ice: Add infrastructure for mqprio support via ndo_setup_tc

Add infrastructure required for "ndo_setup_tc:qdisc_mqprio".
ice_vsi_setup is modified to configure traffic classes based
on mqprio data received from the stack. This includes low-level
functions to configure min, max rate-limit parameters in hardware
for traffic classes. Each traffic class gets mapped to a hardware
channel (VSI) which can be individually configured with different
bandwidth parameters.
Co-developed-by: NTarun Singh <tarun.k.singh@intel.com>
Signed-off-by: NTarun Singh <tarun.k.singh@intel.com>
Signed-off-by: NKiran Patil <kiran.patil@intel.com>
Signed-off-by: NAmritha Nambiar <amritha.nambiar@intel.com>
Signed-off-by: NSudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Tested-by: NBharathi Sreenivas <bharathi.sreenivas@intel.com>
Signed-off-by: NTony Nguyen <anthony.l.nguyen@intel.com>

drivers/net/ethernet/intel/ice/ice.h

@@ -38,6 +38,7 @@
 #include <linux/avf/virtchnl.h>
 #include <linux/cpu_rmap.h>
 #include <linux/dim.h>
+#include <net/pkt_cls.h>
 #include <net/devlink.h>
 #include <net/ipv6.h>
 #include <net/xdp_sock.h>
@@ -104,6 +105,10 @@
 #define ICE_INVAL_VFID		256
 
 #define ICE_MAX_RXQS_PER_TC		256	/* Used when setting VSI context per TC Rx queues */
+
+#define ICE_CHNL_START_TC		1
+#define ICE_CHNL_MAX_TC			16
+
 #define ICE_MAX_RESET_WAIT		20
 
 #define ICE_VSIQF_HKEY_ARRAY_SIZE	((VSIQF_HKEY_MAX_INDEX + 1) *	4)
@@ -145,6 +150,9 @@
 #define ice_for_each_q_vector(vsi, i) \
 	for ((i) = 0; (i) < (vsi)->num_q_vectors; (i)++)
 
+#define ice_for_each_chnl_tc(i)	\
+	for ((i) = ICE_CHNL_START_TC; (i) < ICE_CHNL_MAX_TC; (i)++)
+
 #define ICE_UCAST_PROMISC_BITS (ICE_PROMISC_UCAST_TX | ICE_PROMISC_MCAST_TX | \
 				ICE_PROMISC_UCAST_RX | ICE_PROMISC_MCAST_RX)
 
@@ -172,6 +180,21 @@ enum ice_feature {
 
 DECLARE_STATIC_KEY_FALSE(ice_xdp_locking_key);
 
+struct ice_channel {
+	struct list_head list;
+	u8 type;
+	u16 sw_id;
+	u16 base_q;
+	u16 num_rxq;
+	u16 num_txq;
+	u16 vsi_num;
+	u8 ena_tc;
+	struct ice_aqc_vsi_props info;
+	u64 max_tx_rate;
+	u64 min_tx_rate;
+	struct ice_vsi *ch_vsi;
+};
+
 struct ice_txq_meta {
 	u32 q_teid;	/* Tx-scheduler element identifier */
 	u16 q_id;	/* Entry in VSI's txq_map bitmap */
@@ -189,7 +212,7 @@ struct ice_tc_info {
 
 struct ice_tc_cfg {
 	u8 numtc; /* Total number of enabled TCs */
-	u8 ena_tc; /* Tx map */
+	u16 ena_tc; /* Tx map */
 	struct ice_tc_info tc_info[ICE_MAX_TRAFFIC_CLASS];
 };
 
@@ -361,6 +384,34 @@ struct ice_vsi {
 
 	struct net_device **target_netdevs;
 
