ice: ndo_setup_tc implementation for PF
authorKiran Patil <kiran.patil@intel.com>
Fri, 6 Aug 2021 08:49:05 +0000 (10:49 +0200)
committerTony Nguyen <anthony.l.nguyen@intel.com>
Mon, 11 Oct 2021 16:03:04 +0000 (09:03 -0700)
Implement ndo_setup_tc net device callback for TC HW offload on PF device.

ndo_setup_tc provides support for HW offloading various TC filters.
Add support for configuring the following filter with tc-flower:
- default L2 filters (src/dst mac addresses, ethertype, VLAN)
- variations of L3, L3+L4, L2+L3+L4 filters using advanced filters
(including ipv4 and ipv6 addresses).

Allow for adding/removing TC flows when PF device is configured in
eswitch switchdev mode. Two types of actions are supported at the
moment: FLOW_ACTION_DROP and FLOW_ACTION_REDIRECT.

Co-developed-by: Priyalee Kushwaha <priyalee.kushwaha@intel.com>
Signed-off-by: Priyalee Kushwaha <priyalee.kushwaha@intel.com>
Signed-off-by: Kiran Patil <kiran.patil@intel.com>
Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
drivers/net/ethernet/intel/ice/Makefile
drivers/net/ethernet/intel/ice/ice.h
drivers/net/ethernet/intel/ice/ice_main.c
drivers/net/ethernet/intel/ice/ice_tc_lib.c [new file with mode: 0644]
drivers/net/ethernet/intel/ice/ice_tc_lib.h [new file with mode: 0644]

index 1866be50095db0f8d4d1c9c84bb652715292eaf9..c36faa7d14717a4a5d82cd192f173e7c3e3861f0 100644 (file)
@@ -27,7 +27,8 @@ ice-y := ice_main.o   \
         ice_fw_update.o \
         ice_lag.o      \
         ice_ethtool.o  \
-        ice_repr.o
+        ice_repr.o     \
+        ice_tc_lib.o
 ice-$(CONFIG_PCI_IOV) += ice_virtchnl_allowlist.o
 ice-$(CONFIG_PCI_IOV) += ice_virtchnl_pf.o ice_sriov.o ice_virtchnl_fdir.o
 ice-$(CONFIG_PTP_1588_CLOCK) += ice_ptp.o ice_ptp_hw.o
index 0d44a3767bad9496ce80ea921716889b8c0be604..6bfbf704eddf0263d5f7af3ed3b253bca928cda7 100644 (file)
@@ -64,6 +64,7 @@
 #include "ice_xsk.h"
 #include "ice_arfs.h"
 #include "ice_repr.h"
+#include "ice_eswitch.h"
 #include "ice_lag.h"
 
 #define ICE_BAR0               0
@@ -400,6 +401,7 @@ enum ice_pf_flags {
        ICE_FLAG_PTP,                   /* PTP is enabled by software */
        ICE_FLAG_AUX_ENA,
        ICE_FLAG_ADV_FEATURES,
+       ICE_FLAG_CLS_FLOWER,
        ICE_FLAG_LINK_DOWN_ON_CLOSE_ENA,
        ICE_FLAG_TOTAL_PORT_SHUTDOWN_ENA,
        ICE_FLAG_NO_MEDIA,
@@ -512,6 +514,8 @@ struct ice_pf {
        int aux_idx;
        u32 sw_int_count;
 
+       struct hlist_head tc_flower_fltr_list;
+
        __le64 nvm_phy_type_lo; /* NVM PHY type low */
        __le64 nvm_phy_type_hi; /* NVM PHY type high */
        struct ice_link_default_override_tlv link_dflt_override;
index b3066cfca6b7b2003eee0dd871c0d0c616b4cf79..ceb0912e5850f985f44e4cc8fe8265accdb43730 100644 (file)
@@ -20,6 +20,7 @@
 #define CREATE_TRACE_POINTS
 #include "ice_trace.h"
 #include "ice_eswitch.h"
+#include "ice_tc_lib.h"
 
 #define DRV_SUMMARY    "Intel(R) Ethernet Connection E800 Series Linux Driver"
 static const char ice_driver_string[] = DRV_SUMMARY;
@@ -3107,6 +3108,9 @@ static void ice_set_netdev_features(struct net_device *netdev)
 
        /* enable features */
        netdev->features |= netdev->hw_features;
+
+       netdev->hw_features |= NETIF_F_HW_TC;
+
        /* encap and VLAN devices inherit default, csumo and tso features */
        netdev->hw_enc_features |= dflt_features | csumo_features |
                                   tso_features;
@@ -7069,6 +7073,72 @@ static void ice_tx_timeout(struct net_device *netdev, unsigned int txqueue)
        pf->tx_timeout_recovery_level++;
 }
 
+/**
+ * ice_setup_tc_cls_flower - flower classifier offloads
+ * @np: net device to configure
+ * @filter_dev: device on which filter is added
+ * @cls_flower: offload data
+ */
+static int
+ice_setup_tc_cls_flower(struct ice_netdev_priv *np,
+                       struct net_device *filter_dev,
+                       struct flow_cls_offload *cls_flower)
+{
+       struct ice_vsi *vsi = np->vsi;
+
+       if (cls_flower->common.chain_index)
+               return -EOPNOTSUPP;
+
+       switch (cls_flower->command) {
+       case FLOW_CLS_REPLACE:
+               return ice_add_cls_flower(filter_dev, vsi, cls_flower);
+       case FLOW_CLS_DESTROY:
+               return ice_del_cls_flower(vsi, cls_flower);
+       default:
+               return -EINVAL;
+       }
+}
+
+/**
+ * ice_setup_tc_block_cb - callback handler registered for TC block
+ * @type: TC SETUP type
+ * @type_data: TC flower offload data that contains user input
+ * @cb_priv: netdev private data
+ */
+static int
+ice_setup_tc_block_cb(enum tc_setup_type type, void *type_data, void *cb_priv)
+{
+       struct ice_netdev_priv *np = cb_priv;
+
+       switch (type) {
+       case TC_SETUP_CLSFLOWER:
+               return ice_setup_tc_cls_flower(np, np->vsi->netdev,
+                                              type_data);
+       default:
+               return -EOPNOTSUPP;
+       }
+}
+
+static LIST_HEAD(ice_block_cb_list);
+
+static int
+ice_setup_tc(struct net_device *netdev, enum tc_setup_type type,
+            void *type_data)
+{
+       struct ice_netdev_priv *np = netdev_priv(netdev);
+
+       switch (type) {
+       case TC_SETUP_BLOCK:
+               return flow_block_cb_setup_simple(type_data,
+                                                 &ice_block_cb_list,
+                                                 ice_setup_tc_block_cb,
+                                                 np, np, true);
+       default:
+               return -EOPNOTSUPP;
