#include <stdint.h>
#include <string.h>
-/* Verbs header. */
-/* ISO C doesn't support unnamed structs/unions, disabling -pedantic. */
-#ifdef PEDANTIC
-#pragma GCC diagnostic ignored "-Wpedantic"
-#endif
-#include <infiniband/verbs.h>
-#ifdef PEDANTIC
-#pragma GCC diagnostic error "-Wpedantic"
-#endif
-
#include <rte_common.h>
#include <rte_ether.h>
#include <rte_ethdev_driver.h>
#include <mlx5_glue.h>
#include <mlx5_prm.h>
+#include <mlx5_malloc.h>
#include "mlx5_defs.h"
#include "mlx5.h"
#define VERBS_SPEC_INNER(item_flags) \
(!!((item_flags) & MLX5_FLOW_LAYER_TUNNEL) ? IBV_FLOW_SPEC_INNER : 0)
+/* Map of Verbs to Flow priority with 8 Verbs priorities. */
+static const uint32_t priority_map_3[][MLX5_PRIORITY_MAP_MAX] = {
+ { 0, 1, 2 }, { 2, 3, 4 }, { 5, 6, 7 },
+};
+
+/* Map of Verbs to Flow priority with 16 Verbs priorities. */
+static const uint32_t priority_map_5[][MLX5_PRIORITY_MAP_MAX] = {
+ { 0, 1, 2 }, { 3, 4, 5 }, { 6, 7, 8 },
+ { 9, 10, 11 }, { 12, 13, 14 },
+};
+
+/**
+ * Discover the maximum number of priority available.
+ *
+ * @param[in] dev
+ * Pointer to the Ethernet device structure.
+ *
+ * @return
+ * number of supported flow priority on success, a negative errno
+ * value otherwise and rte_errno is set.
+ */
+int
+mlx5_flow_discover_priorities(struct rte_eth_dev *dev)
+{
+ struct mlx5_priv *priv = dev->data->dev_private;
+ struct {
+ struct ibv_flow_attr attr;
+ struct ibv_flow_spec_eth eth;
+ struct ibv_flow_spec_action_drop drop;
+ } flow_attr = {
+ .attr = {
+ .num_of_specs = 2,
+ .port = (uint8_t)priv->dev_port,
+ },
+ .eth = {
+ .type = IBV_FLOW_SPEC_ETH,
+ .size = sizeof(struct ibv_flow_spec_eth),
+ },
+ .drop = {
+ .size = sizeof(struct ibv_flow_spec_action_drop),
+ .type = IBV_FLOW_SPEC_ACTION_DROP,
+ },
+ };
+ struct ibv_flow *flow;
+ struct mlx5_hrxq *drop = mlx5_hrxq_drop_new(dev);
+ uint16_t vprio[] = { 8, 16 };
+ int i;
+ int priority = 0;
+
+ if (!drop) {
+ rte_errno = ENOTSUP;
+ return -rte_errno;
+ }
+ for (i = 0; i != RTE_DIM(vprio); i++) {
+ flow_attr.attr.priority = vprio[i] - 1;
+ flow = mlx5_glue->create_flow(drop->qp, &flow_attr.attr);
+ if (!flow)
+ break;
+ claim_zero(mlx5_glue->destroy_flow(flow));
+ priority = vprio[i];
+ }
+ mlx5_hrxq_drop_release(dev);
+ switch (priority) {
+ case 8:
+ priority = RTE_DIM(priority_map_3);
+ break;
+ case 16:
+ priority = RTE_DIM(priority_map_5);
+ break;
+ default:
+ rte_errno = ENOTSUP;
+ DRV_LOG(ERR,
+ "port %u verbs maximum priority: %d expected 8/16",
+ dev->data->port_id, priority);
+ return -rte_errno;
+ }
+ DRV_LOG(INFO, "port %u flow maximum priority: %d",
+ dev->data->port_id, priority);
+ return priority;
+}
+
+/**
+ * Adjust flow priority based on the highest layer and the request priority.