+	struct tc_mqprio_qopt_offload mqprio_qopt; /* queue parameters */
+
+	/* Channel Specific Fields */
+	struct ice_vsi *tc_map_vsi[ICE_CHNL_MAX_TC];
+	u16 cnt_q_avail;
+	u16 next_base_q;	/* next queue to be used for channel setup */
+	struct list_head ch_list;
+	u16 num_chnl_rxq;
+	u16 num_chnl_txq;
+	u16 ch_rss_size;
+	/* store away rss size info before configuring ADQ channels so that,
+	 * it can be used after tc-qdisc delete, to get back RSS setting as
+	 * they were before
+	 */
+	u16 orig_rss_size;
+	/* this keeps tracks of all enabled TC with and without DCB
+	 * and inclusive of ADQ, vsi->mqprio_opt keeps track of queue
+	 * information
+	 */
+	u8 all_numtc;
+	u16 all_enatc;
+
+	/* store away TC info, to be used for rebuild logic */
+	u8 old_numtc;
+	u16 old_ena_tc;
+
+	struct ice_channel *ch;
+
 	/* setup back reference, to which aggregator node this VSI
 	 * corresponds to
 	 */
@@ -407,6 +458,7 @@ enum ice_pf_flags {
 	ICE_FLAG_PTP,			/* PTP is enabled by software */
 	ICE_FLAG_AUX_ENA,
 	ICE_FLAG_ADV_FEATURES,
+	ICE_FLAG_TC_MQPRIO,		/* support for Multi queue TC */
 	ICE_FLAG_CLS_FLOWER,
 	ICE_FLAG_LINK_DOWN_ON_CLOSE_ENA,
 	ICE_FLAG_TOTAL_PORT_SHUTDOWN_ENA,
@@ -704,6 +756,30 @@ static inline void ice_clear_sriov_cap(struct ice_pf *pf)
 			((base_idx) * ICE_FD_STAT_CTR_BLOCK_COUNT)
 #define ICE_FD_SB_STAT_IDX(base_idx) ICE_FD_STAT_PF_IDX(base_idx)
 
+/**
+ * ice_is_adq_active - any active ADQs
+ * @pf: pointer to PF
+ *
+ * This function returns true if there are any ADQs configured (which is
+ * determined by looking at VSI type (which should be VSI_PF), numtc, and
+ * TC_MQPRIO flag) otherwise return false
+ */
+static inline bool ice_is_adq_active(struct ice_pf *pf)
+{
+	struct ice_vsi *vsi;
+
+	vsi = ice_get_main_vsi(pf);
+	if (!vsi)
+		return false;
+
+	/* is ADQ configured */
+	if (vsi->tc_cfg.numtc > ICE_CHNL_START_TC &&
+	    test_bit(ICE_FLAG_TC_MQPRIO, pf->flags))
+		return true;
+
+	return false;
+}
+
 bool netif_is_ice(struct net_device *dev);
 int ice_vsi_setup_tx_rings(struct ice_vsi *vsi);
 int ice_vsi_setup_rx_rings(struct ice_vsi *vsi);

drivers/net/ethernet/intel/ice/ice_base.c

@@ -213,6 +213,9 @@ static u16 ice_calc_txq_handle(struct ice_vsi *vsi, struct ice_tx_ring *ring, u8
 {
 	WARN_ONCE(ice_ring_is_xdp(ring) && tc, "XDP ring can't belong to TC other than 0\n");
 
+	if (ring->ch)
+		return ring->q_index - ring->ch->base_q;
+
 	/* Idea here for calculation is that we subtract the number of queue
 	 * count from TC that ring belongs to from it's absolute queue index
 	 * and as a result we get the queue's index within TC.
@@ -300,7 +303,10 @@ ice_setup_tx_ctx(struct ice_tx_ring *ring, struct ice_tlan_ctx *tlan_ctx, u16 pf
 	case ICE_VSI_LB:
 	case ICE_VSI_CTRL:
 	case ICE_VSI_PF:
-		tlan_ctx->vmvf_type = ICE_TLAN_CTX_VMVF_TYPE_PF;
+		if (ring->ch)
+			tlan_ctx->vmvf_type = ICE_TLAN_CTX_VMVF_TYPE_VMQ;
+		else
+			tlan_ctx->vmvf_type = ICE_TLAN_CTX_VMVF_TYPE_PF;
 		break;
 	case ICE_VSI_VF:
 		/* Firmware expects vmvf_num to be absolute VF ID */
@@ -315,7 +321,10 @@ ice_setup_tx_ctx(struct ice_tx_ring *ring, struct ice_tlan_ctx *tlan_ctx, u16 pf
 	}
 
 	/* make sure the context is associated with the right VSI */
-	tlan_ctx->src_vsi = ice_get_hw_vsi_num(hw, vsi->idx);
+	if (ring->ch)
+		tlan_ctx->src_vsi = ring->ch->vsi_num;
+	else
+		tlan_ctx->src_vsi = ice_get_hw_vsi_num(hw, vsi->idx);
 
