ice: Flesh out implementation of support for SRIOV on bonded interface
authorDave Ertman <david.m.ertman@intel.com>
Tue, 20 Jun 2023 22:18:50 +0000 (15:18 -0700)
committerTony Nguyen <anthony.l.nguyen@intel.com>
Thu, 27 Jul 2023 17:56:32 +0000 (10:56 -0700)
Add in the functions that will allow a VF created on the primary interface
of a bond to "fail-over" to another PF interface in the bond and continue
to Tx and Rx.

Add in an ordered take-down path for the bonded interface.

Reviewed-by: Daniel Machon <daniel.machon@microchip.com>
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Tested-by: Sujai Buvaneswaran <sujai.buvaneswaran@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
drivers/net/ethernet/intel/ice/ice_lag.c

index ac4deb5..f054bf6 100644 (file)
 #define ICE_LAG_RES_SHARED     BIT(14)
 #define ICE_LAG_RES_VALID      BIT(15)
 
+#define LACP_TRAIN_PKT_LEN             16
+static const u8 lacp_train_pkt[LACP_TRAIN_PKT_LEN] = { 0, 0, 0, 0, 0, 0,
+                                                      0, 0, 0, 0, 0, 0,
+                                                      0x88, 0x09, 0, 0 };
+
 #define ICE_RECIPE_LEN                 64
 static const u8 ice_dflt_vsi_rcp[ICE_RECIPE_LEN] = {
        0x05, 0, 0, 0, 0x20, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
@@ -58,6 +63,39 @@ static void ice_lag_set_backup(struct ice_lag *lag)
 }
 
 /**
+ * netif_is_same_ice - determine if netdev is on the same ice NIC as local PF
+ * @pf: local PF struct
+ * @netdev: netdev we are evaluating
+ */
+static bool netif_is_same_ice(struct ice_pf *pf, struct net_device *netdev)
+{
+       struct ice_netdev_priv *np;
+       struct ice_pf *test_pf;
+       struct ice_vsi *vsi;
+
+       if (!netif_is_ice(netdev))
+               return false;
+
+       np = netdev_priv(netdev);
+       if (!np)
+               return false;
+
+       vsi = np->vsi;
+       if (!vsi)
+               return false;
+
+       test_pf = vsi->back;
+       if (!test_pf)
+               return false;
+
+       if (pf->pdev->bus != test_pf->pdev->bus ||
+           pf->pdev->slot != test_pf->pdev->slot)
+               return false;
+
+       return true;
+}
+
+/**
  * ice_netdev_to_lag - return pointer to associated lag struct from netdev
  * @netdev: pointer to net_device struct to query
  */
@@ -81,6 +119,38 @@ static struct ice_lag *ice_netdev_to_lag(struct net_device *netdev)
 }
 
 /**
+ * ice_lag_find_hw_by_lport - return an hw struct from bond members lport
+ * @lag: lag struct
+ * @lport: lport value to search for
+ */
+static struct ice_hw *
+ice_lag_find_hw_by_lport(struct ice_lag *lag, u8 lport)
+{
+       struct ice_lag_netdev_list *entry;
+       struct net_device *tmp_netdev;
+       struct ice_netdev_priv *np;
+       struct list_head *tmp;
+       struct ice_hw *hw;
+
+       list_for_each(tmp, lag->netdev_head) {
+               entry = list_entry(tmp, struct ice_lag_netdev_list, node);
+               tmp_netdev = entry->netdev;
+               if (!tmp_netdev || !netif_is_ice(tmp_netdev))
+                       continue;
+
+               np = netdev_priv(tmp_netdev);
+               if (!np || !np->vsi)
+                       continue;
+
+               hw = &np->vsi->back->hw;
+               if (hw->port_info->lport == lport)
+                       return hw;
+       }
+
+       return NULL;
+}
+
+/**
  * ice_lag_find_primary - returns pointer to primary interfaces lag struct
  * @lag: local interfaces lag struct
  */
@@ -234,6 +304,123 @@ static void ice_display_lag_info(struct ice_lag *lag)
 }
 