+       }
+       return -EOPNOTSUPP;
+}
+
 /**
  * ice_open - Called when a network interface becomes active
  * @netdev: network interface device structure
@@ -7273,6 +7343,7 @@ static const struct net_device_ops ice_netdev_ops = {
        .ndo_get_vf_stats = ice_get_vf_stats,
        .ndo_vlan_rx_add_vid = ice_vlan_rx_add_vid,
        .ndo_vlan_rx_kill_vid = ice_vlan_rx_kill_vid,
+       .ndo_setup_tc = ice_setup_tc,
        .ndo_set_features = ice_set_features,
        .ndo_bridge_getlink = ice_bridge_getlink,
        .ndo_bridge_setlink = ice_bridge_setlink,
diff --git a/drivers/net/ethernet/intel/ice/ice_tc_lib.c b/drivers/net/ethernet/intel/ice/ice_tc_lib.c
new file mode 100644 (file)
index 0000000..d2b6490
--- /dev/null
@@ -0,0 +1,838 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (C) 2019-2021, Intel Corporation. */
+
+#include "ice.h"
+#include "ice_tc_lib.h"
+#include "ice_lib.h"
+#include "ice_fltr.h"
+
+/**
+ * ice_tc_count_lkups - determine lookup count for switch filter
+ * @flags: TC-flower flags
+ * @headers: Pointer to TC flower filter header structure
+ * @fltr: Pointer to outer TC filter structure
+ *
+ * Determine lookup count based on TC flower input for switch filter.
+ */
+static int
+ice_tc_count_lkups(u32 flags, struct ice_tc_flower_lyr_2_4_hdrs *headers,
+                  struct ice_tc_flower_fltr *fltr)
+{
+       int lkups_cnt = 0;
+
+       if (flags & ICE_TC_FLWR_FIELD_ETH_TYPE_ID)
+               lkups_cnt++;
+
+       /* are MAC fields specified? */
+       if (flags & (ICE_TC_FLWR_FIELD_DST_MAC | ICE_TC_FLWR_FIELD_SRC_MAC))
+               lkups_cnt++;
+
+       /* is VLAN specified? */
+       if (flags & ICE_TC_FLWR_FIELD_VLAN)
+               lkups_cnt++;
+
+       /* are IPv[4|6] fields specified? */
+       if (flags & (ICE_TC_FLWR_FIELD_DEST_IPV4 | ICE_TC_FLWR_FIELD_SRC_IPV4))
+               lkups_cnt++;
+       else if (flags & (ICE_TC_FLWR_FIELD_DEST_IPV6 |
+                         ICE_TC_FLWR_FIELD_SRC_IPV6))
+               lkups_cnt++;
+
+       /* is L4 (TCP/UDP/any other L4 protocol fields) specified? */
+       if (flags & (ICE_TC_FLWR_FIELD_DEST_L4_PORT |
+                    ICE_TC_FLWR_FIELD_SRC_L4_PORT))
+               lkups_cnt++;
+
+       return lkups_cnt;
+}
+
+/**
+ * ice_tc_fill_rules - fill filter rules based on TC fltr
+ * @hw: pointer to HW structure
+ * @flags: tc flower field flags
+ * @tc_fltr: pointer to TC flower filter
+ * @list: list of advance rule elements
+ * @rule_info: pointer to information about rule
+ * @l4_proto: pointer to information such as L4 proto type
+ *
+ * Fill ice_adv_lkup_elem list based on TC flower flags and
+ * TC flower headers. This list should be used to add
+ * advance filter in hardware.
+ */
+static int
+ice_tc_fill_rules(struct ice_hw *hw, u32 flags,
+                 struct ice_tc_flower_fltr *tc_fltr,
+                 struct ice_adv_lkup_elem *list,
+                 struct ice_adv_rule_info *rule_info,
+                 u16 *l4_proto)
+{
+       struct ice_tc_flower_lyr_2_4_hdrs *headers = &tc_fltr->outer_headers;
+       int i = 0;
+
+       if (flags & ICE_TC_FLWR_FIELD_ETH_TYPE_ID) {
+               list[i].type = ICE_ETYPE_OL;
+               list[i].h_u.ethertype.ethtype_id = headers->l2_key.n_proto;
+               list[i].m_u.ethertype.ethtype_id = headers->l2_mask.n_proto;
+               i++;
+       }
+
+       if (flags & (ICE_TC_FLWR_FIELD_DST_MAC |
+                    ICE_TC_FLWR_FIELD_SRC_MAC)) {
+               struct ice_tc_l2_hdr *l2_key, *l2_mask;
+
+               l2_key = &headers->l2_key;
+               l2_mask = &headers->l2_mask;
+
+               list[i].type = ICE_MAC_OFOS;
+               if (flags & ICE_TC_FLWR_FIELD_DST_MAC) {
+                       ether_addr_copy(list[i].h_u.eth_hdr.dst_addr,
+                                       l2_key->dst_mac);
+                       ether_addr_copy(list[i].m_u.eth_hdr.dst_addr,
+                                       l2_mask->dst_mac);
+               }
+               if (flags & ICE_TC_FLWR_FIELD_SRC_MAC) {
+                       ether_addr_copy(list[i].h_u.eth_hdr.src_addr,
+                                       l2_key->src_mac);
+                       ether_addr_copy(list[i].m_u.eth_hdr.src_addr,
+                                       l2_mask->src_mac);
+               }
+               i++;
+       }
+
+       /* copy VLAN info */
+       if (flags & ICE_TC_FLWR_FIELD_VLAN) {
+               list[i].type = ICE_VLAN_OFOS;
+               list[i].h_u.vlan_hdr.vlan = headers->vlan_hdr.vlan_id;
+               list[i].m_u.vlan_hdr.vlan = cpu_to_be16(0xFFFF);
+               i++;
+       }
+
+       /* copy L3 (IPv[4|6]: src, dest) address */
+       if (flags & (ICE_TC_FLWR_FIELD_DEST_IPV4 |
+                    ICE_TC_FLWR_FIELD_SRC_IPV4)) {
+               struct ice_tc_l3_hdr *l3_key, *l3_mask;
+
+               list[i].type = ICE_IPV4_OFOS;
+               l3_key = &headers->l3_key;
+               l3_mask = &headers->l3_mask;
+               if (flags & ICE_TC_FLWR_FIELD_DEST_IPV4) {
+                       list[i].h_u.ipv4_hdr.dst_addr = l3_key->dst_ipv4;
+                       list[i].m_u.ipv4_hdr.dst_addr = l3_mask->dst_ipv4;
+               }
+               if (flags & ICE_TC_FLWR_FIELD_SRC_IPV4) {
+                       list[i].h_u.ipv4_hdr.src_addr = l3_key->src_ipv4;
+                       list[i].m_u.ipv4_hdr.src_addr = l3_mask->src_ipv4;
+               }
+               i++;