 	/* Restrict Tx timestamps to the PF VSI */
 	switch (vsi->type) {
@@ -747,6 +756,7 @@ ice_vsi_cfg_txq(struct ice_vsi *vsi, struct ice_tx_ring *ring,
 	u8 buf_len = struct_size(qg_buf, txqs, 1);
 	struct ice_tlan_ctx tlan_ctx = { 0 };
 	struct ice_aqc_add_txqs_perq *txq;
+	struct ice_channel *ch = ring->ch;
 	struct ice_pf *pf = vsi->back;
 	struct ice_hw *hw = &pf->hw;
 	enum ice_status status;
@@ -785,8 +795,14 @@ ice_vsi_cfg_txq(struct ice_vsi *vsi, struct ice_tx_ring *ring,
 		ring->q_handle = ice_calc_txq_handle(vsi, ring, tc);
 	}
 
-	status = ice_ena_vsi_txq(vsi->port_info, vsi->idx, tc, ring->q_handle,
-				 1, qg_buf, buf_len, NULL);
+	if (ch)
+		status = ice_ena_vsi_txq(vsi->port_info, ch->ch_vsi->idx, 0,
+					 ring->q_handle, 1, qg_buf, buf_len,
+					 NULL);
+	else
+		status = ice_ena_vsi_txq(vsi->port_info, vsi->idx, tc,
+					 ring->q_handle, 1, qg_buf, buf_len,
+					 NULL);
 	if (status) {
 		dev_err(ice_pf_to_dev(pf), "Failed to set LAN Tx queue context, error: %s\n",
 			ice_stat_str(status));
@@ -967,6 +983,7 @@ void
 ice_fill_txq_meta(struct ice_vsi *vsi, struct ice_tx_ring *ring,
 		  struct ice_txq_meta *txq_meta)
 {
+	struct ice_channel *ch = ring->ch;
 	u8 tc;
 
 	if (IS_ENABLED(CONFIG_DCB))
@@ -977,6 +994,11 @@ ice_fill_txq_meta(struct ice_vsi *vsi, struct ice_tx_ring *ring,
 	txq_meta->q_id = ring->reg_idx;
 	txq_meta->q_teid = ring->txq_teid;
 	txq_meta->q_handle = ring->q_handle;
-	txq_meta->vsi_idx = vsi->idx;
-	txq_meta->tc = tc;
+	if (ch) {
+		txq_meta->vsi_idx = ch->ch_vsi->idx;
+		txq_meta->tc = 0;
+	} else {
+		txq_meta->vsi_idx = vsi->idx;
+		txq_meta->tc = tc;
+	}
 }

drivers/net/ethernet/intel/ice/ice_dcb_lib.c

@@ -4,53 +4,11 @@
 #include "ice_dcb_lib.h"
 #include "ice_dcb_nl.h"
 
-/**
- * ice_vsi_cfg_netdev_tc - Setup the netdev TC configuration
- * @vsi: the VSI being configured
- * @ena_tc: TC map to be enabled
- */
-void ice_vsi_cfg_netdev_tc(struct ice_vsi *vsi, u8 ena_tc)
-{
-	struct net_device *netdev = vsi->netdev;
-	struct ice_pf *pf = vsi->back;
-	struct ice_dcbx_cfg *dcbcfg;
-	u8 netdev_tc;
-	int i;
-
-	if (!netdev)
-		return;
-
-	if (!ena_tc) {
-		netdev_reset_tc(netdev);
-		return;
-	}
-
-	if (netdev_set_num_tc(netdev, vsi->tc_cfg.numtc))
-		return;
-
-	dcbcfg = &pf->hw.port_info->qos_cfg.local_dcbx_cfg;
-
-	ice_for_each_traffic_class(i)
-		if (vsi->tc_cfg.ena_tc & BIT(i))
-			netdev_set_tc_queue(netdev,
-					    vsi->tc_cfg.tc_info[i].netdev_tc,
-					    vsi->tc_cfg.tc_info[i].qcount_tx,
-					    vsi->tc_cfg.tc_info[i].qoffset);
-
-	for (i = 0; i < ICE_MAX_USER_PRIORITY; i++) {
-		u8 ets_tc = dcbcfg->etscfg.prio_table[i];
-
-		/* Get the mapped netdev TC# for the UP */
-		netdev_tc = vsi->tc_cfg.tc_info[ets_tc].netdev_tc;
-		netdev_set_prio_tc_map(netdev, i, netdev_tc);
-	}
-}
-
 /**
  * ice_dcb_get_ena_tc - return bitmap of enabled TCs
  * @dcbcfg: DCB config to evaluate for enabled TCs
  */
-u8 ice_dcb_get_ena_tc(struct ice_dcbx_cfg *dcbcfg)
+static u8 ice_dcb_get_ena_tc(struct ice_dcbx_cfg *dcbcfg)
 {
 	u8 i, num_tc, ena_tc = 1;
 