+ *
+ * @param[in] dev
+ * Pointer to the Ethernet device structure.
+ * @param[in] priority
+ * The rule base priority.
+ * @param[in] subpriority
+ * The priority based on the items.
+ *
+ * @return
+ * The new priority.
+ */
+uint32_t
+mlx5_flow_adjust_priority(struct rte_eth_dev *dev, int32_t priority,
+ uint32_t subpriority)
+{
+ uint32_t res = 0;
+ struct mlx5_priv *priv = dev->data->dev_private;
+
+ switch (priv->config.flow_prio) {
+ case RTE_DIM(priority_map_3):
+ res = priority_map_3[priority][subpriority];
+ break;
+ case RTE_DIM(priority_map_5):
+ res = priority_map_5[priority][subpriority];
+ break;
+ }
+ return res;
+}
+
/**
* Get Verbs flow counter by index.
*
MLX5_ASSERT(pool);
if (ppool)
*ppool = pool;
- return &pool->counters_raw[idx % MLX5_COUNTERS_PER_POOL];
+ return MLX5_POOL_GET_CNT(pool, idx % MLX5_COUNTERS_PER_POOL);
}
/**
pool = cont->pools[pool_idx];
if (!pool)
continue;
- cnt = TAILQ_FIRST(&pool->counters);
+ cnt = TAILQ_FIRST(&pool->counters[0]);
if (cnt)
break;
}
/* Resize the container pool array. */
size = sizeof(struct mlx5_flow_counter_pool *) *
(n_valid + MLX5_CNT_CONTAINER_RESIZE);
- pools = rte_zmalloc(__func__, size, 0);
+ pools = mlx5_malloc(MLX5_MEM_ZERO, size, 0,
+ SOCKET_ID_ANY);
if (!pools)
return 0;
if (n_valid) {
memcpy(pools, cont->pools,
sizeof(struct mlx5_flow_counter_pool *) *
n_valid);
- rte_free(cont->pools);
+ mlx5_free(cont->pools);
}
cont->pools = pools;
cont->n += MLX5_CNT_CONTAINER_RESIZE;
}
/* Allocate memory for new pool*/
- size = sizeof(*pool) + sizeof(*cnt_ext) *
+ size = sizeof(*pool) + (sizeof(*cnt_ext) + sizeof(*cnt)) *
MLX5_COUNTERS_PER_POOL;
- pool = rte_calloc(__func__, 1, size, 0);
+ pool = mlx5_malloc(MLX5_MEM_ZERO, size, 0, SOCKET_ID_ANY);
if (!pool)
return 0;
+ pool->type |= CNT_POOL_TYPE_EXT;
for (i = 0; i < MLX5_COUNTERS_PER_POOL; ++i) {
- cnt = &pool->counters_raw[i];
- TAILQ_INSERT_HEAD(&pool->counters, cnt, next);
+ cnt = MLX5_POOL_GET_CNT(pool, i);
+ TAILQ_INSERT_HEAD(&pool->counters[0], cnt, next);
}
- cnt = &pool->counters_raw[0];
+ cnt = MLX5_POOL_GET_CNT(pool, 0);
cont->pools[n_valid] = pool;
pool_idx = n_valid;
rte_atomic16_add(&cont->n_valid, 1);
TAILQ_INSERT_HEAD(&cont->pool_list, pool, next);
}
- i = cnt - pool->counters_raw;
+ i = MLX5_CNT_ARRAY_IDX(pool, cnt);
cnt_ext = MLX5_GET_POOL_CNT_EXT(pool, i);
cnt_ext->id = id;
cnt_ext->shared = shared;
/* Create counter with Verbs. */
ret = flow_verbs_counter_create(dev, cnt_ext);
if (!ret) {
- TAILQ_REMOVE(&pool->counters, cnt, next);
+ TAILQ_REMOVE(&pool->counters[0], cnt, next);
return MLX5_MAKE_CNT_IDX(pool_idx, i);
}
/* Some error occurred in Verbs library. */
claim_zero(mlx5_glue->destroy_counters(cnt_ext->cs));
cnt_ext->cs = NULL;
#endif
- TAILQ_INSERT_HEAD(&pool->counters, cnt, next);
+ TAILQ_INSERT_HEAD(&pool->counters[0], cnt, next);
}
}