+       } else if (flags & (ICE_TC_FLWR_FIELD_DEST_IPV6 |
+                           ICE_TC_FLWR_FIELD_SRC_IPV6)) {
+               struct ice_ipv6_hdr *ipv6_hdr, *ipv6_mask;
+               struct ice_tc_l3_hdr *l3_key, *l3_mask;
+
+               list[i].type = ICE_IPV6_OFOS;
+               ipv6_hdr = &list[i].h_u.ipv6_hdr;
+               ipv6_mask = &list[i].m_u.ipv6_hdr;
+               l3_key = &headers->l3_key;
+               l3_mask = &headers->l3_mask;
+
+               if (flags & ICE_TC_FLWR_FIELD_DEST_IPV6) {
+                       memcpy(&ipv6_hdr->dst_addr, &l3_key->dst_ipv6_addr,
+                              sizeof(l3_key->dst_ipv6_addr));
+                       memcpy(&ipv6_mask->dst_addr, &l3_mask->dst_ipv6_addr,
+                              sizeof(l3_mask->dst_ipv6_addr));
+               }
+               if (flags & ICE_TC_FLWR_FIELD_SRC_IPV6) {
+                       memcpy(&ipv6_hdr->src_addr, &l3_key->src_ipv6_addr,
+                              sizeof(l3_key->src_ipv6_addr));
+                       memcpy(&ipv6_mask->src_addr, &l3_mask->src_ipv6_addr,
+                              sizeof(l3_mask->src_ipv6_addr));
+               }
+               i++;
+       }
+
+       /* copy L4 (src, dest) port */
+       if (flags & (ICE_TC_FLWR_FIELD_DEST_L4_PORT |
+                    ICE_TC_FLWR_FIELD_SRC_L4_PORT)) {
+               struct ice_tc_l4_hdr *l4_key, *l4_mask;
+
+               l4_key = &headers->l4_key;
+               l4_mask = &headers->l4_mask;
+               if (headers->l3_key.ip_proto == IPPROTO_TCP) {
+                       list[i].type = ICE_TCP_IL;
+                       /* detected L4 proto is TCP */
+                       if (l4_proto)
+                               *l4_proto = IPPROTO_TCP;
+               } else if (headers->l3_key.ip_proto == IPPROTO_UDP) {
+                       list[i].type = ICE_UDP_ILOS;
+                       /* detected L4 proto is UDP */
+                       if (l4_proto)
+                               *l4_proto = IPPROTO_UDP;
+               }
+               if (flags & ICE_TC_FLWR_FIELD_DEST_L4_PORT) {
+                       list[i].h_u.l4_hdr.dst_port = l4_key->dst_port;
+                       list[i].m_u.l4_hdr.dst_port = l4_mask->dst_port;
+               }
+               if (flags & ICE_TC_FLWR_FIELD_SRC_L4_PORT) {
+                       list[i].h_u.l4_hdr.src_port = l4_key->src_port;
+                       list[i].m_u.l4_hdr.src_port = l4_mask->src_port;
+               }
+               i++;
+       }
+
+       return i;
+}
+
+static int
+ice_eswitch_tc_parse_action(struct ice_tc_flower_fltr *fltr,
+                           struct flow_action_entry *act)
+{
+       struct ice_repr *repr;
+
+       switch (act->id) {
+       case FLOW_ACTION_DROP:
+               fltr->action.fltr_act = ICE_DROP_PACKET;
+               break;
+
+       case FLOW_ACTION_REDIRECT:
+               fltr->action.fltr_act = ICE_FWD_TO_VSI;
+
+               if (ice_is_port_repr_netdev(act->dev)) {
+                       repr = ice_netdev_to_repr(act->dev);
+
+                       fltr->dest_vsi = repr->src_vsi;
+                       fltr->direction = ICE_ESWITCH_FLTR_INGRESS;
+               } else if (netif_is_ice(act->dev)) {
+                       struct ice_netdev_priv *np = netdev_priv(act->dev);
+
+                       fltr->dest_vsi = np->vsi;
+                       fltr->direction = ICE_ESWITCH_FLTR_EGRESS;
+               } else {
+                       NL_SET_ERR_MSG_MOD(fltr->extack, "Unsupported netdevice in switchdev mode");
+                       return -EINVAL;
+               }
+
+               break;
+
+       default:
+               NL_SET_ERR_MSG_MOD(fltr->extack, "Unsupported action in switchdev mode");
+               return -EINVAL;
+       }
+
+       return 0;
+}
+
+static int
+ice_eswitch_add_tc_fltr(struct ice_vsi *vsi, struct ice_tc_flower_fltr *fltr)
+{
+       struct ice_tc_flower_lyr_2_4_hdrs *headers = &fltr->outer_headers;
+       struct ice_adv_rule_info rule_info = { 0 };
+       struct ice_rule_query_data rule_added;
+       struct ice_hw *hw = &vsi->back->hw;
+       struct ice_adv_lkup_elem *list;
+       u32 flags = fltr->flags;
+       enum ice_status status;
+       int lkups_cnt;
+       int ret = 0;
+       int i;
+
+       if (!flags || (flags & (ICE_TC_FLWR_FIELD_ENC_DEST_IPV4 |
+                               ICE_TC_FLWR_FIELD_ENC_SRC_IPV4 |
+                               ICE_TC_FLWR_FIELD_ENC_DEST_IPV6 |
+                               ICE_TC_FLWR_FIELD_ENC_SRC_IPV6 |
+                               ICE_TC_FLWR_FIELD_ENC_SRC_L4_PORT))) {
+               NL_SET_ERR_MSG_MOD(fltr->extack, "Unsupported encap field(s)");
+               return -EOPNOTSUPP;
+       }
+
+       lkups_cnt = ice_tc_count_lkups(flags, headers, fltr);
+       list = kcalloc(lkups_cnt, sizeof(*list), GFP_ATOMIC);
+       if (!list)
+               return -ENOMEM;
+
+       i = ice_tc_fill_rules(hw, flags, fltr, list, &rule_info, NULL);
+       if (i != lkups_cnt) {
+               ret = -EINVAL;
+               goto exit;
+       }
+
+       rule_info.sw_act.fltr_act = fltr->action.fltr_act;
+       if (fltr->action.fltr_act != ICE_DROP_PACKET)
+               rule_info.sw_act.vsi_handle = fltr->dest_vsi->idx;
+       /* For now, making priority to be highest, and it also becomes
+        * the priority for recipe which will get created as a result of
+        * new extraction sequence based on input set.