 /**
+ * ice_lag_qbuf_recfg - generate a buffer of queues for a reconfigure command
+ * @hw: HW struct that contains the queue contexts
+ * @qbuf: pointer to buffer to populate
+ * @vsi_num: index of the VSI in PF space
+ * @numq: number of queues to search for
+ * @tc: traffic class that contains the queues
+ *
+ * function returns the number of valid queues in buffer
+ */
+static u16
+ice_lag_qbuf_recfg(struct ice_hw *hw, struct ice_aqc_cfg_txqs_buf *qbuf,
+                  u16 vsi_num, u16 numq, u8 tc)
+{
+       struct ice_q_ctx *q_ctx;
+       u16 qid, count = 0;
+       struct ice_pf *pf;
+       int i;
+
+       pf = hw->back;
+       for (i = 0; i < numq; i++) {
+               q_ctx = ice_get_lan_q_ctx(hw, vsi_num, tc, i);
+               if (!q_ctx) {
+                       dev_dbg(ice_hw_to_dev(hw), "%s queue %d NO Q CONTEXT\n",
+                               __func__, i);
+                       continue;
+               }
+               if (q_ctx->q_teid == ICE_INVAL_TEID) {
+                       dev_dbg(ice_hw_to_dev(hw), "%s queue %d INVAL TEID\n",
+                               __func__, i);
+                       continue;
+               }
+               if (q_ctx->q_handle == ICE_INVAL_Q_HANDLE) {
+                       dev_dbg(ice_hw_to_dev(hw), "%s queue %d INVAL Q HANDLE\n",
+                               __func__, i);
+                       continue;
+               }
+
+               qid = pf->vsi[vsi_num]->txq_map[q_ctx->q_handle];
+               qbuf->queue_info[count].q_handle = cpu_to_le16(qid);
+               qbuf->queue_info[count].tc = tc;
+               qbuf->queue_info[count].q_teid = cpu_to_le32(q_ctx->q_teid);
+               count++;
+       }
+
+       return count;
+}
+
+/**
+ * ice_lag_get_sched_parent - locate or create a sched node parent
+ * @hw: HW struct for getting parent in
+ * @tc: traffic class on parent/node
+ */
+static struct ice_sched_node *
+ice_lag_get_sched_parent(struct ice_hw *hw, u8 tc)
+{
+       struct ice_sched_node *tc_node, *aggnode, *parent = NULL;
+       u16 num_nodes[ICE_AQC_TOPO_MAX_LEVEL_NUM] = { 0 };
+       struct ice_port_info *pi = hw->port_info;
+       struct device *dev;
+       u8 aggl, vsil;
+       int n;
+
+       dev = ice_hw_to_dev(hw);
+
+       tc_node = ice_sched_get_tc_node(pi, tc);
+       if (!tc_node) {
+               dev_warn(dev, "Failure to find TC node for LAG move\n");
+               return parent;
+       }
+
+       aggnode = ice_sched_get_agg_node(pi, tc_node, ICE_DFLT_AGG_ID);
+       if (!aggnode) {
+               dev_warn(dev, "Failure to find aggregate node for LAG move\n");
+               return parent;
+       }
+
+       aggl = ice_sched_get_agg_layer(hw);
+       vsil = ice_sched_get_vsi_layer(hw);
+
+       for (n = aggl + 1; n < vsil; n++)
+               num_nodes[n] = 1;
+
+       for (n = 0; n < aggnode->num_children; n++) {
+               parent = ice_sched_get_free_vsi_parent(hw, aggnode->children[n],
+                                                      num_nodes);
+               if (parent)
+                       return parent;
+       }
+
+       /* if free parent not found - add one */
+       parent = aggnode;
+       for (n = aggl + 1; n < vsil; n++) {
+               u16 num_nodes_added;
+               u32 first_teid;
+               int err;
+
+               err = ice_sched_add_nodes_to_layer(pi, tc_node, parent, n,
+                                                  num_nodes[n], &first_teid,
+                                                  &num_nodes_added);
+               if (err || num_nodes[n] != num_nodes_added)
+                       return NULL;
+
+               if (num_nodes_added)
+                       parent = ice_sched_find_node_by_teid(tc_node,
+                                                            first_teid);
+               else
+                       parent = parent->children[0];
+               if (!parent) {
+                       dev_warn(dev, "Failure to add new parent for LAG move\n");
+                       return parent;
+               }
+       }
+
+       return parent;
+}
+
+/**
  * ice_lag_move_vf_node_tc - move scheduling nodes for one VF on one TC
  * @lag: lag info struct
  * @oldport: lport of previous nodes location
@@ -245,6 +432,110 @@ static void
 ice_lag_move_vf_node_tc(struct ice_lag *lag, u8 oldport, u8 newport,
                        u16 vsi_num, u8 tc)
 {
+       u16 numq, valq, buf_size, num_moved, qbuf_size;
+       struct device *dev = ice_pf_to_dev(lag->pf);
+       struct ice_aqc_cfg_txqs_buf *qbuf;
+       struct ice_aqc_move_elem *buf;
+       struct ice_sched_node *n_prt;
+       struct ice_hw *new_hw = NULL;
+       __le32 teid, parent_teid;
+       struct ice_vsi_ctx *ctx;
+       u32 tmp_teid;
+
+       ctx = ice_get_vsi_ctx(&lag->pf->hw, vsi_num);
+       if (!ctx) {
+               dev_warn(dev, "Unable to locate VSI context for LAG failover\n");
+               return;
+       }
+
+       /* check to see if this VF is enabled on this TC */
+       if (!ctx->sched.vsi_node[tc])
+               return;
+
+       /* locate HW struct for destination port */
+       new_hw = ice_lag_find_hw_by_lport(lag, newport);
+       if (!new_hw) {
+               dev_warn(dev, "Unable to locate HW struct for LAG node destination\n");
+               return;
+       }
+
+       numq = ctx->num_lan_q_entries[tc];
+       teid = ctx->sched.vsi_node[tc]->info.node_teid;
+       tmp_teid = le32_to_cpu(teid);
+       parent_teid = ctx->sched.vsi_node[tc]->info.parent_teid;
+       /* if no teid assigned or numq == 0, then this TC is not active */
+       if (!tmp_teid || !numq)
+               return;
+
+       /* suspend VSI subtree for Traffic Class "tc" on
+        * this VF's VSI
+        */
+       if (ice_sched_suspend_resume_elems(&lag->pf->hw, 1, &tmp_teid, true))
+               dev_dbg(dev, "Problem suspending traffic for LAG node move\n");
+
+       /* reconfigure all VF's queues on this Traffic Class
+        * to new port
+        */
+       qbuf_size = struct_size(qbuf, queue_info, numq);
+       qbuf = kzalloc(qbuf_size, GFP_KERNEL);
+       if (!qbuf) {
+               dev_warn(dev, "Failure allocating memory for VF queue recfg buffer\n");
+               goto resume_traffic;
+       }
+
+       /* add the per queue info for the reconfigure command buffer */
+       valq = ice_lag_qbuf_recfg(&lag->pf->hw, qbuf, vsi_num, numq, tc);
+       if (!valq) {
+               dev_dbg(dev, "No valid queues found for LAG failover\n");
+               goto qbuf_none;
+       }
+
+       if (ice_aq_cfg_lan_txq(&lag->pf->hw, qbuf, qbuf_size, valq, oldport,
+                              newport, NULL)) {
+               dev_warn(dev, "Failure to configure queues for LAG failover\n");
+               goto qbuf_err;
+       }
+
+qbuf_none:
+       kfree(qbuf);
+
+       /* find new parent in destination port's tree for VF VSI node on this
+        * Traffic Class
+        */
+       n_prt = ice_lag_get_sched_parent(new_hw, tc);
+       if (!n_prt)
+               goto resume_traffic;
+
+       /* Move Vf's VSI node for this TC to newport's scheduler tree */
+       buf_size = struct_size(buf, teid, 1);
+       buf = kzalloc(buf_size, GFP_KERNEL);
+       if (!buf) {
+               dev_warn(dev, "Failure to alloc memory for VF node failover\n");
+               goto resume_traffic;
+       }
+
+       buf->hdr.src_parent_teid = parent_teid;
+       buf->hdr.dest_parent_teid = n_prt->info.node_teid;
+       buf->hdr.num_elems = cpu_to_le16(1);
+       buf->hdr.mode = ICE_AQC_MOVE_ELEM_MODE_KEEP_OWN;
+       buf->teid[0] = teid;
+
+       if (ice_aq_move_sched_elems(&lag->pf->hw, 1, buf, buf_size, &num_moved,
+                                   NULL))
+               dev_warn(dev, "Failure to move VF nodes for failover\n");
+       else
+               ice_sched_update_parent(n_prt, ctx->sched.vsi_node[tc]);
+
+       kfree(buf);
+       goto resume_traffic;
+
+qbuf_err:
+       kfree(qbuf);
+
+resume_traffic:
+       /* restart traffic for VSI node */
+       if (ice_sched_suspend_resume_elems(&lag->pf->hw, 1, &tmp_teid, false))
+               dev_dbg(dev, "Problem restarting traffic for LAG node move\n");
 }
 
