net/mlx5: fix Tx checksum offloads

Tx checksum offloads are correctly handled in a single Tx burst function
whereas the capability is always set.
This causes VXLAN packet with checksum offloads request to be ignored when
the (E)MPS Tx functions are selected.

Fixes: f5fde52051 ("net/mlx5: add hardware checksum offload for tunnel packets")
Cc: stable@dpdk.org

Signed-off-by: Nelio Laranjeiro <nelio.laranjeiro@6wind.com>
Acked-by: Yongseok Koh <yskoh@mellanox.com>
This commit is contained in:
Nélio Laranjeiro 2017-11-20 16:35:47 +01:00 committed by Ferruh Yigit
parent 42473d6782
commit 4aa15eb19a
6 changed files with 48 additions and 86 deletions

View File

@ -800,7 +800,7 @@ mlx5_pci_probe(struct rte_pci_driver *pci_drv, struct rte_pci_device *pci_dev)
priv->hw_csum_l2tun = !!(exp_device_attr.exp_device_cap_flags &
IBV_DEVICE_VXLAN_SUPPORT);
#endif
DEBUG("L2 tunnel checksum offloads are %ssupported",
DEBUG("Rx L2 tunnel checksum offloads are %ssupported",
(priv->hw_csum_l2tun ? "" : "not "));
#ifdef HAVE_IBV_DEVICE_COUNTERS_SET_SUPPORT

View File

@ -372,7 +372,7 @@ mlx5_tx_burst(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n)
uint16_t pkt_inline_sz = MLX5_WQE_DWORD_SIZE + 2;
uint16_t tso_header_sz = 0;
uint16_t ehdr;
uint8_t cs_flags = 0;
uint8_t cs_flags;
uint64_t tso = 0;
uint16_t tso_segsz = 0;
#ifdef MLX5_PMD_SOFT_COUNTERS
@ -415,23 +415,7 @@ mlx5_tx_burst(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n)
if (pkts_n - i > 1)
rte_prefetch0(
rte_pktmbuf_mtod(*(pkts + 1), volatile void *));
/* Should we enable HW CKSUM offload */
if (buf->ol_flags &
(PKT_TX_IP_CKSUM | PKT_TX_TCP_CKSUM | PKT_TX_UDP_CKSUM)) {
const uint64_t is_tunneled = buf->ol_flags &
(PKT_TX_TUNNEL_GRE |
PKT_TX_TUNNEL_VXLAN);
if (is_tunneled && txq->tunnel_en) {
cs_flags = MLX5_ETH_WQE_L3_INNER_CSUM |
MLX5_ETH_WQE_L4_INNER_CSUM;
if (buf->ol_flags & PKT_TX_OUTER_IP_CKSUM)
cs_flags |= MLX5_ETH_WQE_L3_CSUM;
} else {
cs_flags = MLX5_ETH_WQE_L3_CSUM |
MLX5_ETH_WQE_L4_CSUM;
}
}
cs_flags = txq_ol_cksum_to_cs(txq, buf);
raw = ((uint8_t *)(uintptr_t)wqe) + 2 * MLX5_WQE_DWORD_SIZE;
/* Replace the Ethernet type by the VLAN if necessary. */
if (buf->ol_flags & PKT_TX_VLAN_PKT) {
@ -845,7 +829,7 @@ mlx5_tx_burst_mpw(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n)
struct rte_mbuf *buf = *(pkts++);
uint32_t length;
unsigned int segs_n = buf->nb_segs;
uint32_t cs_flags = 0;
uint32_t cs_flags;
/*
* Make sure there is enough room to store this packet and
@ -861,10 +845,7 @@ mlx5_tx_burst_mpw(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n)
}
max_elts -= segs_n;
--pkts_n;
/* Should we enable HW CKSUM offload */
if (buf->ol_flags &
(PKT_TX_IP_CKSUM | PKT_TX_TCP_CKSUM | PKT_TX_UDP_CKSUM))
cs_flags = MLX5_ETH_WQE_L3_CSUM | MLX5_ETH_WQE_L4_CSUM;
cs_flags = txq_ol_cksum_to_cs(txq, buf);
/* Retrieve packet information. */
length = PKT_LEN(buf);
assert(length);
@ -1070,7 +1051,7 @@ mlx5_tx_burst_mpw_inline(void *dpdk_txq, struct rte_mbuf **pkts,
uintptr_t addr;
uint32_t length;
unsigned int segs_n = buf->nb_segs;
uint32_t cs_flags = 0;
uint8_t cs_flags;
/*
* Make sure there is enough room to store this packet and
@ -1091,10 +1072,7 @@ mlx5_tx_burst_mpw_inline(void *dpdk_txq, struct rte_mbuf **pkts,
* iteration.
*/
max_wqe = (1u << txq->wqe_n) - (txq->wqe_ci - txq->wqe_pi);
/* Should we enable HW CKSUM offload */
if (buf->ol_flags &
(PKT_TX_IP_CKSUM | PKT_TX_TCP_CKSUM | PKT_TX_UDP_CKSUM))
cs_flags = MLX5_ETH_WQE_L3_CSUM | MLX5_ETH_WQE_L4_CSUM;
cs_flags = txq_ol_cksum_to_cs(txq, buf);
/* Retrieve packet information. */
length = PKT_LEN(buf);
/* Start new session if packet differs. */
@ -1363,7 +1341,7 @@ mlx5_tx_burst_empw(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n)
unsigned int do_inline = 0; /* Whether inline is possible. */
uint32_t length;
unsigned int segs_n = buf->nb_segs;
uint32_t cs_flags = 0;
uint8_t cs_flags;
/*
* Make sure there is enough room to store this packet and
@ -1377,10 +1355,7 @@ mlx5_tx_burst_empw(void *dpdk_txq, struct rte_mbuf **pkts, uint16_t pkts_n)
txq->stats.oerrors++;
break;
}
/* Should we enable HW CKSUM offload. */
if (buf->ol_flags &
(PKT_TX_IP_CKSUM | PKT_TX_TCP_CKSUM | PKT_TX_UDP_CKSUM))
cs_flags = MLX5_ETH_WQE_L3_CSUM | MLX5_ETH_WQE_L4_CSUM;
cs_flags = txq_ol_cksum_to_cs(txq, buf);
/* Retrieve packet information. */
length = PKT_LEN(buf);
/* Start new session if:

View File

@ -616,4 +616,38 @@ mlx5_tx_dbrec(struct mlx5_txq_data *txq, volatile struct mlx5_wqe *wqe)
mlx5_tx_dbrec_cond_wmb(txq, wqe, 1);
}
/**
* Convert the Checksum offloads to Verbs.
*
* @param txq_data
* Pointer to the Tx queue.
* @param buf
* Pointer to the mbuf.
*
* @return
* the converted cs_flags.
*/
static __rte_always_inline uint8_t
txq_ol_cksum_to_cs(struct mlx5_txq_data *txq_data, struct rte_mbuf *buf)
{
uint8_t cs_flags = 0;
/* Should we enable HW CKSUM offload */
if (buf->ol_flags &
(PKT_TX_IP_CKSUM | PKT_TX_TCP_CKSUM | PKT_TX_UDP_CKSUM)) {
if (txq_data->tunnel_en &&
(buf->ol_flags &
(PKT_TX_TUNNEL_GRE | PKT_TX_TUNNEL_VXLAN))) {
cs_flags = MLX5_ETH_WQE_L3_INNER_CSUM |
MLX5_ETH_WQE_L4_INNER_CSUM;
if (buf->ol_flags & PKT_TX_OUTER_IP_CKSUM)
cs_flags |= MLX5_ETH_WQE_L3_CSUM;
} else {
cs_flags = MLX5_ETH_WQE_L3_CSUM |
MLX5_ETH_WQE_L4_CSUM;
}
}
return cs_flags;
}
#endif /* RTE_PMD_MLX5_RXTX_H_ */