+        * Priority '7' is max val for switch recipe, higher the number
+        * results into order of switch rule evaluation.
+        */
+       rule_info.priority = 7;
+
+       if (fltr->direction == ICE_ESWITCH_FLTR_INGRESS) {
+               rule_info.sw_act.flag |= ICE_FLTR_RX;
+               rule_info.sw_act.src = hw->pf_id;
+               rule_info.rx = true;
+       } else {
+               rule_info.sw_act.flag |= ICE_FLTR_TX;
+               rule_info.sw_act.src = vsi->idx;
+               rule_info.rx = false;
+       }
+
+       /* specify the cookie as filter_rule_id */
+       rule_info.fltr_rule_id = fltr->cookie;
+
+       status = ice_add_adv_rule(hw, list, lkups_cnt, &rule_info, &rule_added);
+       if (status == ICE_ERR_ALREADY_EXISTS) {
+               NL_SET_ERR_MSG_MOD(fltr->extack, "Unable to add filter because it already exist");
+               ret = -EINVAL;
+               goto exit;
+       } else if (status) {
+               NL_SET_ERR_MSG_MOD(fltr->extack, "Unable to add filter due to error");
+               ret = -EIO;
+               goto exit;
+       }
+
+       /* store the output params, which are needed later for removing
+        * advanced switch filter
+        */
+       fltr->rid = rule_added.rid;
+       fltr->rule_id = rule_added.rule_id;
+
+       if (fltr->direction == ICE_ESWITCH_FLTR_EGRESS) {
+               if (ice_fltr_update_flags(vsi, fltr->rule_id, fltr->rid,
+                                         ICE_SINGLE_ACT_LAN_ENABLE))
+                       ice_rem_adv_rule_by_id(hw, &rule_added);
+       }
+
+exit:
+       kfree(list);
+       return ret;
+}
+
+/**
+ * ice_tc_set_ipv4 - Parse IPv4 addresses from TC flower filter
+ * @match: Pointer to flow match structure
+ * @fltr: Pointer to filter structure
+ * @headers: inner or outer header fields
+ */
+static int
+ice_tc_set_ipv4(struct flow_match_ipv4_addrs *match,
+               struct ice_tc_flower_fltr *fltr,
+               struct ice_tc_flower_lyr_2_4_hdrs *headers)
+{
+       if (match->key->dst) {
+               fltr->flags |= ICE_TC_FLWR_FIELD_DEST_IPV4;
+               headers->l3_key.dst_ipv4 = match->key->dst;
+               headers->l3_mask.dst_ipv4 = match->mask->dst;
+       }
+       if (match->key->src) {
+               fltr->flags |= ICE_TC_FLWR_FIELD_SRC_IPV4;
+               headers->l3_key.src_ipv4 = match->key->src;
+               headers->l3_mask.src_ipv4 = match->mask->src;
+       }
+       return 0;
+}
+
+/**
+ * ice_tc_set_ipv6 - Parse IPv6 addresses from TC flower filter
+ * @match: Pointer to flow match structure
+ * @fltr: Pointer to filter structure
+ * @headers: inner or outer header fields
+ */
+static int
+ice_tc_set_ipv6(struct flow_match_ipv6_addrs *match,
+               struct ice_tc_flower_fltr *fltr,
+               struct ice_tc_flower_lyr_2_4_hdrs *headers)
+{
+       struct ice_tc_l3_hdr *l3_key, *l3_mask;
+
+       /* src and dest IPV6 address should not be LOOPBACK
+        * (0:0:0:0:0:0:0:1), which can be represented as ::1
+        */
+       if (ipv6_addr_loopback(&match->key->dst) ||
+           ipv6_addr_loopback(&match->key->src)) {
+               NL_SET_ERR_MSG_MOD(fltr->extack, "Bad IPv6, addr is LOOPBACK");
+               return -EINVAL;
+       }
+       /* if src/dest IPv6 address is *,* error */
+       if (ipv6_addr_any(&match->mask->dst) &&
+           ipv6_addr_any(&match->mask->src)) {
+               NL_SET_ERR_MSG_MOD(fltr->extack, "Bad src/dest IPv6, addr is any");
+               return -EINVAL;
+       }
+       if (!ipv6_addr_any(&match->mask->dst))
+               fltr->flags |= ICE_TC_FLWR_FIELD_DEST_IPV6;
+       if (!ipv6_addr_any(&match->mask->src))
+               fltr->flags |= ICE_TC_FLWR_FIELD_SRC_IPV6;
+
+       l3_key = &headers->l3_key;
+       l3_mask = &headers->l3_mask;
+
+       if (fltr->flags & ICE_TC_FLWR_FIELD_SRC_IPV6) {
+               memcpy(&l3_key->src_ipv6_addr, &match->key->src.s6_addr,
+                      sizeof(match->key->src.s6_addr));
+               memcpy(&l3_mask->src_ipv6_addr, &match->mask->src.s6_addr,
+                      sizeof(match->mask->src.s6_addr));
+       }
+       if (fltr->flags & ICE_TC_FLWR_FIELD_DEST_IPV6) {
+               memcpy(&l3_key->dst_ipv6_addr, &match->key->dst.s6_addr,
+                      sizeof(match->key->dst.s6_addr));
+               memcpy(&l3_mask->dst_ipv6_addr, &match->mask->dst.s6_addr,
+                      sizeof(match->mask->dst.s6_addr));
+       }
+
+       return 0;
+}
+
+/**
+ * ice_tc_set_port - Parse ports from TC flower filter
+ * @match: Flow match structure
+ * @fltr: Pointer to filter structure
+ * @headers: inner or outer header fields
+ */
+static int
+ice_tc_set_port(struct flow_match_ports match,
+               struct ice_tc_flower_fltr *fltr,
+               struct ice_tc_flower_lyr_2_4_hdrs *headers)
+{
+       if (match.key->dst) {
+               fltr->flags |= ICE_TC_FLWR_FIELD_DEST_L4_PORT;
+               headers->l4_key.dst_port = match.key->dst;
+               headers->l4_mask.dst_port = match.mask->dst;
+       }
+       if (match.key->src) {
+               fltr->flags |= ICE_TC_FLWR_FIELD_SRC_L4_PORT;
