ligang/vpp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

rdma: add RSS support for IPv6 and TCP

Browse Source 
Type: feature

Change-Id: I8b0d918e6f13325954b29bf34e4ef224c1315c51
Signed-off-by: Benoît Ganne <bganne@cisco.com>
(cherry picked from commit 91603958d1d4fc3114739f9b264808940942e5c8)
This commit is contained in:
Benoît Ganne
2020-10-16 17:12:41 +02:00
committed by Andrew Yourtchenko
parent 9c73329588
commit 32c2b2ae60
2 changed files with 58 additions and 32 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

81
src/plugins/rdma/device.c
View File

@ -57,7 +57,7 @@ rdma_main_t rdma_main;
static struct ibv_flow * static struct ibv_flow *
rdma_rxq_init_flow (const rdma_device_t * rd, struct ibv_qp *qp, rdma_rxq_init_flow (const rdma_device_t * rd, struct ibv_qp *qp,
const mac_address_t * mac, const mac_address_t * mask, const mac_address_t * mac, const mac_address_t * mask,
u32 flags) u16 ether_type, u32 flags)
{ {
struct ibv_flow *flow; struct ibv_flow *flow;
struct raw_eth_flow_attr struct raw_eth_flow_attr
@ -76,6 +76,12 @@ rdma_rxq_init_flow (const rdma_device_t * rd, struct ibv_qp *qp,
memcpy (fa.spec_eth.val.dst_mac, mac, sizeof (fa.spec_eth.val.dst_mac)); memcpy (fa.spec_eth.val.dst_mac, mac, sizeof (fa.spec_eth.val.dst_mac));
memcpy (fa.spec_eth.mask.dst_mac, mask, sizeof (fa.spec_eth.mask.dst_mac)); memcpy (fa.spec_eth.mask.dst_mac, mask, sizeof (fa.spec_eth.mask.dst_mac));
if (ether_type)
{
fa.spec_eth.val.ether_type = ether_type;
fa.spec_eth.mask.ether_type = 0xffff;
}
flow = ibv_create_flow (qp, &fa.attr); flow = ibv_create_flow (qp, &fa.attr);
if (!flow) if (!flow)
rdma_log (VLIB_LOG_LEVEL_ERR, rd, "ibv_create_flow() failed"); rdma_log (VLIB_LOG_LEVEL_ERR, rd, "ibv_create_flow() failed");
@ -104,16 +110,17 @@ rdma_dev_set_promisc (rdma_device_t * rd)
const mac_address_t all = {.bytes = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0} }; const mac_address_t all = {.bytes = {0x0, 0x0, 0x0, 0x0, 0x0, 0x0} };
int err; int err;
err = rdma_rxq_destroy_flow (rd, &rd->flow_mcast); err = rdma_rxq_destroy_flow (rd, &rd->flow_mcast6);
err |= rdma_rxq_destroy_flow (rd, &rd->flow_ucast6);
err |= rdma_rxq_destroy_flow (rd, &rd->flow_mcast4);
err |= rdma_rxq_destroy_flow (rd, &rd->flow_ucast4);
if (err) if (err)
return ~0; return ~0;
err = rdma_rxq_destroy_flow (rd, &rd->flow_ucast); rd->flow_ucast6 =
if (err) rdma_rxq_init_flow (rd, rd->rx_qp6, &all, &all, ntohs (ETH_P_IPV6), 0);
return ~0; rd->flow_ucast4 = rdma_rxq_init_flow (rd, rd->rx_qp4, &all, &all, 0, 0);
if (!rd->flow_ucast6 || !rd->flow_ucast4)
rd->flow_ucast = rdma_rxq_init_flow (rd, rd->rx_qp, &all, &all, 0);
if (!rd->flow_ucast)
return ~0; return ~0;
rd->flags |= RDMA_DEVICE_F_PROMISC; rd->flags |= RDMA_DEVICE_F_PROMISC;
@ -128,25 +135,30 @@ rdma_dev_set_ucast (rdma_device_t * rd)
const mac_address_t mcast = {.bytes = {0x1, 0x0, 0x0, 0x0, 0x0, 0x0} }; const mac_address_t mcast = {.bytes = {0x1, 0x0, 0x0, 0x0, 0x0, 0x0} };
int err; int err;
err = rdma_rxq_destroy_flow (rd, &rd->flow_mcast); err = rdma_rxq_destroy_flow (rd, &rd->flow_mcast6);
err |= rdma_rxq_destroy_flow (rd, &rd->flow_ucast6);
err |= rdma_rxq_destroy_flow (rd, &rd->flow_mcast4);
err |= rdma_rxq_destroy_flow (rd, &rd->flow_ucast4);
if (err) if (err)
return ~0; return ~0;
err = rdma_rxq_destroy_flow (rd, &rd->flow_ucast); rd->flow_ucast6 =
if (err) rdma_rxq_init_flow (rd, rd->rx_qp6, &rd->hwaddr, &ucast,
return ~0; ntohs (ETH_P_IPV6), 0);
rd->flow_mcast6 =
/* receive only packets with src = our MAC */ rdma_rxq_init_flow (rd, rd->rx_qp6, &mcast, &mcast, ntohs (ETH_P_IPV6),
rd->flow_ucast = rdma_rxq_init_flow (rd, rd->rx_qp, &rd->hwaddr, &ucast, 0); IBV_FLOW_ATTR_FLAGS_DONT_TRAP