View File

@ -123,24 +123,7 @@ txq_calc_offload(struct mlx5_txq_data *txq, struct rte_mbuf **pkts,
for (pos = 1; pos < pkts_n; ++pos)
if ((pkts[pos]->ol_flags ^ pkts[0]->ol_flags) & ol_mask)
break;
/* Should open another MPW session for the rest. */
if (pkts[0]->ol_flags &
(PKT_TX_IP_CKSUM | PKT_TX_TCP_CKSUM | PKT_TX_UDP_CKSUM)) {
const uint64_t is_tunneled =
pkts[0]->ol_flags &
(PKT_TX_TUNNEL_GRE |
PKT_TX_TUNNEL_VXLAN);
if (is_tunneled && txq->tunnel_en) {
*cs_flags = MLX5_ETH_WQE_L3_INNER_CSUM |
MLX5_ETH_WQE_L4_INNER_CSUM;
if (pkts[0]->ol_flags & PKT_TX_OUTER_IP_CKSUM)
*cs_flags |= MLX5_ETH_WQE_L3_CSUM;
} else {
*cs_flags = MLX5_ETH_WQE_L3_CSUM |
MLX5_ETH_WQE_L4_CSUM;
}
}
*cs_flags = txq_ol_cksum_to_cs(txq, pkts[0]);
return pos;
}

View File

@ -149,7 +149,7 @@ txq_scatter_v(struct mlx5_txq_data *txq, struct rte_mbuf **pkts,
11, 10, 9, 8, /* bswap32 */
12, 13, 14, 15
};
uint8_t cs_flags = 0;
uint8_t cs_flags;
uint16_t max_elts;
uint16_t max_wqe;
uint8x16_t *t_wqe;
@ -168,22 +168,7 @@ txq_scatter_v(struct mlx5_txq_data *txq, struct rte_mbuf **pkts,
break;
wqe = &((volatile struct mlx5_wqe64 *)
txq->wqes)[wqe_ci & wq_mask].hdr;
if (buf->ol_flags &
(PKT_TX_IP_CKSUM | PKT_TX_TCP_CKSUM | PKT_TX_UDP_CKSUM)) {
const uint64_t is_tunneled =
buf->ol_flags & (PKT_TX_TUNNEL_GRE |
PKT_TX_TUNNEL_VXLAN);
if (is_tunneled && txq->tunnel_en) {
cs_flags = MLX5_ETH_WQE_L3_INNER_CSUM |
MLX5_ETH_WQE_L4_INNER_CSUM;
if (buf->ol_flags & PKT_TX_OUTER_IP_CKSUM)
cs_flags |= MLX5_ETH_WQE_L3_CSUM;
} else {
cs_flags = MLX5_ETH_WQE_L3_CSUM |
MLX5_ETH_WQE_L4_CSUM;
}
}
cs_flags = txq_ol_cksum_to_cs(txq, buf);
/* Title WQEBB pointer. */
t_wqe = (uint8x16_t *)wqe;
dseg = (uint8_t *)(wqe + 1);

View File

@ -148,7 +148,7 @@ txq_scatter_v(struct mlx5_txq_data *txq, struct rte_mbuf **pkts,
8, 9, 10, 11, /* bswap32 */
4, 5, 6, 7, /* bswap32 */
0, 1, 2, 3 /* bswap32 */);
uint8_t cs_flags = 0;
uint8_t cs_flags;
uint16_t max_elts;
uint16_t max_wqe;
__m128i *t_wqe, *dseg;
@ -170,22 +170,7 @@ txq_scatter_v(struct mlx5_txq_data *txq, struct rte_mbuf **pkts,
}
wqe = &((volatile struct mlx5_wqe64 *)
txq->wqes)[wqe_ci & wq_mask].hdr;
if (buf->ol_flags &
(PKT_TX_IP_CKSUM | PKT_TX_TCP_CKSUM | PKT_TX_UDP_CKSUM)) {
const uint64_t is_tunneled =
buf->ol_flags & (PKT_TX_TUNNEL_GRE |
PKT_TX_TUNNEL_VXLAN);
if (is_tunneled && txq->tunnel_en) {
cs_flags = MLX5_ETH_WQE_L3_INNER_CSUM |
MLX5_ETH_WQE_L4_INNER_CSUM;
if (buf->ol_flags & PKT_TX_OUTER_IP_CKSUM)
cs_flags |= MLX5_ETH_WQE_L3_CSUM;
} else {
cs_flags = MLX5_ETH_WQE_L3_CSUM |
MLX5_ETH_WQE_L4_CSUM;
}
}
cs_flags = txq_ol_cksum_to_cs(txq, buf);
/* Title WQEBB pointer. */
t_wqe = (__m128i *)wqe;
dseg = (__m128i *)(wqe + 1);