+               headers->l4_key.src_port = match.key->src;
+               headers->l4_mask.src_port = match.mask->src;
+       }
+       return 0;
+}
+
+/**
+ * ice_parse_cls_flower - Parse TC flower filters provided by kernel
+ * @vsi: Pointer to the VSI
+ * @filter_dev: Pointer to device on which filter is being added
+ * @f: Pointer to struct flow_cls_offload
+ * @fltr: Pointer to filter structure
+ */
+static int
+ice_parse_cls_flower(struct net_device *filter_dev, struct ice_vsi *vsi,
+                    struct flow_cls_offload *f,
+                    struct ice_tc_flower_fltr *fltr)
+{
+       struct ice_tc_flower_lyr_2_4_hdrs *headers = &fltr->outer_headers;
+       struct flow_rule *rule = flow_cls_offload_flow_rule(f);
+       u16 n_proto_mask = 0, n_proto_key = 0, addr_type = 0;
+       struct flow_dissector *dissector;
+
+       dissector = rule->match.dissector;
+
+       if (dissector->used_keys &
+           ~(BIT(FLOW_DISSECTOR_KEY_CONTROL) |
+             BIT(FLOW_DISSECTOR_KEY_BASIC) |
+             BIT(FLOW_DISSECTOR_KEY_ETH_ADDRS) |
+             BIT(FLOW_DISSECTOR_KEY_VLAN) |
+             BIT(FLOW_DISSECTOR_KEY_IPV4_ADDRS) |
+             BIT(FLOW_DISSECTOR_KEY_IPV6_ADDRS) |
+             BIT(FLOW_DISSECTOR_KEY_ENC_CONTROL) |
+             BIT(FLOW_DISSECTOR_KEY_ENC_IP) |
+             BIT(FLOW_DISSECTOR_KEY_PORTS))) {
+               NL_SET_ERR_MSG_MOD(fltr->extack, "Unsupported key used");
+               return -EOPNOTSUPP;
+       }
+
+       if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_BASIC)) {
+               struct flow_match_basic match;
+
+               flow_rule_match_basic(rule, &match);
+
+               n_proto_key = ntohs(match.key->n_proto);
+               n_proto_mask = ntohs(match.mask->n_proto);
+
+               if (n_proto_key == ETH_P_ALL || n_proto_key == 0) {
+                       n_proto_key = 0;
+                       n_proto_mask = 0;
+               } else {
+                       fltr->flags |= ICE_TC_FLWR_FIELD_ETH_TYPE_ID;
+               }
+
+               headers->l2_key.n_proto = cpu_to_be16(n_proto_key);
+               headers->l2_mask.n_proto = cpu_to_be16(n_proto_mask);
+               headers->l3_key.ip_proto = match.key->ip_proto;
+       }
+
+       if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_ETH_ADDRS)) {
+               struct flow_match_eth_addrs match;
+
+               flow_rule_match_eth_addrs(rule, &match);
+
+               if (!is_zero_ether_addr(match.key->dst)) {
+                       ether_addr_copy(headers->l2_key.dst_mac,
+                                       match.key->dst);
+                       ether_addr_copy(headers->l2_mask.dst_mac,
+                                       match.mask->dst);
+                       fltr->flags |= ICE_TC_FLWR_FIELD_DST_MAC;
+               }
+
+               if (!is_zero_ether_addr(match.key->src)) {
+                       ether_addr_copy(headers->l2_key.src_mac,
+                                       match.key->src);
+                       ether_addr_copy(headers->l2_mask.src_mac,
+                                       match.mask->src);
+                       fltr->flags |= ICE_TC_FLWR_FIELD_SRC_MAC;
+               }
+       }
+
+       if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_VLAN) ||
+           is_vlan_dev(filter_dev)) {
+               struct flow_dissector_key_vlan mask;
+               struct flow_dissector_key_vlan key;
+               struct flow_match_vlan match;
+
+               if (is_vlan_dev(filter_dev)) {
+                       match.key = &key;
+                       match.key->vlan_id = vlan_dev_vlan_id(filter_dev);
+                       match.key->vlan_priority = 0;
+                       match.mask = &mask;
+                       memset(match.mask, 0xff, sizeof(*match.mask));
+                       match.mask->vlan_priority = 0;
+               } else {
+                       flow_rule_match_vlan(rule, &match);
+               }
+
+               if (match.mask->vlan_id) {
+                       if (match.mask->vlan_id == VLAN_VID_MASK) {
+                               fltr->flags |= ICE_TC_FLWR_FIELD_VLAN;
+                       } else {
+                               NL_SET_ERR_MSG_MOD(fltr->extack, "Bad VLAN mask");
+                               return -EINVAL;
+                       }
+               }
+
+               headers->vlan_hdr.vlan_id =
+                               cpu_to_be16(match.key->vlan_id & VLAN_VID_MASK);
+               if (match.mask->vlan_priority)
+                       headers->vlan_hdr.vlan_prio = match.key->vlan_priority;
+       }
+
+       if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_CONTROL)) {
+               struct flow_match_control match;
+
+               flow_rule_match_control(rule, &match);
+
+               addr_type = match.key->addr_type;
+       }
+
+       if (addr_type == FLOW_DISSECTOR_KEY_IPV4_ADDRS) {
+               struct flow_match_ipv4_addrs match;
+
+               flow_rule_match_ipv4_addrs(rule, &match);
+               if (ice_tc_set_ipv4(&match, fltr, headers))
+                       return -EINVAL;
+       }
+
+       if (addr_type == FLOW_DISSECTOR_KEY_IPV6_ADDRS) {