 /**
@@ -274,6 +565,66 @@ ice_lag_move_single_vf_nodes(struct ice_lag *lag, u8 oldport, u8 newport,
  */
 void ice_lag_move_new_vf_nodes(struct ice_vf *vf)
 {
+       struct ice_lag_netdev_list ndlist;
+       struct list_head *tmp, *n;
+       u8 pri_port, act_port;
+       struct ice_lag *lag;
+       struct ice_vsi *vsi;
+       struct ice_pf *pf;
+
+       vsi = ice_get_vf_vsi(vf);
+
+       if (WARN_ON(!vsi))
+               return;
+
+       if (WARN_ON(vsi->type != ICE_VSI_VF))
+               return;
+
+       pf = vf->pf;
+       lag = pf->lag;
+
+       mutex_lock(&pf->lag_mutex);
+       if (!lag->bonded)
+               goto new_vf_unlock;
+
+       pri_port = pf->hw.port_info->lport;
+       act_port = lag->active_port;
+
+       if (lag->upper_netdev) {
+               struct ice_lag_netdev_list *nl;
+               struct net_device *tmp_nd;
+
+               INIT_LIST_HEAD(&ndlist.node);
+               rcu_read_lock();
+               for_each_netdev_in_bond_rcu(lag->upper_netdev, tmp_nd) {
+                       nl = kzalloc(sizeof(*nl), GFP_KERNEL);
+                       if (!nl)
+                               break;
+
+                       nl->netdev = tmp_nd;
+                       list_add(&nl->node, &ndlist.node);
+               }
+               rcu_read_unlock();
+       }
+
+       lag->netdev_head = &ndlist.node;
+
+       if (ice_is_feature_supported(pf, ICE_F_SRIOV_LAG) &&
+           lag->bonded && lag->primary && pri_port != act_port &&
+           !list_empty(lag->netdev_head))
+               ice_lag_move_single_vf_nodes(lag, pri_port, act_port, vsi->idx);
+
+       list_for_each_safe(tmp, n, &ndlist.node) {
+               struct ice_lag_netdev_list *entry;
+
+               entry = list_entry(tmp, struct ice_lag_netdev_list, node);
+               list_del(&entry->node);
+               kfree(entry);
+       }
+       lag->netdev_head = NULL;
+
+new_vf_unlock:
+       mutex_unlock(&pf->lag_mutex);
 }
 
