X-Git-Url: http://git.droids-corp.org/?a=blobdiff_plain;f=drivers%2Fnet%2Fmlx5%2Fmlx5_flow.c;h=c347f8130e114166974367881cf3c5edc2d8817b;hb=08c2772fc7476efaed612b41e67eb43d604a68e6;hp=1e6ed629c55a160bb3fb47cf7dc87e43a7eeb80c;hpb=eab3ca48580c1599da754c621c1a941be27d3e9c;p=dpdk.git diff --git a/drivers/net/mlx5/mlx5_flow.c b/drivers/net/mlx5/mlx5_flow.c index 1e6ed629c5..c347f8130e 100644 --- a/drivers/net/mlx5/mlx5_flow.c +++ b/drivers/net/mlx5/mlx5_flow.c @@ -12,7 +12,7 @@ #include #include -#include +#include #include #include #include @@ -73,7 +73,7 @@ const struct mlx5_flow_driver_ops mlx5_flow_null_drv_ops; const struct mlx5_flow_driver_ops *flow_drv_ops[] = { [MLX5_FLOW_TYPE_MIN] = &mlx5_flow_null_drv_ops, -#ifdef HAVE_IBV_FLOW_DV_SUPPORT +#if defined(HAVE_IBV_FLOW_DV_SUPPORT) || !defined(HAVE_INFINIBAND_VERBS_H) [MLX5_FLOW_TYPE_DV] = &mlx5_flow_dv_drv_ops, #endif [MLX5_FLOW_TYPE_VERBS] = &mlx5_flow_verbs_drv_ops, @@ -212,6 +212,8 @@ mlx5_flow_expand_rss_item_complete(const struct rte_flow_item *item) return ret; } +#define MLX5_RSS_EXP_ELT_N 8 + /** * Expand RSS flows into several possible flows according to the RSS hash * fields requested and the driver capabilities. @@ -242,13 +244,12 @@ mlx5_flow_expand_rss(struct mlx5_flow_expand_rss *buf, size_t size, const struct mlx5_flow_expand_node graph[], int graph_root_index) { - const int elt_n = 8; const struct rte_flow_item *item; const struct mlx5_flow_expand_node *node = &graph[graph_root_index]; const int *next_node; - const int *stack[elt_n]; + const int *stack[MLX5_RSS_EXP_ELT_N]; int stack_pos = 0; - struct rte_flow_item flow_items[elt_n]; + struct rte_flow_item flow_items[MLX5_RSS_EXP_ELT_N]; unsigned int i; size_t lsize; size_t user_pattern_size = 0; @@ -261,10 +262,10 @@ mlx5_flow_expand_rss(struct mlx5_flow_expand_rss *buf, size_t size, memset(&missed_item, 0, sizeof(missed_item)); lsize = offsetof(struct mlx5_flow_expand_rss, entry) + - elt_n * sizeof(buf->entry[0]); + MLX5_RSS_EXP_ELT_N * sizeof(buf->entry[0]); if (lsize <= size) { buf->entry[0].priority = 0; - buf->entry[0].pattern = (void *)&buf->entry[elt_n]; + buf->entry[0].pattern = (void *)&buf->entry[MLX5_RSS_EXP_ELT_N]; buf->entries = 0; addr = buf->entry[0].pattern; } @@ -367,7 +368,7 @@ mlx5_flow_expand_rss(struct mlx5_flow_expand_rss *buf, size_t size, /* Go deeper. */ if (node->next) { next_node = node->next; - if (stack_pos++ == elt_n) { + if (stack_pos++ == MLX5_RSS_EXP_ELT_N) { rte_errno = E2BIG; return -rte_errno; } @@ -498,7 +499,8 @@ static const struct mlx5_flow_expand_node mlx5_support_expansion[] = { (MLX5_EXPANSION_OUTER_IPV6_UDP, MLX5_EXPANSION_OUTER_IPV6_TCP, MLX5_EXPANSION_IPV4, - MLX5_EXPANSION_IPV6), + MLX5_EXPANSION_IPV6, + MLX5_EXPANSION_GRE), .type = RTE_FLOW_ITEM_TYPE_IPV6, .rss_types = ETH_RSS_IPV6 | ETH_RSS_FRAG_IPV6 | ETH_RSS_NONFRAG_IPV6_OTHER, @@ -526,7 +528,8 @@ static const struct mlx5_flow_expand_node mlx5_support_expansion[] = { .type = RTE_FLOW_ITEM_TYPE_VXLAN_GPE, }, [MLX5_EXPANSION_GRE] = { - .next = MLX5_FLOW_EXPAND_RSS_NEXT(MLX5_EXPANSION_IPV4), + .next = MLX5_FLOW_EXPAND_RSS_NEXT(MLX5_EXPANSION_IPV4, + MLX5_EXPANSION_IPV6), .type = RTE_FLOW_ITEM_TYPE_GRE, }, [MLX5_EXPANSION_MPLS] = { @@ -695,11 +698,6 @@ static struct mlx5_flow_tunnel_info tunnels_info[] = { }, }; -/* Key of thread specific flow workspace data. */ -static pthread_key_t key_workspace; - -/* Thread specific flow workspace data once initialization data. */ -static pthread_once_t key_workspace_init; /** @@ -777,6 +775,7 @@ mlx5_flow_get_reg_id(struct rte_eth_dev *dev, return priv->mtr_color_reg != REG_C_2 ? REG_C_2 : REG_C_3; case MLX5_MTR_COLOR: + case MLX5_ASO_FLOW_HIT: /* Both features use the same REG_C. */ MLX5_ASSERT(priv->mtr_color_reg != REG_NON); return priv->mtr_color_reg; case MLX5_COPY_MARK: @@ -796,7 +795,7 @@ mlx5_flow_get_reg_id(struct rte_eth_dev *dev, start_reg = priv->mtr_color_reg != REG_C_2 ? REG_C_2 : (priv->mtr_reg_share ? REG_C_3 : REG_C_4); skip_mtr_reg = !!(priv->mtr_en && start_reg == REG_C_2); - if (id > (REG_C_7 - start_reg)) + if (id > (uint32_t)(REG_C_7 - start_reg)) return rte_flow_error_set(error, EINVAL, RTE_FLOW_ERROR_TYPE_ITEM, NULL, "invalid tag id"); @@ -812,7 +811,7 @@ mlx5_flow_get_reg_id(struct rte_eth_dev *dev, */ if (skip_mtr_reg && config->flow_mreg_c [id + start_reg - REG_C_0] >= priv->mtr_color_reg) { - if (id >= (REG_C_7 - start_reg)) + if (id >= (uint32_t)(REG_C_7 - start_reg)) return rte_flow_error_set(error, EINVAL, RTE_FLOW_ERROR_TYPE_ITEM, NULL, "invalid tag id"); @@ -858,6 +857,58 @@ mlx5_flow_ext_mreg_supported(struct rte_eth_dev *dev) return config->flow_mreg_c[2] != REG_NON; } +/** + * Get the lowest priority. + * + * @param[in] dev + * Pointer to the Ethernet device structure. + * @param[in] attributes + * Pointer to device flow rule attributes. + * + * @return + * The value of lowest priority of flow. + */ +uint32_t +mlx5_get_lowest_priority(struct rte_eth_dev *dev, + const struct rte_flow_attr *attr) +{ + struct mlx5_priv *priv = dev->data->dev_private; + + if (!attr->group && !attr->transfer) + return priv->config.flow_prio - 2; + return MLX5_NON_ROOT_FLOW_MAX_PRIO - 1; +} + +/** + * Calculate matcher priority of the flow. + * + * @param[in] dev + * Pointer to the Ethernet device structure. + * @param[in] attr + * Pointer to device flow rule attributes. + * @param[in] subpriority + * The priority based on the items. + * @return + * The matcher priority of the flow. + */ +uint16_t +mlx5_get_matcher_priority(struct rte_eth_dev *dev, + const struct rte_flow_attr *attr, + uint32_t subpriority) +{ + uint16_t priority = (uint16_t)attr->priority; + struct mlx5_priv *priv = dev->data->dev_private; + + if (!attr->group && !attr->transfer) { + if (attr->priority == MLX5_FLOW_LOWEST_PRIO_INDICATOR) + priority = priv->config.flow_prio - 1; + return mlx5_os_flow_adjust_priority(dev, priority, subpriority); + } + if (attr->priority == MLX5_FLOW_LOWEST_PRIO_INDICATOR) + priority = MLX5_NON_ROOT_FLOW_MAX_PRIO; + return priority * 3 + subpriority; +} + /** * Verify the @p item specifications (spec, last, mask) are compatible with the * NIC capabilities. @@ -1000,17 +1051,29 @@ flow_drv_rxq_flags_set(struct rte_eth_dev *dev, struct mlx5_priv *priv = dev->data->dev_private; const int mark = dev_handle->mark; const int tunnel = !!