@@ -178,6 +136,67 @@ u8 ice_dcb_get_num_tc(struct ice_dcbx_cfg *dcbcfg)
 	return ret;
 }
 
+/**
+ * ice_get_first_droptc - returns number of first droptc
+ * @vsi: used to find the first droptc
+ *
+ * This function returns the value of first_droptc.
+ * When DCB is enabled, first droptc information is derived from enabled_tc
+ * and PFC enabled bits. otherwise this function returns 0 as there is one
+ * TC without DCB (tc0)
+ */
+static u8 ice_get_first_droptc(struct ice_vsi *vsi)
+{
+	struct ice_dcbx_cfg *cfg = &vsi->port_info->qos_cfg.local_dcbx_cfg;
+	struct device *dev = ice_pf_to_dev(vsi->back);
+	u8 num_tc, ena_tc_map, pfc_ena_map;
+	u8 i;
+
+	num_tc = ice_dcb_get_num_tc(cfg);
+
+	/* get bitmap of enabled TCs */
+	ena_tc_map = ice_dcb_get_ena_tc(cfg);
+
+	/* get bitmap of PFC enabled TCs */
+	pfc_ena_map = cfg->pfc.pfcena;
+
+	/* get first TC that is not PFC enabled */
+	for (i = 0; i < num_tc; i++) {
+		if ((ena_tc_map & BIT(i)) && (!(pfc_ena_map & BIT(i)))) {
+			dev_dbg(dev, "first drop tc = %d\n", i);
+			return i;
+		}
+	}
+
+	dev_dbg(dev, "first drop tc = 0\n");
+	return 0;
+}
+
+/**
+ * ice_vsi_set_dcb_tc_cfg - Set VSI's TC based on DCB configuration
+ * @vsi: pointer to the VSI instance
+ */
+void ice_vsi_set_dcb_tc_cfg(struct ice_vsi *vsi)
+{
+	struct ice_dcbx_cfg *cfg = &vsi->port_info->qos_cfg.local_dcbx_cfg;
+
+	switch (vsi->type) {
+	case ICE_VSI_PF:
+		vsi->tc_cfg.ena_tc = ice_dcb_get_ena_tc(cfg);
+		vsi->tc_cfg.numtc = ice_dcb_get_num_tc(cfg);
+		break;
+	case ICE_VSI_CHNL:
+		vsi->tc_cfg.ena_tc = BIT(ice_get_first_droptc(vsi));
+		vsi->tc_cfg.numtc = 1;
+		break;
+	case ICE_VSI_CTRL:
+	case ICE_VSI_LB:
+	default:
+		vsi->tc_cfg.ena_tc = ICE_DFLT_TRAFFIC_CLASS;
+		vsi->tc_cfg.numtc = 1;
+	}
+}
+
 /**
  * ice_dcb_get_tc - Get the TC associated with the queue
  * @vsi: ptr to the VSI
@@ -218,11 +237,68 @@ void ice_vsi_cfg_dcb_rings(struct ice_vsi *vsi)
 