 /**
@@ -308,6 +659,50 @@ static void ice_lag_move_vf_nodes(struct ice_lag *lag, u8 oldport, u8 newport)
 static void
 ice_lag_cfg_cp_fltr(struct ice_lag *lag, bool add)
 {
+       struct ice_sw_rule_lkup_rx_tx *s_rule = NULL;
+       struct ice_vsi *vsi;
+       u16 buf_len, opc;
+
+       vsi = lag->pf->vsi[0];
+
+       buf_len = ICE_SW_RULE_RX_TX_HDR_SIZE(s_rule,
+                                            ICE_LAG_SRIOV_TRAIN_PKT_LEN);
+       s_rule = kzalloc(buf_len, GFP_KERNEL);
+       if (!s_rule) {
+               netdev_warn(lag->netdev, "-ENOMEM error configuring CP filter\n");
+               return;
+       }
+
+       if (add) {
+               s_rule->hdr.type = cpu_to_le16(ICE_AQC_SW_RULES_T_LKUP_RX);
+               s_rule->recipe_id = cpu_to_le16(ICE_LAG_SRIOV_CP_RECIPE);
+               s_rule->src = cpu_to_le16(vsi->port_info->lport);
+               s_rule->act = cpu_to_le32(ICE_FWD_TO_VSI |
+                                         ICE_SINGLE_ACT_LAN_ENABLE |
+                                         ICE_SINGLE_ACT_VALID_BIT |
+                                         ((vsi->vsi_num <<
+                                           ICE_SINGLE_ACT_VSI_ID_S) &
+                                          ICE_SINGLE_ACT_VSI_ID_M));
+               s_rule->hdr_len = cpu_to_le16(ICE_LAG_SRIOV_TRAIN_PKT_LEN);
+               memcpy(s_rule->hdr_data, lacp_train_pkt, LACP_TRAIN_PKT_LEN);
+               opc = ice_aqc_opc_add_sw_rules;
+       } else {
+               opc = ice_aqc_opc_remove_sw_rules;
+               s_rule->index = cpu_to_le16(lag->cp_rule_idx);
+       }
+       if (ice_aq_sw_rules(&lag->pf->hw, s_rule, buf_len, 1, opc, NULL)) {
+               netdev_warn(lag->netdev, "Error %s CP rule for fail-over\n",
+                           add ? "ADDING" : "REMOVING");
+               goto cp_free;
+       }
+
+       if (add)
+               lag->cp_rule_idx = le16_to_cpu(s_rule->index);
+       else
+               lag->cp_rule_idx = 0;
+
+cp_free:
+       kfree(s_rule);
 }
 