(dev_handle->layers & MLX5_FLOW_LAYER_TUNNEL); - struct mlx5_hrxq *hrxq; + struct mlx5_ind_table_obj *ind_tbl = NULL; unsigned int i; - if (dev_handle->fate_action != MLX5_FLOW_FATE_QUEUE) - return; - hrxq = mlx5_ipool_get(priv->sh->ipool[MLX5_IPOOL_HRXQ], + if (dev_handle->fate_action == MLX5_FLOW_FATE_QUEUE) { + struct mlx5_hrxq *hrxq; + + hrxq = mlx5_ipool_get(priv->sh->ipool[MLX5_IPOOL_HRXQ], dev_handle->rix_hrxq); - if (!hrxq) + if (hrxq) + ind_tbl = hrxq->ind_table; + } else if (dev_handle->fate_action == MLX5_FLOW_FATE_SHARED_RSS) { + struct mlx5_shared_action_rss *shared_rss; + + shared_rss = mlx5_ipool_get + (priv->sh->ipool[MLX5_IPOOL_RSS_SHARED_ACTIONS], + dev_handle->rix_srss); + if (shared_rss) + ind_tbl = shared_rss->ind_tbl; + } + if (!ind_tbl) return; - for (i = 0; i != hrxq->ind_table->queues_n; ++i) { - int idx = hrxq->ind_table->queues[i]; + for (i = 0; i != ind_tbl->queues_n; ++i) { + int idx = ind_tbl->queues[i]; struct mlx5_rxq_ctrl *rxq_ctrl = container_of((*priv->rxqs)[idx], struct mlx5_rxq_ctrl, rxq); @@ -1082,18 +1145,30 @@ flow_drv_rxq_flags_trim(struct rte_eth_dev *dev, struct mlx5_priv *priv = dev->data->dev_private; const int mark = dev_handle->mark; const int tunnel = !!(dev_handle->layers & MLX5_FLOW_LAYER_TUNNEL); - struct mlx5_hrxq *hrxq; + struct mlx5_ind_table_obj *ind_tbl = NULL; unsigned int i; - if (dev_handle->fate_action != MLX5_FLOW_FATE_QUEUE) - return; - hrxq = mlx5_ipool_get(priv->sh->ipool[MLX5_IPOOL_HRXQ], + if (dev_handle->fate_action == MLX5_FLOW_FATE_QUEUE) { + struct mlx5_hrxq *hrxq; + + hrxq = mlx5_ipool_get(priv->sh->ipool[MLX5_IPOOL_HRXQ], dev_handle->rix_hrxq); - if (!hrxq) + if (hrxq) + ind_tbl = hrxq->ind_table; + } else if (dev_handle->fate_action == MLX5_FLOW_FATE_SHARED_RSS) { + struct mlx5_shared_action_rss *shared_rss; + + shared_rss = mlx5_ipool_get + (priv->sh->ipool[MLX5_IPOOL_RSS_SHARED_ACTIONS], + dev_handle->rix_srss); + if (shared_rss) + ind_tbl = shared_rss->ind_tbl; + } + if (!ind_tbl) return; MLX5_ASSERT(dev->data->dev_started); - for (i = 0; i != hrxq->ind_table->queues_n; ++i) { - int idx = hrxq->ind_table->queues[i]; + for (i = 0; i != ind_tbl->queues_n; ++i) { + int idx = ind_tbl->queues[i]; struct mlx5_rxq_ctrl *rxq_ctrl = container_of((*priv->rxqs)[idx], struct mlx5_rxq_ctrl, rxq); @@ -1194,10 +1269,14 @@ mlx5_flow_rxq_dynf_metadata_set(struct rte_eth_dev *dev) data->dynf_meta = 0; data->flow_meta_mask = 0; data->flow_meta_offset = -1; + data->flow_meta_port_mask = 0; } else { data->dynf_meta = 1; data->flow_meta_mask = rte_flow_dynf_metadata_mask; data->flow_meta_offset = rte_flow_dynf_metadata_offs; + data->flow_meta_port_mask = (uint32_t)~0; + if (priv->config.dv_xmeta_en == MLX5_XMETA_MODE_META16) + data->flow_meta_port_mask >>= 16; } } } @@ -1411,6 +1490,7 @@ mlx5_validate_action_rss(struct rte_eth_dev *dev, { struct mlx5_priv *priv = dev->data->dev_private; const struct rte_flow_action_rss *rss = action->conf; + enum mlx5_rxq_type rxq_type = MLX5_RXQ_TYPE_UNDEFINED; unsigned int i; if (rss->func != RTE_ETH_HASH_FUNCTION_DEFAULT && @@ -1476,6 +1556,8 @@ mlx5_validate_action_rss(struct rte_eth_dev *dev, RTE_FLOW_ERROR_TYPE_ACTION_CONF, NULL, "No queues configured"); for (i = 0; i != rss->queue_num; ++i) { + struct mlx5_rxq_ctrl *rxq_ctrl; + if (rss->queue[i] >= priv->rxqs_n) return rte_flow_error_set (error, EINVAL, @@ -1485,6 +1567,15 @@ mlx5_validate_action_rss(struct rte_eth_dev *dev, return rte_flow_error_set (error, EINVAL, RTE_FLOW_ERROR_TYPE_ACTION_CONF, &rss->queue[i], "queue is not configured"); + rxq_ctrl = container_of((*priv->rxqs)[rss->queue[i]], + struct mlx5_rxq_ctrl, rxq); + if (i == 0) + rxq_type = rxq_ctrl->type; + if (rxq_type != rxq_ctrl->type) + return rte_flow_error_set + (error, ENOTSUP, RTE_FLOW_ERROR_TYPE_ACTION_CONF, + &rss->queue[i], + "combining hairpin and regular RSS queues is not supported"); } return 0; } @@ -1636,7 +1727,7 @@ mlx5_flow_validate_attributes(struct rte_eth_dev *dev, return rte_flow_error_set(error, ENOTSUP, RTE_FLOW_ERROR_TYPE_ATTR_GROUP, NULL, "groups is not supported"); - if (attributes->priority != MLX5_FLOW_PRIO_RSVD && + if (attributes->priority != MLX5_FLOW_LOWEST_PRIO_INDICATOR && attributes->priority >= priority_max) return rte_flow_error_set(error, ENOTSUP, RTE_FLOW_ERROR_TYPE_ATTR_PRIORITY, @@ -2589,6 +2680,149 @@ mlx5_flow_validate_item_geneve(const struct rte_flow_item *item, return 0; } +/** + * Validate Geneve TLV option item. + * + * @param[in] item + * Item specification. + * @param[in] last_item + * Previous validated item in the pattern items. + * @param[in] geneve_item + * Previous GENEVE item specification. + * @param[in] dev + * Pointer to the rte_eth_dev structure. + * @param[out] error + * Pointer to error structure. + * + * @return + * 0 on success, a negative errno value otherwise and rte_errno is set. + */ +int +mlx5_flow_validate_item_geneve_opt(const struct rte_flow_item *item, + uint64_t last_item, + const struct rte_flow_item *geneve_item, + struct rte_eth_dev *dev, + struct rte_flow_error *error) +{ + struct mlx5_priv *priv = dev->data->dev_private; + struct mlx5_dev_ctx_shared *sh = priv->sh; + struct mlx5_geneve_tlv_option_resource *geneve_opt_resource; + struct mlx5_hca_attr *hca_attr = &priv->config.hca_attr; + uint8_t data_max_supported = + hca_attr->max_geneve_tlv_option_data_len * 4; + struct mlx5_dev_config *config = &priv->config; + const struct rte_flow_item_geneve *geneve_spec; + const struct rte_flow_item_geneve *geneve_mask; + const struct rte_flow_item_geneve_opt *spec = item->spec; + const struct rte_flow_item_geneve_opt *mask = item->mask; + unsigned int i; + unsigned int data_len; + uint8_t tlv_option_len; + uint16_t optlen_m, optlen_v; + const struct rte_flow_item_geneve_opt full_mask = { + .option_class = RTE_BE16(0xffff), + .option_type = 0xff, + .option_len = 0x1f, + }; + + if (!mask) + mask = &rte_flow_item_geneve_opt_mask; + if (!spec) + return rte_flow_error_set + (error, ENOTSUP, RTE_FLOW_ERROR_TYPE_ITEM, item, + "Geneve TLV opt class/type/length must be specified"); + if ((uint32_t)spec->option_len > MLX5_GENEVE_OPTLEN_MASK) + return rte_flow_error_set + (error, ENOTSUP, RTE_FLOW_ERROR_TYPE_ITEM, item, + "Geneve TLV opt length exceeeds the limit (31)"); + /* Check if class type and length masks are full. */ + if (full_mask.option_class != mask->option_class || + full_mask.option_type != mask->option_type || + full_mask.option_len != (mask->option_len & full_mask.option_len)) + return rte_flow_error_set + (error, ENOTSUP, RTE_FLOW_ERROR_TYPE_ITEM, item, + "Geneve TLV opt class/type/length masks must be full"); + /* Check if length is supported */ + if ((uint32_t)spec->option_len > + config->hca_attr.max_geneve_tlv_option_data_len) + return rte_flow_error_set + (error, ENOTSUP, RTE_FLOW_ERROR_TYPE_ITEM, item, + "Geneve TLV opt length not supported"); + if (config->hca_attr.max_geneve_tlv_options > 1) + DRV_LOG(DEBUG, + "max_geneve_tlv_options supports more than 1 option"); + /* Check GENEVE item preceding. */ + if (!geneve_item || !