From patchwork Wed Feb 19 14:27:16 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Bing Zhao X-Patchwork-Id: 65909 X-Patchwork-Delegate: rasland@nvidia.com Return-Path: X-Original-To: patchwork@inbox.dpdk.org Delivered-To: patchwork@inbox.dpdk.org Received: from dpdk.org (dpdk.org [92.243.14.124]) by inbox.dpdk.org (Postfix) with ESMTP id 67897A0555; Wed, 19 Feb 2020 15:27:23 +0100 (CET) Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id 457161BF84; Wed, 19 Feb 2020 15:27:23 +0100 (CET) Received: from git-send-mailer.rdmz.labs.mlnx (unknown [37.142.13.130]) by dpdk.org (Postfix) with ESMTP id 0F75B1B13C; Wed, 19 Feb 2020 15:27:22 +0100 (CET) From: Bing Zhao To: orika@mellanox.com, viacheslavo@mellanox.com Cc: rasland@mellanox.com, matan@mellanox.com, dev@dpdk.org, stable@dpdk.org Date: Wed, 19 Feb 2020 22:27:16 +0800 Message-Id: <1582122436-349956-1-git-send-email-bingz@mellanox.com> X-Mailer: git-send-email 1.8.3.1 Subject: [dpdk-dev] [PATCH] net/mlx5: fix the modify checking in flow validate X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" The header modify actions number supported now has some limitation, and it is decided by both driver and hardware. If the configuration is different or the table to insert the flow is different, the result might be different if the flow contains header modify actions. Currently, the actual action number could only be calculated in the later stage called translate, from user specified value to the driver format. And the action numbers checking is missed in the flow validation. So PMD will return incorrect result to indicate the flow actions are valid by rte_flow_validate but then it will fail when calling rte_flow_create. Adding some simple checking in the validation will help to get rid of this incorrect checking. Most of the actions will only consume 1 SW action field except the MAC address and IPv6 address. And from SW POV, the maximal action fields for these will be consumed even if only part of such field will be modified because that there is no mask in the flow actions and the mask will always be all ONEs. Fixes: 9597330c6844 ("net/mlx5: update modify header action translator") Cc: viacheslavo@mellanox.com Cc: stable@dpdk.org Signed-off-by: Bing Zhao --- drivers/net/mlx5/mlx5_flow.h | 17 +++++++++++++++ drivers/net/mlx5/mlx5_flow_dv.c | 47 ++++++++++++++++++++++++++++++++++++----- 2 files changed, 59 insertions(+), 5 deletions(-) diff --git a/drivers/net/mlx5/mlx5_flow.h b/drivers/net/mlx5/mlx5_flow.h index 791f3bd..45db487 100644 --- a/drivers/net/mlx5/mlx5_flow.h +++ b/drivers/net/mlx5/mlx5_flow.h @@ -328,6 +328,23 @@ enum mlx5_feature_name { #define MLX5_GENEVE_OPT_LEN_0 14 #define MLX5_GENEVE_OPT_LEN_1 63 +/* Software header modify action numbers of a flow. */ +#define MLX5_ACT_NUM_MDF_IPV4 1 +#define MLX5_ACT_NUM_MDF_IPV6 4 +#define MLX5_ACT_NUM_MDF_MAC 2 +#define MLX5_ACT_NUM_MDF_VID 1 +#define MLX5_ACT_NUM_MDF_PORT 2 +#define MLX5_ACT_NUM_MDF_TTL 1 +#define MLX5_ACT_NUM_DEC_TTL MLX5_ACT_NUM_MDF_TTL +#define MLX5_ACT_NUM_MDF_TCPSEQ 1 +#define MLX5_ACT_NUM_MDF_TCPACK 1 +#define MLX5_ACT_NUM_SET_REG 1 +#define MLX5_ACT_NUM_SET_TAG 0 +#define MLX5_ACT_NUM_CPY_MREG MLX5_ACT_NUM_SET_TAG +#define MLX5_ACT_NUM_SET_MARK 1 +#define MLX5_ACT_NUM_SET_META 1 +#define MLX5_ACT_NUM_SET_DSCP 1 + enum mlx5_flow_drv_type { MLX5_FLOW_TYPE_MIN, MLX5_FLOW_TYPE_DV, diff --git a/drivers/net/mlx5/mlx5_flow_dv.c b/drivers/net/mlx5/mlx5_flow_dv.c index 467d1ce..f79203a 