 /**
@@ -362,6 +757,100 @@ static void
 ice_lag_reclaim_vf_tc(struct ice_lag *lag, struct ice_hw *src_hw, u16 vsi_num,
                      u8 tc)
 {
+       u16 numq, valq, buf_size, num_moved, qbuf_size;
+       struct device *dev = ice_pf_to_dev(lag->pf);
+       struct ice_aqc_cfg_txqs_buf *qbuf;
+       struct ice_aqc_move_elem *buf;
+       struct ice_sched_node *n_prt;
+       __le32 teid, parent_teid;
+       struct ice_vsi_ctx *ctx;
+       struct ice_hw *hw;
+       u32 tmp_teid;
+
+       hw = &lag->pf->hw;
+       ctx = ice_get_vsi_ctx(hw, vsi_num);
+       if (!ctx) {
+               dev_warn(dev, "Unable to locate VSI context for LAG reclaim\n");
+               return;
+       }
+
+       /* check to see if this VF is enabled on this TC */
+       if (!ctx->sched.vsi_node[tc])
+               return;
+
+       numq = ctx->num_lan_q_entries[tc];
+       teid = ctx->sched.vsi_node[tc]->info.node_teid;
+       tmp_teid = le32_to_cpu(teid);
+       parent_teid = ctx->sched.vsi_node[tc]->info.parent_teid;
+
+       /* if !teid or !numq, then this TC is not active */
+       if (!tmp_teid || !numq)
+               return;
+
+       /* suspend traffic */
+       if (ice_sched_suspend_resume_elems(hw, 1, &tmp_teid, true))
+               dev_dbg(dev, "Problem suspending traffic for LAG node move\n");
+
+       /* reconfig queues for new port */
+       qbuf_size = struct_size(qbuf, queue_info, numq);
+       qbuf = kzalloc(qbuf_size, GFP_KERNEL);
+       if (!qbuf) {
+               dev_warn(dev, "Failure allocating memory for VF queue recfg buffer\n");
+               goto resume_reclaim;
+       }
+
+       /* add the per queue info for the reconfigure command buffer */
+       valq = ice_lag_qbuf_recfg(hw, qbuf, vsi_num, numq, tc);
+       if (!valq) {
+               dev_dbg(dev, "No valid queues found for LAG reclaim\n");
+               goto reclaim_none;
+       }
+
+       if (ice_aq_cfg_lan_txq(hw, qbuf, qbuf_size, numq,
+                              src_hw->port_info->lport, hw->port_info->lport,
+                              NULL)) {
+               dev_warn(dev, "Failure to configure queues for LAG failover\n");
+               goto reclaim_qerr;
+       }
+
+reclaim_none:
+       kfree(qbuf);
+
+       /* find parent in primary tree */
+       n_prt = ice_lag_get_sched_parent(hw, tc);
+       if (!n_prt)
+               goto resume_reclaim;
+
+       /* Move node to new parent */
+       buf_size = struct_size(buf, teid, 1);
+       buf = kzalloc(buf_size, GFP_KERNEL);
+       if (!buf) {
+               dev_warn(dev, "Failure to alloc memory for VF node failover\n");
+               goto resume_reclaim;
+       }
+
+       buf->hdr.src_parent_teid = parent_teid;
+       buf->hdr.dest_parent_teid = n_prt->info.node_teid;
+       buf->hdr.num_elems = cpu_to_le16(1);
+       buf->hdr.mode = ICE_AQC_MOVE_ELEM_MODE_KEEP_OWN;
+       buf->teid[0] = teid;
+
+       if (ice_aq_move_sched_elems(&lag->pf->hw, 1, buf, buf_size, &num_moved,
+                                   NULL))
+               dev_warn(dev, "Failure to move VF nodes for LAG reclaim\n");
+       else
+               ice_sched_update_parent(n_prt, ctx->sched.vsi_node[tc]);
+
+       kfree(buf);
+       goto resume_reclaim;
+
+reclaim_qerr:
+       kfree(qbuf);
+
+resume_reclaim:
+       /* restart traffic */
+       if (ice_sched_suspend_resume_elems(hw, 1, &tmp_teid, false))
+               dev_warn(dev, "Problem restarting traffic for LAG node reclaim\n");
 }
 
 /**
@@ -476,6 +965,65 @@ static void
 ice_lag_set_swid(u16 primary_swid, struct ice_lag *local_lag,
                 bool link)
 {
+       struct ice_aqc_alloc_free_res_elem *buf;
+       struct ice_aqc_set_port_params *cmd;
+       struct ice_aq_desc desc;
+       u16 buf_len, swid;
+       int status;
+
+       buf_len = struct_size(buf, elem, 1);
+       buf = kzalloc(buf_len, GFP_KERNEL);
+       if (!buf) {
+               dev_err(ice_pf_to_dev(local_lag->pf), "-ENOMEM error setting SWID\n");
+               return;
+       }
+
+       buf->num_elems = cpu_to_le16(1);
+       buf->res_type = cpu_to_le16(ICE_AQC_RES_TYPE_SWID);
+       /* if unlinnking need to free the shared resource */
+       if (!link && local_lag->bond_swid) {
+               buf->elem[0].e.sw_resp = cpu_to_le16(local_lag->bond_swid);
+               status = ice_aq_alloc_free_res(&local_lag->pf->hw, 1, buf,
+                                              buf_len, ice_aqc_opc_free_res,
+                                              NULL);
+               if (status)
+                       dev_err(ice_pf_to_dev(local_lag->pf), "Error freeing SWID during LAG unlink\n");
+               local_lag->bond_swid = 0;
+       }
+
+       if (link) {
+               buf->res_type |=  cpu_to_le16(ICE_LAG_RES_SHARED |
+                                             ICE_LAG_RES_VALID);
+               /* store the primary's SWID in case it leaves bond first */
+               local_lag->bond_swid = primary_swid;
+               buf->elem[0].e.sw_resp = cpu_to_le16(local_lag->bond_swid);
+       } else {
+               buf->elem[0].e.sw_resp =
+                       cpu_to_le16(local_lag->pf->hw.port_info->sw_id);
+       }
+
+       status = ice_aq_alloc_free_res(&local_lag->pf->hw, 1, buf, buf_len,
+                                      ice_aqc_opc_alloc_res, NULL);
+       if (status)
+               dev_err(ice_pf_to_dev(local_lag->pf), "Error subscribing to SWID 0x%04X\n",
+                       local_lag->bond_swid);
+
+       kfree(buf);
+
+       /* Configure port param SWID to correct value */
+       if (link)
+               swid = primary_swid;
+       else
+               swid = local_lag->pf->hw.port_info->sw_id;
+
+       cmd = &desc.params.set_port_params;
+       ice_fill_dflt_direct_cmd_desc(&desc, ice_aqc_opc_set_port_params);
+
+       cmd->swid = cpu_to_le16(ICE_AQC_PORT_SWID_VALID | swid);
+       status = ice_aq_send_cmd(&local_lag->pf->hw, &desc, NULL, 0, NULL);
+       if (status)
+               dev_err(ice_pf_to_dev(local_lag->pf), "Error setting SWID in port params %d\n",
+                       status);
 }
 