(last_item & MLX5_FLOW_LAYER_GENEVE)) + return rte_flow_error_set + (error, ENOTSUP, RTE_FLOW_ERROR_TYPE_ITEM, item, + "Geneve opt item must be preceded with Geneve item"); + geneve_spec = geneve_item->spec; + geneve_mask = geneve_item->mask ? geneve_item->mask : + &rte_flow_item_geneve_mask; + /* Check if GENEVE TLV option size doesn't exceed option length */ + if (geneve_spec && (geneve_mask->ver_opt_len_o_c_rsvd0 || + geneve_spec->ver_opt_len_o_c_rsvd0)) { + tlv_option_len = spec->option_len & mask->option_len; + optlen_v = rte_be_to_cpu_16(geneve_spec->ver_opt_len_o_c_rsvd0); + optlen_v = MLX5_GENEVE_OPTLEN_VAL(optlen_v); + optlen_m = rte_be_to_cpu_16(geneve_mask->ver_opt_len_o_c_rsvd0); + optlen_m = MLX5_GENEVE_OPTLEN_VAL(optlen_m); + if ((optlen_v & optlen_m) <= tlv_option_len) + return rte_flow_error_set + (error, ENOTSUP, RTE_FLOW_ERROR_TYPE_ITEM, item, + "GENEVE TLV option length exceeds optlen"); + } + /* Check if length is 0 or data is 0. */ + if (spec->data == NULL || spec->option_len == 0) + return rte_flow_error_set + (error, ENOTSUP, RTE_FLOW_ERROR_TYPE_ITEM, item, + "Geneve TLV opt with zero data/length not supported"); + /* Check not all data & mask are 0. */ + data_len = spec->option_len * 4; + if (mask->data == NULL) { + for (i = 0; i < data_len; i++) + if (spec->data[i]) + break; + if (i == data_len) + return rte_flow_error_set(error, ENOTSUP, + RTE_FLOW_ERROR_TYPE_ITEM, item, + "Can't match on Geneve option data 0"); + } else { + for (i = 0; i < data_len; i++) + if (spec->data[i] & mask->data[i]) + break; + if (i == data_len) + return rte_flow_error_set(error, ENOTSUP, + RTE_FLOW_ERROR_TYPE_ITEM, item, + "Can't match on Geneve option data and mask 0"); + /* Check data mask supported. */ + for (i = data_max_supported; i < data_len ; i++) + if (mask->data[i]) + return rte_flow_error_set(error, ENOTSUP, + RTE_FLOW_ERROR_TYPE_ITEM, item, + "Data mask is of unsupported size"); + } + /* Check GENEVE option is supported in NIC. */ + if (!config->hca_attr.geneve_tlv_opt) + return rte_flow_error_set + (error, ENOTSUP, RTE_FLOW_ERROR_TYPE_ITEM, item, + "Geneve TLV opt not supported"); + /* Check if we already have geneve option with different type/class. */ + rte_spinlock_lock(&sh->geneve_tlv_opt_sl); + geneve_opt_resource = sh->geneve_tlv_option_resource; + if (geneve_opt_resource != NULL) + if (geneve_opt_resource->option_class != spec->option_class || + geneve_opt_resource->option_type != spec->option_type || + geneve_opt_resource->length != spec->option_len) { + rte_spinlock_unlock(&sh->geneve_tlv_opt_sl); + return rte_flow_error_set(error, ENOTSUP, + RTE_FLOW_ERROR_TYPE_ITEM, item, + "Only one Geneve TLV option supported"); + } + rte_spinlock_unlock(&sh->geneve_tlv_opt_sl); + return 0; +} + /** * Validate MPLS item. * @@ -3136,16 +3370,28 @@ flow_drv_destroy(struct rte_eth_dev *dev, struct rte_flow *flow) static const struct rte_flow_action_rss* flow_get_rss_action(const struct rte_flow_action actions[]) { + const struct rte_flow_action_rss *rss = NULL; + for (; actions->type != RTE_FLOW_ACTION_TYPE_END; actions++) { switch (actions->type) { case RTE_FLOW_ACTION_TYPE_RSS: - return (const struct rte_flow_action_rss *) - actions->conf; + rss = actions->conf; + break; + case RTE_FLOW_ACTION_TYPE_SAMPLE: + { + const struct rte_flow_action_sample *sample = + actions->conf; + const struct rte_flow_action *act = sample->actions; + for (; act->type != RTE_FLOW_ACTION_TYPE_END; act++) + if (act->type == RTE_FLOW_ACTION_TYPE_RSS) + rss = act->conf; + break; + } default: break; } } - return NULL; + return rss; } /** @@ -3510,7 +3756,7 @@ flow_check_hairpin_split(struct rte_eth_dev *dev, if (queue == NULL) return 0; conf = mlx5_rxq_get_hairpin_conf(dev, queue->index); - if (conf != NULL && !!conf->tx_explicit) + if (conf == NULL || conf->tx_explicit != 0) return 0; queue_action = 1; action_n++; @@ -3520,7 +3766,7 @@ flow_check_hairpin_split(struct rte_eth_dev *dev, if (rss == NULL || rss->queue_num == 0) return 0; conf = mlx5_rxq_get_hairpin_conf(dev, rss->queue[0]); - if (conf != NULL && !!conf->tx_explicit) + if (conf == NULL || conf->tx_explicit != 0) return 0; queue_action = 1; action_n++; @@ -3535,9 +3781,7 @@ flow_check_hairpin_split(struct rte_eth_dev *dev, break; case RTE_FLOW_ACTION_TYPE_RAW_ENCAP: raw_encap = actions->conf; - if (raw_encap->size > - (sizeof(struct rte_flow_item_eth) + - sizeof(struct rte_flow_item_ipv4))) + if (raw_encap->size > MLX5_ENCAPSULATION_DECISION_SIZE) split++; action_n++; break; @@ -3563,6 +3807,17 @@ static void flow_list_destroy(struct rte_eth_dev *dev, uint32_t *list, uint32_t flow_idx); +int +flow_dv_mreg_match_cb(struct mlx5_hlist *list __rte_unused, + struct mlx5_hlist_entry *entry, + uint64_t key, void *cb_ctx __rte_unused) +{ + struct mlx5_flow_mreg_copy_resource *mcp_res = + container_of(entry, typeof(*mcp_res), hlist_ent); + + return mcp_res->mark_id != key; +} + struct mlx5_hlist_entry * flow_dv_mreg_create_cb(struct mlx5_hlist *list, uint64_t key, void *cb_ctx) @@ -3640,7 +3895,7 @@ flow_dv_mreg_create_cb(struct mlx5_hlist *list, uint64_t key, }; } else { /* Default rule, wildcard match. */ - attr.priority = MLX5_FLOW_PRIO_RSVD; + attr.priority = MLX5_FLOW_LOWEST_PRIO_INDICATOR; items[0] = (struct rte_flow_item){ .type = RTE_FLOW_ITEM_TYPE_END, }; @@ -3664,6 +3919,7 @@ flow_dv_mreg_create_cb(struct mlx5_hlist *list, uint64_t key, return NULL; } mcp_res->idx = idx; + mcp_res->mark_id = mark_id; /* * The copy Flows are not included in any list. There * ones are referenced from other Flows and can not @@ -3972,9 +4228,7 @@ flow_hairpin_split(struct rte_eth_dev *dev, break; case RTE_FLOW_ACTION_TYPE_RAW_ENCAP: raw_encap = actions->conf; - if (raw_encap->size > - (sizeof(struct rte_flow_item_eth) + - sizeof(struct rte_flow_item_ipv4))) { + if (raw_encap->size > MLX5_ENCAPSULATION_DECISION_SIZE) { memcpy(actions_tx, actions, sizeof(struct rte_flow_action)); actions_tx++; @@ -3987,9 +4241,7 @@ flow_hairpin_split(struct rte_eth_dev *dev, break; case RTE_FLOW_ACTION_TYPE_RAW_DECAP: raw_decap = actions->conf; - if (raw_decap->size < - (sizeof(struct rte_flow_item_eth) + - sizeof(struct rte_flow_item_ipv4))) { + if (raw_decap->size < MLX5_ENCAPSULATION_DECISION_SIZE) { memcpy(actions_tx, actions, sizeof(struct rte_flow_action)); actions_tx++; @@ -4446,6 +4698,8 @@ flow_mreg_tx_copy_prep(struct rte_eth_dev *dev, * Pointer to the position of the matched action if exists, otherwise is -1. * @param[out] qrss_action_pos * Pointer to the position of the Queue/RSS action if exists, otherwise is -1. + * @param[out] modify_after_mirror + * Pointer to the flag of modify action after FDB mirroring. * * @return * > 0 the total number of actions. @@ -4455,14 +4709,15 @@ static int flow_check_match_action(const struct rte_flow_action actions[], const struct rte_flow_attr *attr, enum rte_flow_action_type action, - int *match_action_pos, int *qrss_action_pos) + int *match_action_pos, int *qrss_action_pos, + int *modify_after_mirror) { const struct rte_flow_action_sample *sample; int actions_n = 0; - int jump_flag = 0; uint32_t ratio = 0; int sub_type = 0; int flag = 0; + int fdb_mirror = 0; *match_action_pos = -1; *qrss_action_pos = -1; @@ -4471,27 +4726,53 @@ flow_check_match_action(const struct rte_flow_action actions[], flag = 1; *match_action_pos = actions_n; } - if (actions->type == RTE_FLOW_ACTION_TYPE_QUEUE || - actions->type == RTE_FLOW_ACTION_TYPE_RSS) + switch (actions->type) { + case RTE_FLOW_ACTION_TYPE_QUEUE: + case RTE_FLOW_ACTION_TYPE_RSS: *qrss_action_pos = actions_n; - if (actions->type == RTE_FLOW_ACTION_TYPE_JUMP) - jump_flag = 1; - if (actions->type == RTE_FLOW_ACTION_TYPE_SAMPLE) { + break; + case RTE_FLOW_ACTION_TYPE_SAMPLE: sample = actions->conf; ratio = sample->ratio; sub_type = ((const struct rte_flow_action *) (sample->actions))->type; + if (ratio == 1 && attr->transfer) + fdb_mirror = 1; + break; + case RTE_FLOW_ACTION_TYPE_SET_MAC_SRC: + case RTE_FLOW_ACTION_TYPE_SET_MAC_DST: + case RTE_FLOW_ACTION_TYPE_SET_IPV4_SRC: + case RTE_FLOW_ACTION_TYPE_SET_IPV4_DST: + case RTE_FLOW_ACTION_TYPE_SET_IPV6_SRC: + case RTE_FLOW_ACTION_TYPE_SET_IPV6_DST: + case RTE_FLOW_ACTION_TYPE_SET_TP_SRC: + case RTE_FLOW_ACTION_TYPE_SET_TP_DST: + case RTE_FLOW_ACTION_TYPE_DEC_TTL: + case RTE_FLOW_ACTION_TYPE_SET_TTL: + case RTE_FLOW_ACTION_TYPE_INC_TCP_SEQ: + case RTE_FLOW_ACTION_TYPE_DEC_TCP_SEQ: + case RTE_FLOW_ACTION_TYPE_INC_TCP_ACK: + case RTE_FLOW_ACTION_TYPE_DEC_TCP_ACK: + case RTE_FLOW_ACTION_TYPE_SET_IPV4_DSCP: + case RTE_FLOW_ACTION_TYPE_SET_IPV6_DSCP: + case RTE_FLOW_ACTION_TYPE_FLAG: + case RTE_FLOW_ACTION_TYPE_MARK: + case RTE_FLOW_ACTION_TYPE_SET_META: + case RTE_FLOW_ACTION_TYPE_SET_TAG: + if (fdb_mirror) + *modify_after_mirror = 1; + break; + default: + break; } actions_n++; } - if (flag && action == RTE_FLOW_ACTION_TYPE_SAMPLE && attr->transfer) { - if (ratio == 1) { - /* JUMP Action not support for Mirroring; - * Mirroring support multi-destination; - */ - if (!jump_flag && sub_type != RTE_FLOW_ACTION_TYPE_END) - flag = 0; - } + if (flag && fdb_mirror && !*modify_after_mirror) { + /* FDB mirroring uses the destination array to implement + * instead of FLOW_SAMPLER object. + */ + if (sub_type != RTE_FLOW_ACTION_TYPE_END) + flag = 0; } /* Count RTE_FLOW_ACTION_TYPE_END. */ return flag ? actions_n + 1 : 0; @@ -4510,8 +4791,8 @@ flow_check_match_action(const struct rte_flow_action actions[], * * @param dev * Pointer to Ethernet device. - * @param[in] fdb_tx - * FDB egress flow flag. + * @param[in] add_tag + * Add extra tag action flag. * @param[out] sfx_items * Suffix flow match items (list terminated by the END pattern item). * @param[in] actions @@ -4526,6 +4807,8 @@ flow_check_match_action(const struct rte_flow_action actions[], * The sample action position. * @param[in] qrss_action_pos * The Queue/RSS action position. + * @param[in] jump_table + * Add extra jump action flag. * @param[out] error * Perform verbose error reporting if not NULL. * @@ -4535,7 +4818,7 @@ flow_check_match_action(const struct rte_flow_action actions[], */ static int flow_sample_split_prep(struct rte_eth_dev *dev, - uint32_t fdb_tx, + int add_tag, struct rte_flow_item sfx_items[], const struct rte_flow_action actions[], struct rte_flow_action actions_sfx[], @@ -4543,14 +4826,17 @@ flow_sample_split_prep(struct rte_eth_dev *dev, int actions_n, int sample_action_pos, int qrss_action_pos, + int jump_table, struct rte_flow_error *error) { struct mlx5_priv *priv = dev->data->dev_private; struct mlx5_rte_flow_action_set_tag *set_tag; struct mlx5_rte_flow_item_tag *tag_spec; struct mlx5_rte_flow_item_tag *tag_mask; + struct rte_flow_action_jump *jump_action; uint32_t tag_id = 0; int index; + int append_index = 0; int ret; if (sample_action_pos < 0) @@ -4558,9 +4844,37 @@ flow_sample_split_prep(struct rte_eth_dev *dev, RTE_FLOW_ERROR_TYPE_ACTION, NULL, "invalid position of sample " "action in list"); - if (!fdb_tx) { + /* Prepare the actions for prefix and suffix flow. */ + if (qrss_action_pos >= 0 && qrss_action_pos < sample_action_pos) { + index = qrss_action_pos; + /* Put the preceding the Queue/RSS action into prefix flow. */ + if (index != 0) + memcpy(actions_pre, actions, + sizeof(struct rte_flow_action) * index); + /* Put others preceding the sample action into prefix flow. */ + if (sample_action_pos > index + 1) + memcpy(actions_pre + index, actions + index + 1, + sizeof(struct rte_flow_action) * + (sample_action_pos - index - 1)); + index = sample_action_pos - 1; + /* Put Queue/RSS action into Suffix flow. */ + memcpy(actions_sfx, actions + qrss_action_pos, + sizeof(struct rte_flow_action)); + actions_sfx++; + } else { + index = sample_action_pos; + if (index != 0) + memcpy(actions_pre, actions, + sizeof(struct rte_flow_action) * index); + } + /* For CX5, add an extra tag action for NIC-RX and E-Switch ingress. + * For CX6DX and above, metadata registers Cx preserve their value, + * add an extra tag action for NIC-RX and E-Switch Domain. + */ + if (add_tag) { /* Prepare the prefix tag action. */ - set_tag = (void *)(actions_pre + actions_n + 1); + append_index++; + set_tag = (void *)(actions_pre + actions_n + append_index); ret = mlx5_flow_get_reg_id(dev, MLX5_APP_TAG, 0, error); if (ret < 0) return ret; @@ -4585,32 +4899,7 @@ flow_sample_split_prep(struct rte_eth_dev *dev, .type = (enum rte_flow_item_type) RTE_FLOW_ITEM_TYPE_END, }; - } - /* Prepare the actions for prefix and suffix flow. */ - if (qrss_action_pos >= 0 && qrss_action_pos < sample_action_pos) { - index = qrss_action_pos; - /* Put the preceding the Queue/RSS action into prefix flow. */ - if (index != 0) - memcpy(actions_pre, actions, - sizeof(struct rte_flow_action) * index); - /* Put others preceding the sample action into prefix flow. */ - if (sample_action_pos > index + 1) - memcpy(actions_pre + index, actions + index + 1, - sizeof(struct rte_flow_action) * - (sample_action_pos - index - 1)); - index = sample_action_pos - 1; - /* Put Queue/RSS action into Suffix flow. */ - memcpy(actions_sfx, actions + qrss_action_pos, - sizeof(struct rte_flow_action)); - actions_sfx++; - } else { - index = sample_action_pos; - if (index != 0) - memcpy(actions_pre, actions, - sizeof(struct rte_flow_action) * index); - } - /* Add the extra tag action for NIC-RX and E-Switch ingress. */ - if (!fdb_tx) { + /* Prepare the tag action in prefix subflow. */ actions_pre[index++] = (struct rte_flow_action){ .type = (enum rte_flow_action_type) @@ -4621,6 +4910,22 @@ flow_sample_split_prep(struct rte_eth_dev *dev, memcpy(actions_pre + index, actions + sample_action_pos, sizeof(struct rte_flow_action)); index += 1; + /* For the modify action after the sample action in E-Switch mirroring, + * Add the extra jump action in prefix subflow and jump into the next + * table, then do the modify action in the new table. + */ + if (jump_table) { + /* Prepare the prefix jump action. */ + append_index++; + jump_action = (void *)(actions_pre + actions_n + append_index); + jump_action->group = jump_table; + actions_pre[index++] = + (struct rte_flow_action){ + .type = (enum rte_flow_action_type) + RTE_FLOW_ACTION_TYPE_JUMP, + .conf = jump_action, + }; + } actions_pre[index] = (struct rte_flow_action){ .type = (enum rte_flow_action_type) RTE_FLOW_ACTION_TYPE_END, @@ -5015,7 +5320,6 @@ flow_create_split_sample(struct rte_eth_dev *dev, struct mlx5_flow_dv_sample_resource *sample_res; struct mlx5_flow_tbl_data_entry *sfx_tbl_data; struct mlx5_flow_tbl_resource *sfx_tbl; - union mlx5_flow_tbl_key sfx_table_key; #endif size_t act_size; size_t item_size; @@ -5024,12 +5328,17 @@ flow_create_split_sample(struct rte_eth_dev *dev, int actions_n = 0; int sample_action_pos; int qrss_action_pos; + int add_tag = 0; + int modify_after_mirror = 0; + uint16_t jump_table = 0; + const uint32_t next_ft_step = 1; int ret = 0; if (priv->sampler_en) actions_n = flow_check_match_action(actions, attr, RTE_FLOW_ACTION_TYPE_SAMPLE, - &sample_action_pos, &qrss_action_pos); + &sample_action_pos, &qrss_action_pos, + &modify_after_mirror); if (actions_n) { /* The prefix actions must includes sample, tag, end. */ act_size = sizeof(struct rte_flow_action) * (actions_n * 2 + 1) @@ -5045,19 +5354,31 @@ flow_create_split_sample(struct rte_eth_dev *dev, "sample flow"); /* The representor_id is -1 for uplink. */ fdb_tx = (attr->transfer && priv->representor_id != -1); - if (!fdb_tx) + /* + * When reg_c_preserve is set, metadata registers Cx preserve + * their value even through packet duplication. + */ + add_tag = (!fdb_tx || priv->config.hca_attr.reg_c_preserve); + if (add_tag) sfx_items = (struct rte_flow_item *)((char *)sfx_actions + act_size); + if (modify_after_mirror) + jump_table = attr->group * MLX5_FLOW_TABLE_FACTOR + + next_ft_step; pre_actions = sfx_actions + actions_n; - tag_id = flow_sample_split_prep(dev, fdb_tx, sfx_items, + tag_id = flow_sample_split_prep(dev, add_tag, sfx_items, actions, sfx_actions, pre_actions, actions_n, sample_action_pos, - qrss_action_pos, error); - if (tag_id < 0 || (!fdb_tx && !tag_id)) { + qrss_action_pos, jump_table, + error); + if (tag_id < 0 || (add_tag && !tag_id)) { ret = -rte_errno; goto exit; } + if (modify_after_mirror) + flow_split_info->skip_scale = + 1 << MLX5_SCALE_JUMP_FLOW_GROUP_BIT; /* Add the prefix subflow. */ ret = flow_create_split_inner(dev, flow, &dev_flow, attr, items, pre_actions, @@ -5068,24 +5389,30 @@ flow_create_split_sample(struct rte_eth_dev *dev, } dev_flow->handle->split_flow_id = tag_id; #ifdef HAVE_IBV_FLOW_DV_SUPPORT - /* Set the sfx group attr. */ - sample_res = (struct mlx5_flow_dv_sample_resource *) - dev_flow->dv.sample_res; - sfx_tbl = (struct mlx5_flow_tbl_resource *) - sample_res->normal_path_tbl; - sfx_tbl_data = container_of(sfx_tbl, - struct mlx5_flow_tbl_data_entry, tbl); - sfx_table_key.v64 = sfx_tbl_data->entry.key; - sfx_attr.group = sfx_attr.transfer ? - (sfx_table_key.table_id - 1) : - sfx_table_key.table_id; + if (!modify_after_mirror) { + /* Set the sfx group attr. */ + sample_res = (struct mlx5_flow_dv_sample_resource *) + dev_flow->dv.sample_res; + sfx_tbl = (struct mlx5_flow_tbl_resource *) + sample_res->normal_path_tbl; + sfx_tbl_data = container_of(sfx_tbl, + struct mlx5_flow_tbl_data_entry, + tbl); + sfx_attr.group = sfx_attr.transfer ? + (sfx_tbl_data->table_id - 1) : + sfx_tbl_data->table_id; + } else { + MLX5_ASSERT(attr->transfer); + sfx_attr.group = jump_table; + } flow_split_info->prefix_layers = flow_get_prefix_layer_flags(dev_flow); flow_split_info->prefix_mark = dev_flow->handle->mark; /* Suffix group level already be scaled with factor, set - * skip_scale to 1 to avoid scale again in translation. + * MLX5_SCALE_FLOW_GROUP_BIT of skip_scale to 1 to avoid scale + * again in translation. */ - flow_split_info->skip_scale = 1; + flow_split_info->skip_scale = 1 << MLX5_SCALE_FLOW_GROUP_BIT; #endif } /* Add the suffix subflow. */ @@ -5236,7 +5563,7 @@ flow_list_create(struct rte_eth_dev *dev, uint32_t *list, struct mlx5_priv *priv = dev->data->dev_private; struct rte_flow *flow = NULL; struct mlx5_flow *dev_flow; - const struct rte_flow_action_rss *rss; + const struct rte_flow_action_rss *rss = NULL; struct mlx5_translated_shared_action shared_actions[MLX5_MAX_SHARED_ACTIONS]; int shared_actions_n = MLX5_MAX_SHARED_ACTIONS; @@ -5314,7 +5641,9 @@ flow_list_create(struct rte_eth_dev *dev, uint32_t *list, MLX5_ASSERT(flow->drv_type > MLX5_FLOW_TYPE_MIN && flow->drv_type < MLX5_FLOW_TYPE_MAX); memset(rss_desc, 0, offsetof(struct mlx5_flow_rss_desc, queue)); - rss = flow_get_rss_action(p_actions_rx); + /* RSS Action only works on NIC RX domain */ + if (attr->ingress && !attr->transfer) + rss = flow_get_rss_action(p_actions_rx); if (rss) { if (flow_rss_workspace_adjust(wks, rss_desc, rss->queue_num)) return 0; @@ -5414,7 +5743,7 @@ flow_list_create(struct rte_eth_dev *dev, uint32_t *list, */ if (external || dev->data->dev_started || (attr->group == MLX5_FLOW_MREG_CP_TABLE_GROUP && - attr->priority == MLX5_FLOW_PRIO_RSVD)) { + attr->priority == MLX5_FLOW_LOWEST_PRIO_INDICATOR)) { ret = flow_drv_apply(dev, flow, error); if (ret < 0) goto error; @@ -5613,11 +5942,8 @@ flow_list_destroy(struct rte_eth_dev *dev, uint32_t *list, if (flow->tunnel) { struct mlx5_flow_tunnel *tunnel; - rte_spinlock_lock(&mlx5_tunnel_hub(dev)->sl); tunnel = mlx5_find_tunnel_id(dev, flow->tunnel_id); RTE_VERIFY(tunnel); - LIST_REMOVE(tunnel, chain); - rte_spinlock_unlock(&mlx5_tunnel_hub(dev)->sl); if (!__atomic_sub_fetch(&tunnel->refctn, 1, __ATOMIC_RELAXED)) mlx5_flow_tunnel_free(dev, tunnel); } @@ -5683,7 +6009,7 @@ mlx5_flow_start_default(struct rte_eth_dev *dev) /** * Release key of thread specific flow workspace data. */ -static void +void flow_release_workspace(void *data) { struct mlx5_flow_workspace *wks = data; @@ -5697,16 +6023,6 @@ flow_release_workspace(void *data) } } -/** - * Initialize key of thread specific flow workspace data. - */ -static void -flow_alloc_workspace(void) -{ - if (pthread_key_create(&key_workspace, flow_release_workspace)) - DRV_LOG(ERR, "Can't create flow workspace data thread key."); -} - /** * Get thread specific current flow workspace. * @@ -5717,7 +6033,7 @@ mlx5_flow_get_thread_workspace(void) { struct mlx5_flow_workspace *data; - data = pthread_getspecific(key_workspace); + data = mlx5_flow_os_get_specific_workspace(); MLX5_ASSERT(data && data->inuse); if (!data || !data->inuse) DRV_LOG(ERR, "flow workspace not initialized."); @@ -5765,11 +6081,7 @@ mlx5_flow_push_thread_workspace(void) struct mlx5_flow_workspace *curr; struct mlx5_flow_workspace *data; - if (pthread_once(&key_workspace_init, flow_alloc_workspace)) { - DRV_LOG(ERR, "Failed to init flow workspace data thread key."); - return NULL; - } - curr = pthread_getspecific(key_workspace); + curr = mlx5_flow_os_get_specific_workspace(); if (!curr) { data = flow_alloc_thread_workspace(); if (!data) @@ -5788,7 +6100,7 @@ mlx5_flow_push_thread_workspace(void) data->inuse = 1; data->flow_idx = 0; /* Set as current workspace */ - if (pthread_setspecific(key_workspace, data)) + if (mlx5_flow_os_set_specific_workspace(data)) DRV_LOG(ERR, "Failed to set flow workspace to thread."); return data; } @@ -5814,7 +6126,7 @@ mlx5_flow_pop_thread_workspace(void) data->inuse = 0; if (!data->prev) return; - if (pthread_setspecific(key_workspace, data->prev)) + if (mlx5_flow_os_set_specific_workspace(data->prev)) DRV_LOG(ERR, "Failed to set flow workspace to thread."); } @@ -5931,7 +6243,7 @@ mlx5_ctrl_flow_vlan(struct rte_eth_dev *dev, struct mlx5_priv *priv = dev->data->dev_private; const struct rte_flow_attr attr = { .ingress = 1, - .priority = MLX5_FLOW_PRIO_RSVD, + .priority = MLX5_FLOW_LOWEST_PRIO_INDICATOR, }; struct rte_flow_item items[] = { { @@ -6120,9 +6432,9 @@ mlx5_flow_isolate(struct rte_eth_dev *dev, } priv->isolated = !!enable; if (enable) - dev->dev_ops = &mlx5_os_dev_ops_isolate; + dev->dev_ops = &mlx5_dev_ops_isolate; else - dev->dev_ops = &mlx5_os_dev_ops; + dev->dev_ops = &mlx5_dev_ops; dev->rx_descriptor_status = mlx5_rx_descriptor_status; dev->tx_descriptor_status = mlx5_tx_descriptor_status; @@ -6186,40 +6498,20 @@ mlx5_flow_query(struct rte_eth_dev *dev, } /** - * Manage filter operations. + * Get rte_flow callbacks. * * @param dev * Pointer to Ethernet device structure. - * @param filter_type - * Filter type. - * @param filter_op - * Operation to perform. - * @param arg + * @param ops * Pointer to operation-specific structure. * - * @return - * 0 on success, a negative errno value otherwise and rte_errno is set. + * @return 0 */ int -mlx5_dev_filter_ctrl(struct rte_eth_dev *dev, - enum rte_filter_type filter_type, - enum rte_filter_op filter_op, - void *arg) -{ - switch (filter_type) { - case RTE_ETH_FILTER_GENERIC: - if (filter_op != RTE_ETH_FILTER_GET) { - rte_errno = EINVAL; - return -rte_errno; - } - *(const void **)arg = &mlx5_flow_ops; - return 0; - default: - DRV_LOG(ERR, "port %u filter type (%d) not supported", - dev->data->port_id, filter_type); - rte_errno = ENOTSUP; - return -rte_errno; - } +mlx5_flow_ops_get(struct rte_eth_dev *dev __rte_unused, + const struct rte_flow_ops **ops) +{ + *ops = &mlx5_flow_ops; return 0; } @@ -6431,7 +6723,7 @@ mlx5_flow_create_counter_stat_mem_mng(struct mlx5_dev_ctx_shared *sh) } mem_mng = (struct mlx5_counter_stats_mem_mng *)(mem + size) - 1; size = sizeof(*raw_data) * MLX5_COUNTERS_PER_POOL * raws_n; - mem_mng->umem = mlx5_glue->devx_umem_reg(sh->ctx, mem, size, + mem_mng->umem = mlx5_os_umem_reg(sh->ctx, mem, size, IBV_ACCESS_LOCAL_WRITE); if (!mem_mng->umem) { rte_errno = errno; @@ -6450,7 +6742,7 @@ mlx5_flow_create_counter_stat_mem_mng(struct mlx5_dev_ctx_shared *sh) mkey_attr.relaxed_ordering_read = sh->cmng.relaxed_ordering_read; mem_mng->dm = mlx5_devx_cmd_mkey_create(sh->ctx, &mkey_attr); if (!mem_mng->dm) { - mlx5_glue->devx_umem_dereg(mem_mng->umem); + mlx5_os_umem_dereg(mem_mng->umem); rte_errno = errno; mlx5_free(mem); return -rte_errno; @@ -6776,7 +7068,7 @@ mlx5_flow_group_to_table(struct rte_eth_dev *dev, standard_translation = true; } DRV_LOG(DEBUG, - "port %u group=%#x transfer=%d external=%d fdb_def_rule=%d translate=%s", + "port %u group=%u transfer=%d external=%d fdb_def_rule=%d translate=%s", dev->data->port_id, group, grp_info->transfer, grp_info->external, grp_info->fdb_def_rule, standard_translation ? "STANDARD" : "TUNNEL"); @@ -6816,7 +7108,7 @@ mlx5_flow_discover_mreg_c(struct rte_eth_dev *dev) for (idx = REG_C_2; idx <= REG_C_7; ++idx) { struct rte_flow_attr attr = { .group = MLX5_FLOW_MREG_CP_TABLE_GROUP, - .priority = MLX5_FLOW_PRIO_RSVD, + .priority = MLX5_FLOW_LOWEST_PRIO_INDICATOR, .ingress = 1, }; struct rte_flow_item items[] = { @@ -7146,12 +7438,12 @@ mlx5_shared_action_flush(struct rte_eth_dev *dev) { struct rte_flow_error error; struct mlx5_priv *priv = dev->data->dev_private; - struct mlx5_shared_action_rss *action; + struct mlx5_shared_action_rss *shared_rss; int ret = 0; uint32_t idx; ILIST_FOREACH(priv->sh->ipool[MLX5_IPOOL_RSS_SHARED_ACTIONS], - priv->rss_shared_actions, idx, action, next) { + priv->rss_shared_actions, idx, shared_rss, next) { ret |= mlx5_shared_action_destroy(dev, (struct rte_flow_shared_action *)(uintptr_t)idx, &error); } @@ -7194,6 +7486,15 @@ union tunnel_offload_mark { }; }; +static bool +mlx5_access_tunnel_offload_db + (struct rte_eth_dev *dev, + bool (*match)(struct rte_eth_dev *, + struct mlx5_flow_tunnel *, const void *), + void (*hit)(struct rte_eth_dev *, struct mlx5_flow_tunnel *, void *), + void (*miss)(struct rte_eth_dev *, void *), + void *ctx, bool lock_op); + static int flow_tunnel_add_default_miss(struct rte_eth_dev *dev, struct rte_flow *flow, @@ -7340,13 +7641,28 @@ mlx5_flow_tunnel_grp2tbl_remove_cb(struct mlx5_hlist *list, mlx5_free(tte); } +static int +mlx5_flow_tunnel_grp2tbl_match_cb(struct mlx5_hlist *list __rte_unused, + struct mlx5_hlist_entry *entry, + uint64_t key, void *cb_ctx __rte_unused) +{ + union tunnel_tbl_key tbl = { + .val = key, + }; + struct tunnel_tbl_entry *tte = container_of(entry, typeof(*tte), hash); + + return tbl.tunnel_id != tte->tunnel_id || tbl.group != tte->group; +} + static struct mlx5_hlist_entry * -mlx5_flow_tunnel_grp2tbl_create_cb(struct mlx5_hlist *list, - uint64_t key __rte_unused, +mlx5_flow_tunnel_grp2tbl_create_cb(struct mlx5_hlist *list, uint64_t key, void *ctx __rte_unused) { struct mlx5_dev_ctx_shared *sh = list->ctx; struct tunnel_tbl_entry *tte; + union tunnel_tbl_key tbl = { + .val = key, + }; tte = mlx5_malloc(MLX5_MEM_SYS | MLX5_MEM_ZERO, sizeof(*tte), 0, @@ -7365,6 +7681,8 @@ mlx5_flow_tunnel_grp2tbl_create_cb(struct mlx5_hlist *list, goto err; } tte->flow_table = tunnel_id_to_flow_tbl(tte->flow_table); + tte->tunnel_id = tbl.tunnel_id; + tte->group = tbl.group; return &tte->hash; err: if (tte) @@ -7406,28 +7724,82 @@ mlx5_flow_tunnel_free(struct rte_eth_dev *dev, struct mlx5_flow_tunnel *tunnel) { struct mlx5_priv *priv = dev->data->dev_private; + struct mlx5_indexed_pool *ipool; DRV_LOG(DEBUG, "port %u release pmd tunnel id=0x%x", dev->data->port_id, tunnel->tunnel_id); - RTE_VERIFY(!__atomic_load_n(&tunnel->refctn, __ATOMIC_RELAXED)); - mlx5_ipool_free(priv->sh->ipool[MLX5_IPOOL_TUNNEL_ID], - tunnel->tunnel_id); + LIST_REMOVE(tunnel, chain); mlx5_hlist_destroy(tunnel->groups); - mlx5_free(tunnel); + ipool = priv->sh->ipool[MLX5_IPOOL_TUNNEL_ID]; + mlx5_ipool_free(ipool, tunnel->tunnel_id); } -static struct mlx5_flow_tunnel * -mlx5_find_tunnel_id(struct rte_eth_dev *dev, uint32_t id) +static bool +mlx5_access_tunnel_offload_db + (struct rte_eth_dev *dev, + bool (*match)(struct rte_eth_dev *, + struct mlx5_flow_tunnel *, const void *), + void (*hit)(struct rte_eth_dev *, struct mlx5_flow_tunnel *, void *), + void (*miss)(struct rte_eth_dev *, void *), + void *ctx, bool lock_op) { + bool verdict = false; struct mlx5_flow_tunnel_hub *thub = mlx5_tunnel_hub(dev); - struct mlx5_flow_tunnel *tun; + struct mlx5_flow_tunnel *tunnel; - LIST_FOREACH(tun, &thub->tunnels, chain) { - if (tun->tunnel_id == id) + rte_spinlock_lock(&thub->sl); + LIST_FOREACH(tunnel, &thub->tunnels, chain) { + verdict = match(dev, tunnel, (const void *)ctx); + if (verdict) break; } + if (!lock_op) + rte_spinlock_unlock(&thub->sl); + if (verdict && hit) + hit(dev, tunnel, ctx); + if (!verdict && miss) + miss(dev, ctx); + if (lock_op) + rte_spinlock_unlock(&thub->sl); + + return verdict; +} + +struct tunnel_db_find_tunnel_id_ctx { + uint32_t tunnel_id; + struct mlx5_flow_tunnel *tunnel; +}; + +static bool +find_tunnel_id_match(struct rte_eth_dev *dev, + struct mlx5_flow_tunnel *tunnel, const void *x) +{ + const struct tunnel_db_find_tunnel_id_ctx *ctx = x; + + RTE_SET_USED(dev); + return tunnel->tunnel_id == ctx->tunnel_id; +} + +static void +find_tunnel_id_hit(struct rte_eth_dev *dev, + struct mlx5_flow_tunnel *tunnel, void *x) +{ + struct tunnel_db_find_tunnel_id_ctx *ctx = x; + RTE_SET_USED(dev); + ctx->tunnel = tunnel; +} + +static struct mlx5_flow_tunnel * +mlx5_find_tunnel_id(struct rte_eth_dev *dev, uint32_t id) +{ + struct tunnel_db_find_tunnel_id_ctx ctx = { + .tunnel_id = id, + }; + + mlx5_access_tunnel_offload_db(dev, find_tunnel_id_match, + find_tunnel_id_hit, NULL, &ctx, true); - return tun; + return ctx.tunnel; } static struct mlx5_flow_tunnel * @@ -7435,39 +7807,25 @@ mlx5_flow_tunnel_allocate(struct rte_eth_dev *dev, const struct rte_flow_tunnel *app_tunnel) { struct mlx5_priv *priv = dev->data->dev_private; + struct mlx5_indexed_pool *ipool; struct mlx5_flow_tunnel *tunnel; uint32_t id; - mlx5_ipool_malloc(priv->sh->ipool[MLX5_IPOOL_RSS_EXPANTION_FLOW_ID], - &id); + ipool = priv->sh->ipool[MLX5_IPOOL_TUNNEL_ID]; + tunnel = mlx5_ipool_zmalloc(ipool, &id); + if (!tunnel) + return NULL; if (id >= MLX5_MAX_TUNNELS) { - mlx5_ipool_free(priv->sh->ipool - [MLX5_IPOOL_RSS_EXPANTION_FLOW_ID], id); + mlx5_ipool_free(ipool, id); DRV_LOG(ERR, "Tunnel ID %d exceed max limit.", id); return NULL; - } else if (!id) { - return NULL; - } - /** - * mlx5 flow tunnel is an auxlilary data structure - * It's not part of IO. No need to allocate it from - * huge pages pools dedicated for IO - */ - tunnel = mlx5_malloc(MLX5_MEM_SYS | MLX5_MEM_ZERO, sizeof(*tunnel), - 0, SOCKET_ID_ANY); - if (!tunnel) { - mlx5_ipool_free(priv->sh->ipool - [MLX5_IPOOL_RSS_EXPANTION_FLOW_ID], id); - return NULL; } tunnel->groups = mlx5_hlist_create("tunnel groups", 1024, 0, 0, mlx5_flow_tunnel_grp2tbl_create_cb, - NULL, + mlx5_flow_tunnel_grp2tbl_match_cb, mlx5_flow_tunnel_grp2tbl_remove_cb); if (!tunnel->groups) { - mlx5_ipool_free(priv->sh->ipool - [MLX5_IPOOL_RSS_EXPANTION_FLOW_ID], id); - mlx5_free(tunnel); + mlx5_ipool_free(ipool, id); return NULL; } tunnel->groups->ctx = priv->sh; @@ -7489,38 +7847,61 @@ mlx5_flow_tunnel_allocate(struct rte_eth_dev *dev, return tunnel; } -static int -mlx5_get_flow_tunnel(struct rte_eth_dev *dev, - const struct rte_flow_tunnel *app_tunnel, - struct mlx5_flow_tunnel **tunnel) +struct tunnel_db_get_tunnel_ctx { + const struct rte_flow_tunnel *app_tunnel; + struct mlx5_flow_tunnel *tunnel; +}; + +static bool get_tunnel_match(struct rte_eth_dev *dev, + struct mlx5_flow_tunnel *tunnel, const void *x) { - int ret; + const