struct mlx5_flow_counter *cnt = flow_verbs_counter_get_by_idx
(dev, flow->counter, &pool);
struct mlx5_flow_counter_ext *cnt_ext = MLX5_CNT_TO_CNT_EXT
- (pool, cnt);
+ (pool, cnt);
struct rte_flow_query_count *qc = data;
uint64_t counters[2] = {0, 0};
#if defined(HAVE_IBV_DEVICE_COUNTERS_SET_V42)
udp.val.src_port &= udp.mask.src_port;
udp.val.dst_port &= udp.mask.dst_port;
}
+ item++;
+ while (item->type == RTE_FLOW_ITEM_TYPE_VOID)
+ item++;
+ if (!(udp.val.dst_port & udp.mask.dst_port)) {
+ switch ((item)->type) {
+ case RTE_FLOW_ITEM_TYPE_VXLAN:
+ udp.val.dst_port = htons(MLX5_UDP_PORT_VXLAN);
+ udp.mask.dst_port = 0xffff;
+ break;
+ case RTE_FLOW_ITEM_TYPE_VXLAN_GPE:
+ udp.val.dst_port = htons(MLX5_UDP_PORT_VXLAN_GPE);
+ udp.mask.dst_port = 0xffff;
+ break;
+ case RTE_FLOW_ITEM_TYPE_MPLS:
+ udp.val.dst_port = htons(MLX5_UDP_PORT_MPLS);
+ udp.mask.dst_port = 0xffff;
+ break;
+ default:
+ break;
+ }
+ }
+
flow_verbs_spec_add(&dev_flow->verbs, &udp, size);
}
* the input is valid and that there is space to insert the requested action
* into the flow.
*
- * @param[in] dev_flow
- * Pointer to mlx5_flow.
+ * @param[in] rss_desc
+ * Pointer to mlx5_flow_rss_desc.
* @param[in] action
* Action configuration.
*/
static void
-flow_verbs_translate_action_queue(struct mlx5_flow *dev_flow,
+flow_verbs_translate_action_queue(struct mlx5_flow_rss_desc *rss_desc,
const struct rte_flow_action *action)
{
const struct rte_flow_action_queue *queue = action->conf;
- struct rte_flow *flow = dev_flow->flow;
- if (flow->rss.queue)
- (*flow->rss.queue)[0] = queue->index;
- flow->rss.queue_num = 1;
+ rss_desc->queue[0] = queue->index;
+ rss_desc->queue_num = 1;
}
/**
* the input is valid and that there is space to insert the requested action
* into the flow.
*
+ * @param[in] rss_desc
+ * Pointer to mlx5_flow_rss_desc.
* @param[in] action
* Action configuration.
- * @param[in, out] action_flags
- * Pointer to the detected actions.
- * @param[in] dev_flow
- * Pointer to mlx5_flow.
*/
static void
-flow_verbs_translate_action_rss(struct mlx5_flow *dev_flow,
+flow_verbs_translate_action_rss(struct mlx5_flow_rss_desc *rss_desc,
const struct rte_flow_action *action)
{
const struct rte_flow_action_rss *rss = action->conf;
const uint8_t *rss_key;
- struct rte_flow *flow = dev_flow->flow;
- if (flow->rss.queue)
- memcpy((*flow->rss.queue), rss->queue,
- rss->queue_num * sizeof(uint16_t));
- flow->rss.queue_num = rss->queue_num;
+ memcpy(rss_desc->queue, rss->queue, rss->queue_num * sizeof(uint16_t));
+ rss_desc->queue_num = rss->queue_num;
/* NULL RSS key indicates default RSS key. */
rss_key = !rss->key ? rss_hash_default_key : rss->key;
- memcpy(flow->rss.key, rss_key, MLX5_RSS_HASH_KEY_LEN);
+ memcpy(rss_desc->key, rss_key, MLX5_RSS_HASH_KEY_LEN);
/*
* rss->level and rss.types should be set in advance when expanding
* items for RSS.