 /**
@@ -504,6 +1052,38 @@ static void ice_lag_primary_swid(struct ice_lag *lag, bool link)
  */
 static void ice_lag_add_prune_list(struct ice_lag *lag, struct ice_pf *event_pf)
 {
+       u16 num_vsi, rule_buf_sz, vsi_list_id, event_vsi_num, prim_vsi_idx;
+       struct ice_sw_rule_vsi_list *s_rule = NULL;
+       struct device *dev;
+
+       num_vsi = 1;
+
+       dev = ice_pf_to_dev(lag->pf);
+       event_vsi_num = event_pf->vsi[0]->vsi_num;
+       prim_vsi_idx = lag->pf->vsi[0]->idx;
+
+       if (!ice_find_vsi_list_entry(&lag->pf->hw, ICE_SW_LKUP_VLAN,
+                                    prim_vsi_idx, &vsi_list_id)) {
+               dev_warn(dev, "Could not locate prune list when setting up SRIOV LAG\n");
+               return;
+       }
+
+       rule_buf_sz = (u16)ICE_SW_RULE_VSI_LIST_SIZE(s_rule, num_vsi);
+       s_rule = kzalloc(rule_buf_sz, GFP_KERNEL);
+       if (!s_rule) {
+               dev_warn(dev, "Error allocating space for prune list when configuring SRIOV LAG\n");
+               return;
+       }
+
+       s_rule->hdr.type = cpu_to_le16(ICE_AQC_SW_RULES_T_PRUNE_LIST_SET);
+       s_rule->index = cpu_to_le16(vsi_list_id);
+       s_rule->number_vsi = cpu_to_le16(num_vsi);
+       s_rule->vsi[0] = cpu_to_le16(event_vsi_num);
+
+       if (ice_aq_sw_rules(&event_pf->hw, s_rule, rule_buf_sz, 1,
+                           ice_aqc_opc_update_sw_rules, NULL))
+               dev_warn(dev, "Error adding VSI prune list\n");
+       kfree(s_rule);
 }
 
 /**
@@ -513,6 +1093,39 @@ static void ice_lag_add_prune_list(struct ice_lag *lag, struct ice_pf *event_pf)
  */
 static void ice_lag_del_prune_list(struct ice_lag *lag, struct ice_pf *event_pf)
 {
+       u16 num_vsi, vsi_num, vsi_idx, rule_buf_sz, vsi_list_id;
+       struct ice_sw_rule_vsi_list *s_rule = NULL;
+       struct device *dev;
+
+       num_vsi = 1;
+
+       dev = ice_pf_to_dev(lag->pf);
+       vsi_num = event_pf->vsi[0]->vsi_num;
+       vsi_idx = lag->pf->vsi[0]->idx;
+
+       if (!ice_find_vsi_list_entry(&lag->pf->hw, ICE_SW_LKUP_VLAN,
+                                    vsi_idx, &vsi_list_id)) {
+               dev_warn(dev, "Could not locate prune list when unwinding SRIOV LAG\n");
+               return;
+       }
+
+       rule_buf_sz = (u16)ICE_SW_RULE_VSI_LIST_SIZE(s_rule, num_vsi);
+       s_rule = kzalloc(rule_buf_sz, GFP_KERNEL);
+       if (!s_rule) {
+               dev_warn(dev, "Error allocating prune list when unwinding SRIOV LAG\n");
+               return;
+       }
+
+       s_rule->hdr.type = cpu_to_le16(ICE_AQC_SW_RULES_T_PRUNE_LIST_CLEAR);
+       s_rule->index = cpu_to_le16(vsi_list_id);
+       s_rule->number_vsi = cpu_to_le16(num_vsi);
+       s_rule->vsi[0] = cpu_to_le16(vsi_num);
+
+       if (ice_aq_sw_rules(&event_pf->hw, (struct ice_aqc_sw_rules *)s_rule,
+                           rule_buf_sz, 1, ice_aqc_opc_update_sw_rules, NULL))
+               dev_warn(dev, "Error clearing VSI prune list\n");
+
+       kfree(s_rule);
 }
 