+               struct flow_match_ipv6_addrs match;
+
+               flow_rule_match_ipv6_addrs(rule, &match);
+               if (ice_tc_set_ipv6(&match, fltr, headers))
+                       return -EINVAL;
+       }
+
+       if (flow_rule_match_key(rule, FLOW_DISSECTOR_KEY_PORTS)) {
+               struct flow_match_ports match;
+
+               flow_rule_match_ports(rule, &match);
+               if (ice_tc_set_port(match, fltr, headers))
+                       return -EINVAL;
+               switch (headers->l3_key.ip_proto) {
+               case IPPROTO_TCP:
+               case IPPROTO_UDP:
+                       break;
+               default:
+                       NL_SET_ERR_MSG_MOD(fltr->extack, "Only UDP and TCP transport are supported");
+                       return -EINVAL;
+               }
+       }
+       return 0;
+}
+
+/**
+ * ice_add_switch_fltr - Add TC flower filters
+ * @vsi: Pointer to VSI
+ * @fltr: Pointer to struct ice_tc_flower_fltr
+ *
+ * Add filter in HW switch block
+ */
+static int
+ice_add_switch_fltr(struct ice_vsi *vsi, struct ice_tc_flower_fltr *fltr)
+{
+       if (ice_is_eswitch_mode_switchdev(vsi->back))
+               return ice_eswitch_add_tc_fltr(vsi, fltr);
+
+       return -EOPNOTSUPP;
+}
+
+/**
+ * ice_handle_tclass_action - Support directing to a traffic class
+ * @vsi: Pointer to VSI
+ * @cls_flower: Pointer to TC flower offload structure
+ * @fltr: Pointer to TC flower filter structure
+ *
+ * Support directing traffic to a traffic class
+ */
+static int
+ice_handle_tclass_action(struct ice_vsi *vsi,
+                        struct flow_cls_offload *cls_flower,
+                        struct ice_tc_flower_fltr *fltr)
+{
+       int tc = tc_classid_to_hwtc(vsi->netdev, cls_flower->classid);
+
+       if (tc < 0) {
+               NL_SET_ERR_MSG_MOD(fltr->extack, "Unable to add filter because specified destination is invalid");
+               return -EINVAL;
+       }
+       if (!tc) {
+               NL_SET_ERR_MSG_MOD(fltr->extack, "Unable to add filter because of invalid destination");
+               return -EINVAL;
+       }
+
+       if (!(vsi->tc_cfg.ena_tc & BIT(tc))) {
+               NL_SET_ERR_MSG_MOD(fltr->extack, "Unable to add filter because of non-existence destination");
+               return -EINVAL;
+       }
+
+       /* Redirect to a TC class or Queue Group */
+       fltr->action.fltr_act = ICE_FWD_TO_QGRP;
+       fltr->action.tc_class = tc;
+
+       return 0;
+}
+
+/**
+ * ice_parse_tc_flower_actions - Parse the actions for a TC filter
+ * @vsi: Pointer to VSI
+ * @cls_flower: Pointer to TC flower offload structure
+ * @fltr: Pointer to TC flower filter structure
+ *
+ * Parse the actions for a TC filter
+ */
+static int
+ice_parse_tc_flower_actions(struct ice_vsi *vsi,
+                           struct flow_cls_offload *cls_flower,
+                           struct ice_tc_flower_fltr *fltr)
+{
+       struct flow_rule *rule = flow_cls_offload_flow_rule(cls_flower);
+       struct flow_action *flow_action = &rule->action;
+       struct flow_action_entry *act;
+       int i;
+
+       if (cls_flower->classid)
+               return ice_handle_tclass_action(vsi, cls_flower, fltr);
+
+       if (!flow_action_has_entries(flow_action))
+               return -EINVAL;
+
+       flow_action_for_each(i, act, flow_action) {
+               if (ice_is_eswitch_mode_switchdev(vsi->back)) {
+                       int err = ice_eswitch_tc_parse_action(fltr, act);
+
+                       if (err)
+                               return err;
+                       continue;
+               }
+               /* Allow only one rule per filter */
+
+               /* Drop action */
+               if (act->id == FLOW_ACTION_DROP) {
+                       fltr->action.fltr_act = ICE_DROP_PACKET;
+                       return 0;
+               }
+               fltr->action.fltr_act = ICE_FWD_TO_VSI;
+       }
+       return 0;
+}
+
+/**
+ * ice_del_tc_fltr - deletes a filter from HW table
+ * @vsi: Pointer to VSI
+ * @fltr: Pointer to struct ice_tc_flower_fltr
+ *
+ * This function deletes a filter from HW table and manages book-keeping
+ */
+static int ice_del_tc_fltr(struct ice_vsi *vsi, struct ice_tc_flower_fltr *fltr)
+{
+       struct ice_rule_query_data rule_rem;
+       struct ice_pf *pf = vsi->back;
+       int err;
+
+       rule_rem.rid = fltr->rid;
+       rule_rem.rule_id = fltr->rule_id;
+       rule_rem.vsi_handle = fltr->dest_id;
+       err = ice_rem_adv_rule_by_id(&pf->hw, &rule_rem);
+       if (err) {
+               if (err == ICE_ERR_DOES_NOT_EXIST) {
+                       NL_SET_ERR_MSG_MOD(fltr->extack, "Filter does not exist");
+                       return -ENOENT;
+               }
+               NL_SET_ERR_MSG_MOD(fltr->extack, "Failed to delete TC flower filter");
+               return -EIO;
+       }
+
+       return 0;
+}
+
+/**
+ * ice_add_tc_fltr - adds a TC flower filter
+ * @netdev: Pointer to netdev
+ * @vsi: Pointer to VSI
+ * @f: Pointer to flower offload structure
+ * @__fltr: Pointer to struct ice_tc_flower_fltr
+ *
+ * This function parses TC-flower input fields, parses action,
+ * and adds a filter.