* Pointer to the list of actions.
* @param[in] external
* This flow rule is created by request external to PMD.
+ * @param[in] hairpin
+ * Number of hairpin TX actions, 0 means classic flow.
* @param[out] error
* Pointer to the error structure.
*
const struct rte_flow_item items[],
const struct rte_flow_action actions[],
bool external __rte_unused,
+ int hairpin __rte_unused,
struct rte_flow_error *error)
{
int ret;
dev_flow->verbs.size = 0;
dev_flow->verbs.attr.num_of_specs = 0;
dev_flow->ingress = attr->ingress;
+ dev_flow->hash_fields = 0;
/* Need to set transfer attribute: not supported in Verbs mode. */
return dev_flow;
}
uint64_t priority = attr->priority;
uint32_t subpriority = 0;
struct mlx5_priv *priv = dev->data->dev_private;
+ struct mlx5_flow_rss_desc *rss_desc = &((struct mlx5_flow_rss_desc *)
+ priv->rss_desc)
+ [!!priv->flow_nested_idx];
if (priority == MLX5_FLOW_PRIO_RSVD)
priority = priv->config.flow_prio - 1;
case RTE_FLOW_ACTION_TYPE_FLAG:
flow_verbs_translate_action_flag(dev_flow, actions);
action_flags |= MLX5_FLOW_ACTION_FLAG;
+ dev_flow->handle->mark = 1;
break;
case RTE_FLOW_ACTION_TYPE_MARK:
flow_verbs_translate_action_mark(dev_flow, actions);
action_flags |= MLX5_FLOW_ACTION_MARK;
+ dev_flow->handle->mark = 1;
break;
case RTE_FLOW_ACTION_TYPE_DROP:
flow_verbs_translate_action_drop(dev_flow, actions);
action_flags |= MLX5_FLOW_ACTION_DROP;
+ dev_flow->handle->fate_action = MLX5_FLOW_FATE_DROP;
break;
case RTE_FLOW_ACTION_TYPE_QUEUE:
- flow_verbs_translate_action_queue(dev_flow, actions);
+ flow_verbs_translate_action_queue(rss_desc, actions);
action_flags |= MLX5_FLOW_ACTION_QUEUE;
+ dev_flow->handle->fate_action = MLX5_FLOW_FATE_QUEUE;
break;
case RTE_FLOW_ACTION_TYPE_RSS:
- flow_verbs_translate_action_rss(dev_flow, actions);
+ flow_verbs_translate_action_rss(rss_desc, actions);
action_flags |= MLX5_FLOW_ACTION_RSS;
+ dev_flow->handle->fate_action = MLX5_FLOW_FATE_QUEUE;
break;
case RTE_FLOW_ACTION_TYPE_COUNT:
ret = flow_verbs_translate_action_count(dev_flow,
"action not supported");
}
}
- dev_flow->handle->act_flags = action_flags;
+ dev_flow->act_flags = action_flags;
for (; items->type != RTE_FLOW_ITEM_TYPE_END; items++) {
int tunnel = !!(item_flags & MLX5_FLOW_LAYER_TUNNEL);
subpriority = MLX5_PRIORITY_MAP_L3;
dev_flow->hash_fields |=
mlx5_flow_hashfields_adjust
- (dev_flow, tunnel,
+ (rss_desc, tunnel,
MLX5_IPV4_LAYER_TYPES,
MLX5_IPV4_IBV_RX_HASH);
item_flags |= tunnel ? MLX5_FLOW_LAYER_INNER_L3_IPV4 :
subpriority = MLX5_PRIORITY_MAP_L3;
dev_flow->hash_fields |=