 /**
@@ -539,8 +1152,6 @@ static void ice_lag_init_feature_support_flag(struct ice_pf *pf)
  * ice_lag_changeupper_event - handle LAG changeupper event
  * @lag: LAG info struct
  * @ptr: opaque pointer data
- *
- * ptr is to be cast into netdev_notifier_changeupper_info
  */
 static void ice_lag_changeupper_event(struct ice_lag *lag, void *ptr)
 {
@@ -600,6 +1211,40 @@ static void ice_lag_changeupper_event(struct ice_lag *lag, void *ptr)
  */
 static void ice_lag_monitor_link(struct ice_lag *lag, void *ptr)
 {
+       struct netdev_notifier_changeupper_info *info;
+       struct ice_hw *prim_hw, *active_hw;
+       struct net_device *event_netdev;
+       struct ice_pf *pf;
+       u8 prim_port;
+
+       if (!lag->primary)
+               return;
+
+       event_netdev = netdev_notifier_info_to_dev(ptr);
+       if (!netif_is_same_ice(lag->pf, event_netdev))
+               return;
+
+       pf = lag->pf;
+       prim_hw = &pf->hw;
+       prim_port = prim_hw->port_info->lport;
+
+       info = (struct netdev_notifier_changeupper_info *)ptr;
+       if (info->upper_dev != lag->upper_netdev)
+               return;
+
+       if (!info->linking) {
+               /* Since there are only two interfaces allowed in SRIOV+LAG, if
+                * one port is leaving, then nodes need to be on primary
+                * interface.
+                */
+               if (prim_port != lag->active_port &&
+                   lag->active_port != ICE_LAG_INVALID_PORT) {
+                       active_hw = ice_lag_find_hw_by_lport(lag,
+                                                            lag->active_port);
+                       ice_lag_reclaim_vf_nodes(lag, active_hw);
+                       lag->active_port = ICE_LAG_INVALID_PORT;
+               }
+       }
 }
 
 /**
@@ -612,6 +1257,67 @@ static void ice_lag_monitor_link(struct ice_lag *lag, void *ptr)
  */
 static void ice_lag_monitor_active(struct ice_lag *lag, void *ptr)
 {
+       struct net_device *event_netdev, *event_upper;
+       struct netdev_notifier_bonding_info *info;
+       struct netdev_bonding_info *bonding_info;
+       struct ice_netdev_priv *event_np;
+       struct ice_pf *pf, *event_pf;
+       u8 prim_port, event_port;
+
+       if (!lag->primary)
+               return;
+
+       pf = lag->pf;
+       if (!pf)
+               return;
+
+       event_netdev = netdev_notifier_info_to_dev(ptr);
+       rcu_read_lock();
+       event_upper = netdev_master_upper_dev_get_rcu(event_netdev);
+       rcu_read_unlock();
+       if (!netif_is_ice(event_netdev) || event_upper != lag->upper_netdev)
+               return;
+
+       event_np = netdev_priv(event_netdev);
+       event_pf = event_np->vsi->back;
+       event_port = event_pf->hw.port_info->lport;
+       prim_port = pf->hw.port_info->lport;
+
+       info = (struct netdev_notifier_bonding_info *)ptr;
+       bonding_info = &info->bonding_info;
+
+       if (!bonding_info->slave.state) {
+               /* if no port is currently active, then nodes and filters exist
+                * on primary port, check if we need to move them
+                */
+               if (lag->active_port == ICE_LAG_INVALID_PORT) {
+                       if (event_port != prim_port)
+                               ice_lag_move_vf_nodes(lag, prim_port,
+                                                     event_port);
+                       lag->active_port = event_port;
+                       return;
+               }
+
+               /* active port is already set and is current event port */
+               if (lag->active_port == event_port)
+                       return;
+               /* new active port */
+               ice_lag_move_vf_nodes(lag, lag->active_port, event_port);
+               lag->active_port = event_port;
+       } else {
+               /* port not set as currently active (e.g. new active port
+                * has already claimed the nodes and filters
+                */
+               if (lag->active_port != event_port)
+                       return;
+               /* This is the case when neither port is active (both link down)
+                * Link down on the bond - set active port to invalid and move
+                * nodes and filters back to primary if not already there
+                */
+               if (event_port != prim_port)
+                       ice_lag_move_vf_nodes(lag, event_port, prim_port);
+               lag->active_port = ICE_LAG_INVALID_PORT;
+       }
 }
 
 /**
@@ -622,6 +1328,68 @@ static void ice_lag_monitor_active(struct ice_lag *lag, void *ptr)
 static bool
 ice_lag_chk_comp(struct ice_lag *lag, void *ptr)
 {
+       struct net_device *event_netdev, *event_upper;
+       struct netdev_notifier_bonding_info *info;
+       struct netdev_bonding_info *bonding_info;
+       struct list_head *tmp;
+       int count = 0;
+
+       if (!lag->primary)
+               return true;
+
+       event_netdev = netdev_notifier_info_to_dev(ptr);
+       rcu_read_lock();
+       event_upper = netdev_master_upper_dev_get_rcu(event_netdev);
+       rcu_read_unlock();
+       if (event_upper != lag->upper_netdev)
+               return true;
+
+       info = (struct netdev_notifier_bonding_info *)ptr;
+       bonding_info = &info->bonding_info;
+       lag->bond_mode = bonding_info->master.bond_mode;
+       if (lag->bond_mode != BOND_MODE_ACTIVEBACKUP) {
+               netdev_info(lag->netdev, "Bond Mode not ACTIVE-BACKUP\n");
+               return false;
+       }
+
+       list_for_each(tmp, lag->netdev_head) {
+               struct ice_dcbx_cfg *dcb_cfg, *peer_dcb_cfg;
+               struct ice_lag_netdev_list *entry;
+               struct ice_netdev_priv *peer_np;
+               struct net_device *peer_netdev;
+               struct ice_vsi *vsi, *peer_vsi;
+
+               entry = list_entry(tmp, struct ice_lag_netdev_list, node);
+               peer_netdev = entry->netdev;
+               if (!netif_is_ice(peer_netdev)) {
+                       netdev_info(lag->netdev, "Found non-ice netdev in LAG\n");
+                       return false;
+               }
+
+               count++;
+               if (count > 2) {
+                       netdev_info(lag->netdev, "Found more than two netdevs in LAG\n");
+                       return false;
+               }
+
+               peer_np = netdev_priv(peer_netdev);
+               vsi = ice_get_main_vsi(lag->pf);
+               peer_vsi = peer_np->vsi;
+               if (lag->pf->pdev->bus != peer_vsi->back->pdev->bus ||
+                   lag->pf->pdev->slot != peer_vsi->back->pdev->slot) {
+                       netdev_info(lag->netdev, "Found netdev on different device in LAG\n");
+                       return false;
+               }
+
+               dcb_cfg = &vsi->port_info->qos_cfg.local_dcbx_cfg;
+               peer_dcb_cfg = &peer_vsi->port_info->qos_cfg.local_dcbx_cfg;
+               if (memcmp(dcb_cfg, peer_dcb_cfg,
+                          sizeof(struct ice_dcbx_cfg))) {
+                       netdev_info(lag->netdev, "Found netdev with different DCB config in LAG\n");
+                       return false;
+               }
+       }
+
        return true;
 }
 