if (!rd->flow_ucast) /* let others receive mcast packet too (eg. Linux) */
return ~0;
/* receive multicast packets */
rd->flow_mcast = rdma_rxq_init_flow (rd, rd->rx_qp, &mcast, &mcast,
IBV_FLOW_ATTR_FLAGS_DONT_TRAP
/* let others receive mcast packet too (eg. Linux) */
); );
if (!rd->flow_mcast) rd->flow_ucast4 =
rdma_rxq_init_flow (rd, rd->rx_qp4, &rd->hwaddr, &ucast, 0, 0);
rd->flow_mcast4 =
rdma_rxq_init_flow (rd, rd->rx_qp4, &mcast, &mcast, 0,
IBV_FLOW_ATTR_FLAGS_DONT_TRAP
/* let others receive mcast packet too (eg. Linux) */
);
if (!rd->flow_ucast6 || !rd->flow_mcast6 || !rd->flow_ucast4
|| !rd->flow_mcast4)
return ~0; return ~0;
rd->flags &= ~RDMA_DEVICE_F_PROMISC; rd->flags &= ~RDMA_DEVICE_F_PROMISC;
@ -375,8 +387,10 @@ rdma_dev_cleanup (rdma_device_t * rd)
rdma_log (VLIB_LOG_LEVEL_DEBUG, rd, #fn "() failed (rv = %d)", rv); \ rdma_log (VLIB_LOG_LEVEL_DEBUG, rd, #fn "() failed (rv = %d)", rv); \
} }
_(ibv_destroy_flow, rd->flow_mcast); _(ibv_destroy_flow, rd->flow_mcast6);
_(ibv_destroy_flow, rd->flow_ucast); _(ibv_destroy_flow, rd->flow_ucast6);
_(ibv_destroy_flow, rd->flow_mcast4);
_(ibv_destroy_flow, rd->flow_ucast4);
_(ibv_dereg_mr, rd->mr); _(ibv_dereg_mr, rd->mr);
vec_foreach (txq, rd->txqs) vec_foreach (txq, rd->txqs)
{ {
@ -389,7 +403,8 @@ rdma_dev_cleanup (rdma_device_t * rd)
_(ibv_destroy_cq, rxq->cq); _(ibv_destroy_cq, rxq->cq);
} }
_(ibv_destroy_rwq_ind_table, rd->rx_rwq_ind_tbl); _(ibv_destroy_rwq_ind_table, rd->rx_rwq_ind_tbl);
_(ibv_destroy_qp, rd->rx_qp); _(ibv_destroy_qp, rd->rx_qp6);
_(ibv_destroy_qp, rd->rx_qp4);
_(ibv_dealloc_pd, rd->pd); _(ibv_dealloc_pd, rd->pd);
_(ibv_close_device, rd->ctx); _(ibv_close_device, rd->ctx);
#undef _ #undef _
@ -523,10 +538,18 @@ rdma_rxq_finalize (vlib_main_t * vm, rdma_device_t * rd)
qpia.rx_hash_conf.rx_hash_key_len = sizeof (rdma_rss_hash_key); qpia.rx_hash_conf.rx_hash_key_len = sizeof (rdma_rss_hash_key);
qpia.rx_hash_conf.rx_hash_key = rdma_rss_hash_key; qpia.rx_hash_conf.rx_hash_key = rdma_rss_hash_key;
qpia.rx_hash_conf.rx_hash_function = IBV_RX_HASH_FUNC_TOEPLITZ; qpia.rx_hash_conf.rx_hash_function = IBV_RX_HASH_FUNC_TOEPLITZ;
qpia.rx_hash_conf.rx_hash_fields_mask = qpia.rx_hash_conf.rx_hash_fields_mask =
IBV_RX_HASH_SRC_IPV4 | IBV_RX_HASH_DST_IPV4; IBV_RX_HASH_SRC_IPV4 | IBV_RX_HASH_DST_IPV4 | IBV_RX_HASH_SRC_PORT_TCP |
if ((rd->rx_qp = ibv_create_qp_ex (rd->ctx, &qpia)) == 0) IBV_RX_HASH_DST_PORT_TCP;
return clib_error_return_unix (0, "Queue Pair create failed"); if ((rd->rx_qp4 = ibv_create_qp_ex (rd->ctx, &qpia)) == 0)
return clib_error_return_unix (0, "IPv4 Queue Pair create failed");
qpia.rx_hash_conf.rx_hash_fields_mask =
IBV_RX_HASH_SRC_IPV6 | IBV_RX_HASH_DST_IPV6 | IBV_RX_HASH_SRC_PORT_TCP |
IBV_RX_HASH_DST_PORT_TCP;
if ((rd->rx_qp6 = ibv_create_qp_ex (rd->ctx, &qpia)) == 0)
return clib_error_return_unix (0, "IPv6 Queue Pair create failed");
if (rdma_dev_set_ucast (rd)) if (rdma_dev_set_ucast (rd))
return clib_error_return_unix (0, "Set unicast mode failed"); return clib_error_return_unix (0, "Set unicast mode failed");

9
src/plugins/rdma/rdma.h
View File

@ -172,10 +172,13 @@ typedef struct
struct ibv_context *ctx; struct ibv_context *ctx;
struct ibv_pd *pd; struct ibv_pd *pd;
struct ibv_mr *mr; struct ibv_mr *mr;
struct ibv_qp *rx_qp; struct ibv_qp *rx_qp4;
struct ibv_qp *rx_qp6;
struct ibv_rwq_ind_table *rx_rwq_ind_tbl; struct ibv_rwq_ind_table *rx_rwq_ind_tbl;
struct ibv_flow *flow_ucast; struct ibv_flow *flow_ucast4;
struct ibv_flow *flow_mcast; struct ibv_flow *flow_mcast4;
struct ibv_flow *flow_ucast6;
struct ibv_flow *flow_mcast6;
clib_error_t *error; clib_error_t *error;
} rdma_device_t; } rdma_device_t;