 		qoffset = vsi->tc_cfg.tc_info[n].qoffset;
 		qcount = vsi->tc_cfg.tc_info[n].qcount_tx;
-		for (i = qoffset; i < (qoffset + qcount); i++) {
-			tx_ring = vsi->tx_rings[i];
-			rx_ring = vsi->rx_rings[i];
-			tx_ring->dcb_tc = n;
-			rx_ring->dcb_tc = n;
+		for (i = qoffset; i < (qoffset + qcount); i++)
+			vsi->tx_rings[i]->dcb_tc = n;
+
+		qcount = vsi->tc_cfg.tc_info[n].qcount_rx;
+		for (i = qoffset; i < (qoffset + qcount); i++)
+			vsi->rx_rings[i]->dcb_tc = n;
+	}
+	/* applicable only if "all_enatc" is set, which will be set from
+	 * setup_tc method as part of configuring channels
+	 */
+	if (vsi->all_enatc) {
+		u8 first_droptc = ice_get_first_droptc(vsi);
+
+		/* When DCB is configured, TC for ADQ queues (which are really
+		 * PF queues) should be the first drop TC of the main VSI
+		 */
+		ice_for_each_chnl_tc(n) {
+			if (!(vsi->all_enatc & BIT(n)))
+				break;
+
+			qoffset = vsi->mqprio_qopt.qopt.offset[n];
+			qcount = vsi->mqprio_qopt.qopt.count[n];
+			for (i = qoffset; i < (qoffset + qcount); i++) {
+				vsi->tx_rings[i]->dcb_tc = first_droptc;
+				vsi->rx_rings[i]->dcb_tc = first_droptc;
+			}
+		}
+	}
+}
+
+/**
+ * ice_dcb_ena_dis_vsi - disable certain VSIs for DCB config/reconfig
+ * @pf: pointer to the PF instance
+ * @ena: true to enable VSIs, false to disable
+ * @locked: true if caller holds RTNL lock, false otherwise
+ *
+ * Before a new DCB configuration can be applied, VSIs of type PF, SWITCHDEV
+ * and CHNL need to be brought down. Following completion of DCB configuration
+ * the VSIs that were downed need to be brought up again. This helper function
+ * does both.
+ */
+static void ice_dcb_ena_dis_vsi(struct ice_pf *pf, bool ena, bool locked)
+{
+	int i;
+
+	ice_for_each_vsi(pf, i) {
+		struct ice_vsi *vsi = pf->vsi[i];
+
+		if (!vsi)
+			continue;
+
+		switch (vsi->type) {
+		case ICE_VSI_CHNL:
+		case ICE_VSI_SWITCHDEV_CTRL:
+		case ICE_VSI_PF:
+			if (ena)
+				ice_ena_vsi(vsi, locked);
+			else
+				ice_dis_vsi(vsi, locked);
+			break;
+		default:
+			continue;
 		}
 	}
 }
@@ -331,7 +407,9 @@ int ice_pf_dcb_cfg(struct ice_pf *pf, struct ice_dcbx_cfg *new_cfg, bool locked)
 	 */
 	if (!locked)
 		rtnl_lock();
-	ice_dis_vsi(pf_vsi, true);
+
+	/* disable VSIs affected by DCB changes */
+	ice_dcb_ena_dis_vsi(pf, false, true);
 
 	memcpy(curr_cfg, new_cfg, sizeof(*curr_cfg));
 	memcpy(&curr_cfg->etsrec, &curr_cfg->etscfg, sizeof(curr_cfg->etsrec));
@@ -359,7 +437,8 @@ int ice_pf_dcb_cfg(struct ice_pf *pf, struct ice_dcbx_cfg *new_cfg, bool locked)
 	ice_pf_dcb_recfg(pf);
 
 out:
-	ice_ena_vsi(pf_vsi, true);
+	/* enable previously downed VSIs */
+	ice_dcb_ena_dis_vsi(pf, true, true);
 	if (!locked)
 		rtnl_unlock();
 free_cfg:
@@ -674,6 +753,8 @@ void ice_pf_dcb_recfg(struct ice_pf *pf)
 				tc_map = ICE_DFLT_TRAFFIC_CLASS;
 				ice_dcb_noncontig_cfg(pf);
 			}
+		} else if (vsi->type == ICE_VSI_CHNL) {
+			tc_map = BIT(ice_get_first_droptc(vsi));
 		} else {
 			tc_map = ICE_DFLT_TRAFFIC_CLASS;
 		}
@@ -684,10 +765,11 @@ void ice_pf_dcb_recfg(struct ice_pf *pf)
 				vsi->idx);
 			continue;
 		}
-		/* no need to proceed with remaining cfg if it is switchdev
-		 * VSI
+		/* no need to proceed with remaining cfg if it is CHNL
+		 * or switchdev VSI
 		 */
-		if (vsi->type == ICE_VSI_SWITCHDEV_CTRL)
+		if (vsi->type == ICE_VSI_CHNL ||
+		    vsi->type == ICE_VSI_SWITCHDEV_CTRL)
 			continue;
 
 		ice_vsi_map_rings_to_vectors(vsi);
@@ -862,7 +944,6 @@ ice_dcb_process_lldp_set_mib_change(struct ice_pf *pf,
 	struct ice_dcbx_cfg tmp_dcbx_cfg;
 	bool need_reconfig = false;
 	struct ice_port_info *pi;
-	struct ice_vsi *pf_vsi;
 	u8 mib_type;
 	int ret;
 