@@ -856,17 +1624,40 @@ static void ice_unregister_lag_handler(struct ice_lag *lag)
 /**
  * ice_create_lag_recipe
  * @hw: pointer to HW struct
+ * @rid: pointer to u16 to pass back recipe index
  * @base_recipe: recipe to base the new recipe on
  * @prio: priority for new recipe
  *
  * function returns 0 on error
  */
-static u16 ice_create_lag_recipe(struct ice_hw *hw, const u8 *base_recipe,
-                                u8 prio)
+static int ice_create_lag_recipe(struct ice_hw *hw, u16 *rid,
+                                const u8 *base_recipe, u8 prio)
 {
-       u16 rid = 0;
+       struct ice_aqc_recipe_data_elem *new_rcp;
+       int err;
+
+       err = ice_alloc_recipe(hw, rid);
+       if (err)
+               return err;
+
+       new_rcp = kzalloc(ICE_RECIPE_LEN * ICE_MAX_NUM_RECIPES, GFP_KERNEL);
+       if (!new_rcp)
+               return -ENOMEM;
+
+       memcpy(new_rcp, base_recipe, ICE_RECIPE_LEN);
+       new_rcp->content.act_ctrl_fwd_priority = prio;
+       new_rcp->content.rid = *rid | ICE_AQ_RECIPE_ID_IS_ROOT;
+       new_rcp->recipe_indx = *rid;
+       bitmap_zero((unsigned long *)new_rcp->recipe_bitmap,
+                   ICE_MAX_NUM_RECIPES);
+       set_bit(*rid, (unsigned long *)new_rcp->recipe_bitmap);
 
-       return rid;
+       err = ice_aq_add_recipe(hw, new_rcp, 1, NULL);
+       if (err)
+               *rid = 0;
+
+       kfree(new_rcp);
+       return err;
 }
 
 /**
@@ -881,7 +1672,8 @@ int ice_init_lag(struct ice_pf *pf)
        struct device *dev = ice_pf_to_dev(pf);
        struct ice_lag *lag;
        struct ice_vsi *vsi;
-       int err;
+       u64 recipe_bits = 0;
+       int n, err;
 
        ice_lag_init_feature_support_flag(pf);
 
@@ -900,6 +1692,7 @@ int ice_init_lag(struct ice_pf *pf)
        lag->pf = pf;
        lag->netdev = vsi->netdev;
        lag->role = ICE_LAG_NONE;
+       lag->active_port = ICE_LAG_INVALID_PORT;
        lag->bonded = false;
        lag->upper_netdev = NULL;
        lag->notif_block.notifier_call = NULL;
@@ -910,10 +1703,23 @@ int ice_init_lag(struct ice_pf *pf)
                goto lag_error;
        }
 
-       lag->pf_recipe = ice_create_lag_recipe(&pf->hw, ice_dflt_vsi_rcp, 1);
-       if (!lag->pf_recipe) {
-               err = -EINVAL;
+       err = ice_create_lag_recipe(&pf->hw, &lag->pf_recipe, ice_dflt_vsi_rcp,
+                                   1);
+       if (err)
                goto lag_error;
+
+       /* associate recipes to profiles */
+       for (n = 0; n < ICE_PROFID_IPV6_GTPU_IPV6_TCP_INNER; n++) {
+               err = ice_aq_get_recipe_to_profile(&pf->hw, n,
+                                                  (u8 *)&recipe_bits, NULL);
+               if (err)
+                       continue;
+
+               if (recipe_bits & BIT(ICE_SW_LKUP_DFLT)) {
+                       recipe_bits |= BIT(lag->pf_recipe);
+                       ice_aq_map_recipe_to_profile(&pf->hw, n,
+                                                    (u8 *)&recipe_bits, NULL);
+               }
        }
 
        ice_display_lag_info(lag);