+ */
+static int
+ice_add_tc_fltr(struct net_device *netdev, struct ice_vsi *vsi,
+               struct flow_cls_offload *f,
+               struct ice_tc_flower_fltr **__fltr)
+{
+       struct ice_tc_flower_fltr *fltr;
+       int err;
+
+       /* by default, set output to be INVALID */
+       *__fltr = NULL;
+
+       fltr = kzalloc(sizeof(*fltr), GFP_KERNEL);
+       if (!fltr)
+               return -ENOMEM;
+
+       fltr->cookie = f->cookie;
+       fltr->extack = f->common.extack;
+       fltr->src_vsi = vsi;
+       INIT_HLIST_NODE(&fltr->tc_flower_node);
+
+       err = ice_parse_cls_flower(netdev, vsi, f, fltr);
+       if (err < 0)
+               goto err;
+
+       err = ice_parse_tc_flower_actions(vsi, f, fltr);
+       if (err < 0)
+               goto err;
+
+       err = ice_add_switch_fltr(vsi, fltr);
+       if (err < 0)
+               goto err;
+
+       /* return the newly created filter */
+       *__fltr = fltr;
+
+       return 0;
+err:
+       kfree(fltr);
+       return err;
+}
+
+/**
+ * ice_find_tc_flower_fltr - Find the TC flower filter in the list
+ * @pf: Pointer to PF
+ * @cookie: filter specific cookie
+ */
+static struct ice_tc_flower_fltr *
+ice_find_tc_flower_fltr(struct ice_pf *pf, unsigned long cookie)
+{
+       struct ice_tc_flower_fltr *fltr;
+
+       hlist_for_each_entry(fltr, &pf->tc_flower_fltr_list, tc_flower_node)
+               if (cookie == fltr->cookie)
+                       return fltr;
+
+       return NULL;
+}
+
+/**
+ * ice_add_cls_flower - add TC flower filters
+ * @netdev: Pointer to filter device
+ * @vsi: Pointer to VSI
+ * @cls_flower: Pointer to flower offload structure
+ */
+int
+ice_add_cls_flower(struct net_device *netdev, struct ice_vsi *vsi,
+                  struct flow_cls_offload *cls_flower)
+{
+       struct netlink_ext_ack *extack = cls_flower->common.extack;
+       struct net_device *vsi_netdev = vsi->netdev;
+       struct ice_tc_flower_fltr *fltr;
+       struct ice_pf *pf = vsi->back;
+       int err;
+
+       if (ice_is_reset_in_progress(pf->state))
+               return -EBUSY;
+       if (test_bit(ICE_FLAG_FW_LLDP_AGENT, pf->flags))
+               return -EINVAL;
+
+       if (ice_is_port_repr_netdev(netdev))
+               vsi_netdev = netdev;
+
+       if (!(vsi_netdev->features & NETIF_F_HW_TC) &&
+           !test_bit(ICE_FLAG_CLS_FLOWER, pf->flags)) {
+               /* Based on TC indirect notifications from kernel, all ice
+                * devices get an instance of rule from higher level device.
+                * Avoid triggering explicit error in this case.
+                */
+               if (netdev == vsi_netdev)
+                       NL_SET_ERR_MSG_MOD(extack, "can't apply TC flower filters, turn ON hw-tc-offload and try again");
+               return -EINVAL;
+       }
+
+       /* avoid duplicate entries, if exists - return error */
+       fltr = ice_find_tc_flower_fltr(pf, cls_flower->cookie);
+       if (fltr) {
+               NL_SET_ERR_MSG_MOD(extack, "filter cookie already exists, ignoring");
+               return -EEXIST;
+       }
+
+       /* prep and add TC-flower filter in HW */
+       err = ice_add_tc_fltr(netdev, vsi, cls_flower, &fltr);
+       if (err)
+               return err;
+
+       /* add filter into an ordered list */
+       hlist_add_head(&fltr->tc_flower_node, &pf->tc_flower_fltr_list);
+       return 0;
+}
+
+/**
+ * ice_del_cls_flower - delete TC flower filters
+ * @vsi: Pointer to VSI
+ * @cls_flower: Pointer to struct flow_cls_offload
+ */
+int
+ice_del_cls_flower(struct ice_vsi *vsi, struct flow_cls_offload *cls_flower)
+{
+       struct ice_tc_flower_fltr *fltr;
+       struct ice_pf *pf = vsi->back;
+       int err;
+
+       /* find filter */
+       fltr = ice_find_tc_flower_fltr(pf, cls_flower->cookie);
+       if (!fltr) {
+               if (hlist_empty(&pf->tc_flower_fltr_list))
+                       return 0;
+
+               NL_SET_ERR_MSG_MOD(cls_flower->common.extack, "failed to delete TC flower filter because unable to find it");
+               return -EINVAL;
+       }
+
+       fltr->extack = cls_flower->common.extack;
+       /* delete filter from HW */
+       err = ice_del_tc_fltr(vsi, fltr);
+       if (err)
+               return err;
+
+       /* delete filter from an ordered list */
+       hlist_del(&fltr->tc_flower_node);
+
+       /* free the filter node */
+       kfree(fltr);
+
+       return 0;
+}