mlx5_flow_hashfields_adjust
- (dev_flow, tunnel,
+ (rss_desc, tunnel,
MLX5_IPV6_LAYER_TYPES,
MLX5_IPV6_IBV_RX_HASH);
item_flags |= tunnel ? MLX5_FLOW_LAYER_INNER_L3_IPV6 :
subpriority = MLX5_PRIORITY_MAP_L4;
dev_flow->hash_fields |=
mlx5_flow_hashfields_adjust
- (dev_flow, tunnel, ETH_RSS_TCP,
+ (rss_desc, tunnel, ETH_RSS_TCP,
(IBV_RX_HASH_SRC_PORT_TCP |
IBV_RX_HASH_DST_PORT_TCP));
item_flags |= tunnel ? MLX5_FLOW_LAYER_INNER_L4_TCP :
subpriority = MLX5_PRIORITY_MAP_L4;
dev_flow->hash_fields |=
mlx5_flow_hashfields_adjust
- (dev_flow, tunnel, ETH_RSS_UDP,
+ (rss_desc, tunnel, ETH_RSS_UDP,
(IBV_RX_HASH_SRC_PORT_UDP |
IBV_RX_HASH_DST_PORT_UDP));
item_flags |= tunnel ? MLX5_FLOW_LAYER_INNER_L4_UDP :
case RTE_FLOW_ITEM_TYPE_VXLAN:
flow_verbs_translate_item_vxlan(dev_flow, items,
item_flags);
- subpriority = MLX5_PRIORITY_MAP_L2;
+ subpriority = MLX5_TUNNEL_PRIO_GET(rss_desc);
item_flags |= MLX5_FLOW_LAYER_VXLAN;
break;
case RTE_FLOW_ITEM_TYPE_VXLAN_GPE:
flow_verbs_translate_item_vxlan_gpe(dev_flow, items,
item_flags);
- subpriority = MLX5_PRIORITY_MAP_L2;
+ subpriority = MLX5_TUNNEL_PRIO_GET(rss_desc);
item_flags |= MLX5_FLOW_LAYER_VXLAN_GPE;
break;
case RTE_FLOW_ITEM_TYPE_GRE:
flow_verbs_translate_item_gre(dev_flow, items,
item_flags);
- subpriority = MLX5_PRIORITY_MAP_L2;
+ subpriority = MLX5_TUNNEL_PRIO_GET(rss_desc);
item_flags |= MLX5_FLOW_LAYER_GRE;
break;
case RTE_FLOW_ITEM_TYPE_MPLS:
flow_verbs_translate_item_mpls(dev_flow, items,
item_flags);
- subpriority = MLX5_PRIORITY_MAP_L2;
+ subpriority = MLX5_TUNNEL_PRIO_GET(rss_desc);
item_flags |= MLX5_FLOW_LAYER_MPLS;
break;
default:
/* Other members of attr will be ignored. */
dev_flow->verbs.attr.priority =
mlx5_flow_adjust_priority(dev, priority, subpriority);
- dev_flow->verbs.attr.port = (uint8_t)priv->ibv_port;
+ dev_flow->verbs.attr.port = (uint8_t)priv->dev_port;
return 0;
}
return;
SILIST_FOREACH(priv->sh->ipool[MLX5_IPOOL_MLX5_FLOW], flow->dev_handles,
handle_idx, handle, next) {
- if (handle->ib_flow) {
- claim_zero(mlx5_glue->destroy_flow(handle->ib_flow));
- handle->ib_flow = NULL;
+ if (handle->drv_flow) {
+ claim_zero(mlx5_glue->destroy_flow(handle->drv_flow));
+ handle->drv_flow = NULL;
}
/* hrxq is union, don't touch it only the flag is set. */
- if (handle->hrxq) {
- if (handle->act_flags & MLX5_FLOW_ACTION_DROP) {
+ if (handle->rix_hrxq) {
+ if (handle->fate_action == MLX5_FLOW_FATE_DROP) {
mlx5_hrxq_drop_release(dev);
- handle->hrxq = 0;
- } else if (handle->act_flags &
- (MLX5_FLOW_ACTION_QUEUE |