100644 --- a/drivers/net/mlx5/mlx5_flow_dv.c +++ b/drivers/net/mlx5/mlx5_flow_dv.c @@ -375,6 +375,7 @@ struct field_modify_info modify_tcp[] = { /* Fetch variable byte size mask from the array. */ mask = flow_dv_fetch_field((const uint8_t *)item->mask + field->offset, field->size); + MLX5_ASSERT(mask); if (!mask) { ++field; continue; @@ -4302,7 +4303,9 @@ struct field_modify_info modify_tcp[] = { * Pointer to error structure. * * @return - * 0 on success, a negative errno value otherwise and rte_errno is set. + * - 0 on success and non root table. + * - 1 on success and root table. + * - a negative errno value otherwise and rte_errno is set. */ static int flow_dv_validate_attributes(struct rte_eth_dev *dev, @@ -4312,6 +4315,7 @@ struct field_modify_info modify_tcp[] = { { struct mlx5_priv *priv = dev->data->dev_private; uint32_t priority_max = priv->config.flow_prio - 1; + int ret = 0; #ifndef HAVE_MLX5DV_DR if (attributes->group) @@ -4321,13 +4325,14 @@ struct field_modify_info modify_tcp[] = { "groups are not supported"); #else uint32_t table; - int ret; ret = mlx5_flow_group_to_table(attributes, external, attributes->group, !!priv->fdb_def_rule, &table, error); if (ret) return ret; + if (!table) + ret = MLX5DV_DR_ACTION_FLAGS_ROOT_LEVEL; #endif if (attributes->priority != MLX5_FLOW_PRIO_RSVD && attributes->priority >= priority_max) @@ -4357,7 +4362,7 @@ struct field_modify_info modify_tcp[] = { RTE_FLOW_ERROR_TYPE_ATTR, NULL, "must specify exactly one of " "ingress or egress"); - return 0; + return ret; } /** @@ -4407,12 +4412,15 @@ struct field_modify_info modify_tcp[] = { struct mlx5_priv *priv = dev->data->dev_private; struct mlx5_dev_config *dev_conf = &priv->config; uint16_t queue_index = 0xFFFF; + int16_t rw_act_num = 0; + uint64_t is_root; if (items == NULL) return -1; ret = flow_dv_validate_attributes(dev, attr, external, error); if (ret < 0) return ret; + is_root = (uint64_t)ret; for (; items->type != RTE_FLOW_ITEM_TYPE_END; items++) { int tunnel = !!(item_flags & MLX5_FLOW_LAYER_TUNNEL); int type = items->type; @@ -4684,6 +4692,7 @@ struct field_modify_info modify_tcp[] = { action_flags |= MLX5_FLOW_ACTION_FLAG; ++actions_n; } + rw_act_num += MLX5_ACT_NUM_SET_MARK; break; case RTE_FLOW_ACTION_TYPE_MARK: ret = flow_dv_validate_action_mark(dev, actions, @@ -4702,6 +4711,7 @@ struct field_modify_info modify_tcp[] = { action_flags |= MLX5_FLOW_ACTION_MARK; ++actions_n; } + rw_act_num += MLX5_ACT_NUM_SET_MARK; break; case RTE_FLOW_ACTION_TYPE_SET_META: ret = flow_dv_validate_action_set_meta(dev, actions, @@ -4713,6 +4723,7 @@ struct field_modify_info modify_tcp[] = { if (!(action_flags & MLX5_FLOW_MODIFY_HDR_ACTIONS)) ++actions_n; action_flags |= MLX5_FLOW_ACTION_SET_META; + rw_act_num += MLX5_ACT_NUM_SET_META; break; case RTE_FLOW_ACTION_TYPE_SET_TAG: ret = flow_dv_validate_action_set_tag(dev, actions, @@ -4724,6 +4735,7 @@ struct field_modify_info modify_tcp[] = { if (!(action_flags & MLX5_FLOW_MODIFY_HDR_ACTIONS)) ++actions_n; action_flags |= MLX5_FLOW_ACTION_SET_TAG; + rw_act_num += MLX5_ACT_NUM_SET_TAG; break; case RTE_FLOW_ACTION_TYPE_DROP: ret = mlx5_flow_validate_action_drop(action_flags, @@ -4800,6 +4812,7 @@ struct field_modify_info modify_tcp[] = { return ret; /* Count VID with push_vlan command. */ action_flags |= MLX5_FLOW_ACTION_OF_SET_VLAN_VID; + rw_act_num += MLX5_ACT_NUM_MDF_VID; break; case RTE_FLOW_ACTION_TYPE_VXLAN_ENCAP: case RTE_FLOW_ACTION_TYPE_NVGRE_ENCAP: @@ -4858,8 +4871,15 @@ struct field_modify_info modify_tcp[] = { RTE_FLOW_ACTION_TYPE_SET_MAC_SRC ? MLX5_FLOW_ACTION_SET_MAC_SRC : MLX5_FLOW_ACTION_SET_MAC_DST; + /* + * Even if the source and destination MAC addresses have + * overlap in the header with 4B alignment, the convert + * function will handle them separately and 4 SW actions + * will be created. And 2 actions will be added each + * time no matter how many bytes of address will be set. + */ + rw_act_num += MLX5_ACT_NUM_MDF_MAC; break; - case RTE_FLOW_ACTION_TYPE_SET_IPV4_SRC: case RTE_FLOW_ACTION_TYPE_SET_IPV4_DST: ret = flow_dv_validate_action_modify_ipv4(action_flags, @@ -4875,6 +4895,7 @@ struct field_modify_info modify_tcp[] = { RTE_FLOW_ACTION_TYPE_SET_IPV4_SRC ? MLX5_FLOW_ACTION_SET_IPV4_SRC : MLX5_FLOW_ACTION_SET_IPV4_DST; + rw_act_num += MLX5_ACT_NUM_MDF_IPV4; break; case RTE_FLOW_ACTION_TYPE_SET_IPV6_SRC: case RTE_FLOW_ACTION_TYPE_SET_IPV6_DST: @@ -4897,6 +4918,7 @@ struct field_modify_info modify_tcp[] = { RTE_FLOW_ACTION_TYPE_SET_IPV6_SRC ? MLX5_FLOW_ACTION_SET_IPV6_SRC : MLX5_FLOW_ACTION_SET_IPV6_DST; + rw_act_num += MLX5_ACT_NUM_MDF_IPV6; break; case RTE_FLOW_ACTION_TYPE_SET_TP_SRC: case RTE_FLOW_ACTION_TYPE_SET_TP_DST: @@ -4913,6 +4935,7 @@ struct field_modify_info modify_tcp[] = { RTE_FLOW_ACTION_TYPE_SET_TP_SRC ? MLX5_FLOW_ACTION_SET_TP_SRC : MLX5_FLOW_ACTION_SET_TP_DST; + rw_act_num += MLX5_ACT_NUM_MDF_PORT; break; case RTE_FLOW_ACTION_TYPE_DEC_TTL: case RTE_FLOW_ACTION_TYPE_SET_TTL: @@ -4929,6 +4952,7 @@ struct field_modify_info modify_tcp[] = { RTE_FLOW_ACTION_TYPE_SET_TTL ? MLX5_FLOW_ACTION_SET_TTL : MLX5_FLOW_ACTION_DEC_TTL; + rw_act_num += MLX5_ACT_NUM_MDF_TTL; break; case RTE_FLOW_ACTION_TYPE_JUMP: ret = flow_dv_validate_action_jump(actions, @@ -4956,6 +4980,7 @@ struct field_modify_info modify_tcp[] = { RTE_FLOW_ACTION_TYPE_INC_TCP_SEQ ? MLX5_FLOW_ACTION_INC_TCP_SEQ : MLX5_FLOW_ACTION_DEC_TCP_SEQ; + rw_act_num += MLX5_ACT_NUM_MDF_TCPSEQ; break; case RTE_FLOW_ACTION_TYPE_INC_TCP_ACK: case RTE_FLOW_ACTION_TYPE_DEC_TCP_ACK: @@ -4973,10 +4998,13 @@ struct field_modify_info modify_tcp[] = { RTE_FLOW_ACTION_TYPE_INC_TCP_ACK ? MLX5_FLOW_ACTION_INC_TCP_ACK : MLX5_FLOW_ACTION_DEC_TCP_ACK; + rw_act_num += MLX5_ACT_NUM_MDF_TCPACK; break; - case MLX5_RTE_FLOW_ACTION_TYPE_TAG: case MLX5_RTE_FLOW_ACTION_TYPE_MARK: + break; + case MLX5_RTE_FLOW_ACTION_TYPE_TAG: case MLX5_RTE_FLOW_ACTION_TYPE_COPY_MREG: + rw_act_num += MLX5_ACT_NUM_SET_TAG; break; case RTE_FLOW_ACTION_TYPE_METER: ret = mlx5_flow_validate_action_meter(dev, @@ -5000,6 +5028,7 @@ struct field_modify_info modify_tcp[] = { if (!(action_flags & MLX5_FLOW_MODIFY_HDR_ACTIONS)) ++actions_n; action_flags |= MLX5_FLOW_ACTION_SET_IPV4_DSCP; + rw_act_num += MLX5_ACT_NUM_SET_DSCP; break; case RTE_FLOW_ACTION_TYPE_SET_IPV6_DSCP: ret = flow_dv_validate_action_modify_ipv6_dscp @@ -5013,6 +5042,7 @@ struct field_modify_info modify_tcp[] = { if (!(action_flags & MLX5_FLOW_MODIFY_HDR_ACTIONS)) ++actions_n; action_flags |= MLX5_FLOW_ACTION_SET_IPV6_DSCP; + rw_act_num += MLX5_ACT_NUM_SET_DSCP; break; default: return rte_flow_error_set(error, ENOTSUP, @@ -5085,6 +5115,13 @@ struct field_modify_info modify_tcp[] = { NULL, "encap is not supported" " for ingress traffic"); } + if ((uint32_t)rw_act_num >= + flow_dv_modify_hdr_action_max(dev, is_root)) { + return rte_flow_error_set(error, ENOTSUP, + RTE_FLOW_ERROR_TYPE_ACTION, + NULL, "too many header modify" + " actions to support"); + } return 0; }