
+ uint64_t data;
+ struct {
+ uint64_t l2_len:7; /**< L2 (MAC) Header Length. */
+ uint64_t l3_len:9; /**< L3 (IP) Header Length. */
+ uint64_t l4_len:8; /**< L4 (TCP/UDP) Header Length. */
+ uint64_t tso_segsz:16; /**< TCP TSO segment size */
+ uint64_t outer_l2_len:8; /**< L2 outer Header Length */
+ uint64_t outer_l3_len:16; /**< L2 outer Header Length */
+ };
+};
+
int i40e_dev_rx_queue_start(struct rte_eth_dev *dev, uint16_t rx_queue_id);
int i40e_dev_rx_queue_stop(struct rte_eth_dev *dev, uint16_t rx_queue_id);
int i40e_dev_tx_queue_start(struct rte_eth_dev *dev, uint16_t tx_queue_id);
--
1.7.7.6