- MLX5_FLOW_ACTION_RSS)) {
- mlx5_hrxq_release(dev, handle->hrxq);
- handle->hrxq = 0;
+ handle->rix_hrxq = 0;
+ } else if (handle->fate_action ==
+ MLX5_FLOW_FATE_QUEUE) {
+ mlx5_hrxq_release(dev, handle->rix_hrxq);
+ handle->rix_hrxq = 0;
}
}
if (handle->vf_vlan.tag && handle->vf_vlan.created)
for (idx = priv->flow_idx - 1; idx >= priv->flow_nested_idx; idx--) {
dev_flow = &((struct mlx5_flow *)priv->inter_flows)[idx];
handle = dev_flow->handle;
- if (handle->act_flags & MLX5_FLOW_ACTION_DROP) {
+ if (handle->fate_action == MLX5_FLOW_FATE_DROP) {
hrxq = mlx5_hrxq_drop_new(dev);
if (!hrxq) {
rte_flow_error_set
}
} else {
uint32_t hrxq_idx;
+ struct mlx5_flow_rss_desc *rss_desc =
+ &((struct mlx5_flow_rss_desc *)priv->rss_desc)
+ [!!priv->flow_nested_idx];
- MLX5_ASSERT(flow->rss.queue);
- hrxq_idx = mlx5_hrxq_get(dev, flow->rss.key,
+ MLX5_ASSERT(rss_desc->queue_num);
+ hrxq_idx = mlx5_hrxq_get(dev, rss_desc->key,
MLX5_RSS_HASH_KEY_LEN,
dev_flow->hash_fields,
- (*flow->rss.queue),
- flow->rss.queue_num);
+ rss_desc->queue,
+ rss_desc->queue_num);
if (!hrxq_idx)
- hrxq_idx = mlx5_hrxq_new(dev, flow->rss.key,
+ hrxq_idx = mlx5_hrxq_new(dev, rss_desc->key,
MLX5_RSS_HASH_KEY_LEN,
dev_flow->hash_fields,
- (*flow->rss.queue),
- flow->rss.queue_num,
+ rss_desc->queue,
+ rss_desc->queue_num,
!!(handle->layers &
MLX5_FLOW_LAYER_TUNNEL));
hrxq = mlx5_ipool_get(priv->sh->ipool[MLX5_IPOOL_HRXQ],
"cannot get hash queue");
goto error;
}
- handle->hrxq = hrxq_idx;
+ handle->rix_hrxq = hrxq_idx;
}
MLX5_ASSERT(hrxq);
- handle->ib_flow = mlx5_glue->create_flow(hrxq->qp,
- &dev_flow->verbs.attr);
- if (!handle->ib_flow) {
+ handle->drv_flow = mlx5_glue->create_flow
+ (hrxq->qp, &dev_flow->verbs.attr);
+ if (!handle->drv_flow) {
rte_flow_error_set(error, errno,
RTE_FLOW_ERROR_TYPE_UNSPECIFIED,
NULL,
SILIST_FOREACH(priv->sh->ipool[MLX5_IPOOL_MLX5_FLOW], flow->dev_handles,
dev_handles, handle, next) {
/* hrxq is union, don't touch it only the flag is set. */
- if (handle->hrxq) {
- if (handle->act_flags & MLX5_FLOW_ACTION_DROP) {
+ if (handle->rix_hrxq) {
+ if (handle->fate_action == MLX5_FLOW_FATE_DROP) {
mlx5_hrxq_drop_release(dev);
- handle->hrxq = 0;
- } else if (handle->act_flags &
- (MLX5_FLOW_ACTION_QUEUE |
- MLX5_FLOW_ACTION_RSS)) {
- mlx5_hrxq_release(dev, handle->hrxq);
- handle->hrxq = 0;
+ handle->rix_hrxq = 0;
+ } else if (handle->fate_action ==
+ MLX5_FLOW_FATE_QUEUE) {
+ mlx5_hrxq_release(dev, handle->rix_hrxq);
+ handle->rix_hrxq = 0;
}
}
if (handle->vf_vlan.tag && handle->vf_vlan.created)