struct tunnel_db_get_tunnel_ctx *ctx = x; + + RTE_SET_USED(dev); + return !memcmp(ctx->app_tunnel, &tunnel->app_tunnel, + sizeof(*ctx->app_tunnel)); +} + +static void get_tunnel_hit(struct rte_eth_dev *dev, + struct mlx5_flow_tunnel *tunnel, void *x) +{ + /* called under tunnel spinlock protection */ + struct tunnel_db_get_tunnel_ctx *ctx = x; + + RTE_SET_USED(dev); + tunnel->refctn++; + ctx->tunnel = tunnel; +} + +static void get_tunnel_miss(struct rte_eth_dev *dev, void *x) +{ + /* called under tunnel spinlock protection */ struct mlx5_flow_tunnel_hub *thub = mlx5_tunnel_hub(dev); - struct mlx5_flow_tunnel *tun; + struct tunnel_db_get_tunnel_ctx *ctx = x; + rte_spinlock_unlock(&thub->sl); + ctx->tunnel = mlx5_flow_tunnel_allocate(dev, ctx->app_tunnel); rte_spinlock_lock(&thub->sl); - LIST_FOREACH(tun, &thub->tunnels, chain) { - if (!memcmp(app_tunnel, &tun->app_tunnel, - sizeof(*app_tunnel))) { - *tunnel = tun; - ret = 0; - break; - } + if (ctx->tunnel) { + ctx->tunnel->refctn = 1; + LIST_INSERT_HEAD(&thub->tunnels, ctx->tunnel, chain); } - if (!tun) { - tun = mlx5_flow_tunnel_allocate(dev, app_tunnel); - if (tun) { - LIST_INSERT_HEAD(&thub->tunnels, tun, chain); - *tunnel = tun; - } else { - ret = -ENOMEM; - } - } - rte_spinlock_unlock(&thub->sl); - if (tun) - __atomic_add_fetch(&tun->refctn, 1, __ATOMIC_RELAXED); +} - return ret; + +static int +mlx5_get_flow_tunnel(struct rte_eth_dev *dev, + const struct rte_flow_tunnel *app_tunnel, + struct mlx5_flow_tunnel **tunnel) +{ + struct tunnel_db_get_tunnel_ctx ctx = { + .app_tunnel = app_tunnel, + }; + + mlx5_access_tunnel_offload_db(dev, get_tunnel_match, get_tunnel_hit, + get_tunnel_miss, &ctx, true); + *tunnel = ctx.tunnel; + return ctx.tunnel ? 0 : -ENOMEM; } void mlx5_release_tunnel_hub(struct mlx5_dev_ctx_shared *sh, uint16_t port_id) @@ -7530,7 +7911,7 @@ void mlx5_release_tunnel_hub(struct mlx5_dev_ctx_shared *sh, uint16_t port_id) if (!thub) return; if (!LIST_EMPTY(&thub->tunnels)) - DRV_LOG(WARNING, "port %u tunnels present\n", port_id); + DRV_LOG(WARNING, "port %u tunnels present", port_id); mlx5_hlist_destroy(thub->groups); mlx5_free(thub); } @@ -7546,9 +7927,10 @@ int mlx5_alloc_tunnel_hub(struct mlx5_dev_ctx_shared *sh) return -ENOMEM; LIST_INIT(&thub->tunnels); rte_spinlock_init(&thub->sl); - thub->groups = mlx5_hlist_create("flow groups", MLX5_MAX_TABLES, 0, + thub->groups = mlx5_hlist_create("flow groups", + rte_align32pow2(MLX5_MAX_TABLES), 0, 0, mlx5_flow_tunnel_grp2tbl_create_cb, - NULL, + mlx5_flow_tunnel_grp2tbl_match_cb, mlx5_flow_tunnel_grp2tbl_remove_cb); if (!thub->groups) { err = -rte_errno; @@ -7646,56 +8028,88 @@ mlx5_flow_tunnel_match(struct rte_eth_dev *dev, *num_of_items = 1; return 0; } + +struct tunnel_db_element_release_ctx { + struct rte_flow_item *items; + struct rte_flow_action *actions; + uint32_t num_elements; + struct rte_flow_error *error; + int ret; +}; + +static bool +tunnel_element_release_match(struct rte_eth_dev *dev, + struct mlx5_flow_tunnel *tunnel, const void *x) +{ + const struct tunnel_db_element_release_ctx *ctx = x; + + RTE_SET_USED(dev); + if (ctx->num_elements != 1) + return false; + else if (ctx->items) + return ctx->items == &tunnel->item; + else if (ctx->actions) + return ctx->actions == &tunnel->action; + + return false; +} + +static void +tunnel_element_release_hit(struct rte_eth_dev *dev, + struct mlx5_flow_tunnel *tunnel, void *x) +{ + struct tunnel_db_element_release_ctx *ctx = x; + ctx->ret = 0; + if (!__atomic_sub_fetch(&tunnel->refctn, 1, __ATOMIC_RELAXED)) + mlx5_flow_tunnel_free(dev, tunnel); +} + +static void +tunnel_element_release_miss(struct rte_eth_dev *dev, void *x) +{ + struct tunnel_db_element_release_ctx *ctx = x; + RTE_SET_USED(dev); + ctx->ret = rte_flow_error_set(ctx->error, EINVAL, + RTE_FLOW_ERROR_TYPE_HANDLE, NULL, + "invalid argument"); +} + static int mlx5_flow_tunnel_item_release(struct rte_eth_dev *dev, - struct rte_flow_item *pmd_items, - uint32_t num_items, struct rte_flow_error *err) -{ - struct mlx5_flow_tunnel_hub *thub = mlx5_tunnel_hub(dev); - struct mlx5_flow_tunnel *tun; + struct rte_flow_item *pmd_items, + uint32_t num_items, struct rte_flow_error *err) +{ + struct tunnel_db_element_release_ctx ctx = { + .items = pmd_items, + .actions = NULL, + .num_elements = num_items, + .error = err, + }; - rte_spinlock_lock(&thub->sl); - LIST_FOREACH(tun, &thub->tunnels, chain) { - if (&tun->item == pmd_items) { - LIST_REMOVE(tun, chain); - break; - } - } - rte_spinlock_unlock(&thub->sl); - if (!tun || num_items != 1) - return rte_flow_error_set(err, EINVAL, - RTE_FLOW_ERROR_TYPE_HANDLE, NULL, - "invalid argument"); - if (!__atomic_sub_fetch(&tun->refctn, 1, __ATOMIC_RELAXED)) - mlx5_flow_tunnel_free(dev, tun); - return 0; + mlx5_access_tunnel_offload_db(dev, tunnel_element_release_match, + tunnel_element_release_hit, + tunnel_element_release_miss, &ctx, false); + + return ctx.ret; } static int mlx5_flow_tunnel_action_release(struct rte_eth_dev *dev, - struct rte_flow_action *pmd_actions, - uint32_t num_actions, - struct rte_flow_error *err) -{ - struct mlx5_flow_tunnel_hub *thub = mlx5_tunnel_hub(dev); - struct mlx5_flow_tunnel *tun; + struct rte_flow_action *pmd_actions, + uint32_t num_actions, struct rte_flow_error *err) +{ + struct tunnel_db_element_release_ctx ctx = { + .items = NULL, + .actions = pmd_actions, + .num_elements = num_actions, + .error = err, + }; - rte_spinlock_lock(&thub->sl); - LIST_FOREACH(tun, &thub->tunnels, chain) { - if (&tun->action == pmd_actions) { - LIST_REMOVE(tun, chain); - break; - } - } - rte_spinlock_unlock(&thub->sl); - if (!tun || num_actions != 1) - return rte_flow_error_set(err, EINVAL, - RTE_FLOW_ERROR_TYPE_HANDLE, NULL, - "invalid argument"); - if (!__atomic_sub_fetch(&tun->refctn, 1, __ATOMIC_RELAXED)) - mlx5_flow_tunnel_free(dev, tun); + mlx5_access_tunnel_offload_db(dev, tunnel_element_release_match, + tunnel_element_release_hit, + tunnel_element_release_miss, &ctx, false); - return 0; + return ctx.ret; } static int @@ -7708,6 +8122,11 @@ mlx5_flow_tunnel_get_restore_info(struct rte_eth_dev *dev, const struct mlx5_flow_tbl_data_entry *tble; const uint64_t mask = PKT_RX_FDIR | PKT_RX_FDIR_ID; + if (!is_tunnel_offload_active(dev)) { + info->flags = 0; + return 0; + } + if ((ol_flags & mask) != mask) goto err; tble = tunnel_mark_decode(dev, m->hash.fdir.hi);