diff --git a/drivers/net/ethernet/intel/ice/ice_tc_lib.h b/drivers/net/ethernet/intel/ice/ice_tc_lib.h
new file mode 100644 (file)
index 0000000..883d289
--- /dev/null
@@ -0,0 +1,129 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright (C) 2019-2021, Intel Corporation. */
+
+#ifndef _ICE_TC_LIB_H_
+#define _ICE_TC_LIB_H_
+
+#define ICE_TC_FLWR_FIELD_DST_MAC              BIT(0)
+#define ICE_TC_FLWR_FIELD_SRC_MAC              BIT(1)
+#define ICE_TC_FLWR_FIELD_VLAN                 BIT(2)
+#define ICE_TC_FLWR_FIELD_DEST_IPV4            BIT(3)
+#define ICE_TC_FLWR_FIELD_SRC_IPV4             BIT(4)
+#define ICE_TC_FLWR_FIELD_DEST_IPV6            BIT(5)
+#define ICE_TC_FLWR_FIELD_SRC_IPV6             BIT(6)
+#define ICE_TC_FLWR_FIELD_DEST_L4_PORT         BIT(7)
+#define ICE_TC_FLWR_FIELD_SRC_L4_PORT          BIT(8)
+#define ICE_TC_FLWR_FIELD_TENANT_ID            BIT(9)
+#define ICE_TC_FLWR_FIELD_ENC_DEST_IPV4                BIT(10)
+#define ICE_TC_FLWR_FIELD_ENC_SRC_IPV4         BIT(11)
+#define ICE_TC_FLWR_FIELD_ENC_DEST_IPV6                BIT(12)
+#define ICE_TC_FLWR_FIELD_ENC_SRC_IPV6         BIT(13)
+#define ICE_TC_FLWR_FIELD_ENC_DEST_L4_PORT     BIT(14)
+#define ICE_TC_FLWR_FIELD_ENC_SRC_L4_PORT      BIT(15)
+#define ICE_TC_FLWR_FIELD_ENC_DST_MAC          BIT(16)
+#define ICE_TC_FLWR_FIELD_ETH_TYPE_ID          BIT(17)
+
+struct ice_tc_flower_action {
+       u32 tc_class;
+       enum ice_sw_fwd_act_type fltr_act;
+};
+
+struct ice_tc_vlan_hdr {
+       __be16 vlan_id; /* Only last 12 bits valid */
+       u16 vlan_prio; /* Only last 3 bits valid (valid values: 0..7) */
+};
+
+struct ice_tc_l2_hdr {
+       u8 dst_mac[ETH_ALEN];
+       u8 src_mac[ETH_ALEN];
+       __be16 n_proto;    /* Ethernet Protocol */
+};
+
+struct ice_tc_l3_hdr {
+       u8 ip_proto;    /* IPPROTO value */
+       union {
+               struct {
+                       struct in_addr dst_ip;
+                       struct in_addr src_ip;
+               } v4;
+               struct {
+                       struct in6_addr dst_ip6;
+                       struct in6_addr src_ip6;
+               } v6;
+       } ip;
+#define dst_ipv6       ip.v6.dst_ip6.s6_addr32
+#define dst_ipv6_addr  ip.v6.dst_ip6.s6_addr
+#define src_ipv6       ip.v6.src_ip6.s6_addr32
+#define src_ipv6_addr  ip.v6.src_ip6.s6_addr
+#define dst_ipv4       ip.v4.dst_ip.s_addr
+#define src_ipv4       ip.v4.src_ip.s_addr
+
+       u8 tos;
+       u8 ttl;
+};
+
+struct ice_tc_l4_hdr {
+       __be16 dst_port;
+       __be16 src_port;
+};
+
+struct ice_tc_flower_lyr_2_4_hdrs {
+       /* L2 layer fields with their mask */
+       struct ice_tc_l2_hdr l2_key;
+       struct ice_tc_l2_hdr l2_mask;
+       struct ice_tc_vlan_hdr vlan_hdr;
+       /* L3 (IPv4[6]) layer fields with their mask */
+       struct ice_tc_l3_hdr l3_key;
+       struct ice_tc_l3_hdr l3_mask;
+
+       /* L4 layer fields with their mask */
+       struct ice_tc_l4_hdr l4_key;
+       struct ice_tc_l4_hdr l4_mask;
+};
+
+enum ice_eswitch_fltr_direction {
+       ICE_ESWITCH_FLTR_INGRESS,
+       ICE_ESWITCH_FLTR_EGRESS,
+};
+
+struct ice_tc_flower_fltr {
+       struct hlist_node tc_flower_node;
+
+       /* cookie becomes filter_rule_id if rule is added successfully */
+       unsigned long cookie;
+
+       /* add_adv_rule returns information like recipe ID, rule_id. Store
+        * those values since they are needed to remove advanced rule
+        */
+       u16 rid;
+       u16 rule_id;
+       /* this could be queue/vsi_idx (sw handle)/queue_group, depending upon
+        * destination type
+        */
+       u16 dest_id;
+       /* if dest_id is vsi_idx, then need to store destination VSI ptr */
+       struct ice_vsi *dest_vsi;
+       /* direction of fltr for eswitch use case */
+       enum ice_eswitch_fltr_direction direction;
+
+       /* Parsed TC flower configuration params */
+       struct ice_tc_flower_lyr_2_4_hdrs outer_headers;
+       struct ice_tc_flower_lyr_2_4_hdrs inner_headers;
+       struct ice_vsi *src_vsi;
+       __be32 tenant_id;
+       u32 flags;
+       struct ice_tc_flower_action     action;
+
+       /* cache ptr which is used wherever needed to communicate netlink
+        * messages
+        */
+       struct netlink_ext_ack *extack;
+};
+
+int
+ice_add_cls_flower(struct net_device *netdev, struct ice_vsi *vsi,
+                  struct flow_cls_offload *cls_flower);
+int
+ice_del_cls_flower(struct ice_vsi *vsi, struct flow_cls_offload *cls_flower);
+
+#endif /* _ICE_TC_LIB_H_ */