@@ -938,14 +1019,9 @@ ice_dcb_process_lldp_set_mib_change(struct ice_pf *pf,
 		clear_bit(ICE_FLAG_DCB_ENA, pf->flags);
 	}
 
-	pf_vsi = ice_get_main_vsi(pf);
-	if (!pf_vsi) {
-		dev_dbg(dev, "PF VSI doesn't exist\n");
-		goto out;
-	}
-
 	rtnl_lock();
-	ice_dis_vsi(pf_vsi, true);
+	/* disable VSIs affected by DCB changes */
+	ice_dcb_ena_dis_vsi(pf, false, true);
 
 	ret = ice_query_port_ets(pf->hw.port_info, &buf, sizeof(buf), NULL);
 	if (ret) {
@@ -956,7 +1032,8 @@ ice_dcb_process_lldp_set_mib_change(struct ice_pf *pf,
 	/* changes in configuration update VSI */
 	ice_pf_dcb_recfg(pf);
 
-	ice_ena_vsi(pf_vsi, true);
+	/* enable previously downed VSIs */
+	ice_dcb_ena_dis_vsi(pf, true, true);
 unlock_rtnl:
 	rtnl_unlock();
 out:

drivers/net/ethernet/intel/ice/ice_dcb_lib.h

@@ -16,7 +16,6 @@
 
 void ice_dcb_rebuild(struct ice_pf *pf);
 int ice_dcb_sw_dflt_cfg(struct ice_pf *pf, bool ets_willing, bool locked);
-u8 ice_dcb_get_ena_tc(struct ice_dcbx_cfg *dcbcfg);
 u8 ice_dcb_get_num_tc(struct ice_dcbx_cfg *dcbcfg);
 void ice_vsi_set_dcb_tc_cfg(struct ice_vsi *vsi);
 bool ice_is_pfc_causing_hung_q(struct ice_pf *pf, unsigned int txqueue);
@@ -34,8 +33,6 @@ ice_tx_prepare_vlan_flags_dcb(struct ice_tx_ring *tx_ring,
 void
 ice_dcb_process_lldp_set_mib_change(struct ice_pf *pf,
 				    struct ice_rq_event_info *event);
-void ice_vsi_cfg_netdev_tc(struct ice_vsi *vsi, u8 ena_tc);
-
 /**
  * ice_find_q_in_range
  * @low: start of queue range for a TC i.e. offset of TC
@@ -69,6 +66,12 @@ static inline u8 ice_get_pfc_mode(struct ice_pf *pf)
 #else
 static inline void ice_dcb_rebuild(struct ice_pf *pf) { }
 
+static inline void ice_vsi_set_dcb_tc_cfg(struct ice_vsi *vsi)
+{
+	vsi->tc_cfg.ena_tc = ICE_DFLT_TRAFFIC_CLASS;
+	vsi->tc_cfg.numtc = 1;
+}
+
 static inline u8 ice_dcb_get_ena_tc(struct ice_dcbx_cfg __always_unused *dcbcfg)
 {
 	return ICE_DFLT_TRAFFIC_CLASS;
@@ -130,7 +133,6 @@ static inline void ice_vsi_cfg_dcb_rings(struct ice_vsi *vsi) { }
 static inline void ice_update_dcb_stats(struct ice_pf *pf) { }
 static inline void
 ice_dcb_process_lldp_set_mib_change(struct ice_pf *pf, struct ice_rq_event_info *event) { }
-static inline void ice_vsi_cfg_netdev_tc(struct ice_vsi *vsi, u8 ena_tc) { }
 static inline void ice_set_cgd_num(struct ice_tlan_ctx *tlan_ctx, u8 dcb_tc) { }
 #endif /* CONFIG_DCB */
 #endif /* _ICE_DCB_LIB_H_ */

drivers/net/ethernet/intel/ice/ice_eswitch.c

@@ -329,7 +329,7 @@ static void ice_eswitch_release_env(struct ice_pf *pf)
 static struct ice_vsi *
 ice_eswitch_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi)
 {
-	return ice_vsi_setup(pf, pi, ICE_VSI_SWITCHDEV_CTRL, ICE_INVAL_VFID);
+	return ice_vsi_setup(pf, pi, ICE_VSI_SWITCHDEV_CTRL, ICE_INVAL_VFID, NULL);
 }
 
 /**

drivers/net/ethernet/intel/ice/ice_lib.h

@@ -51,13 +51,18 @@ void ice_cfg_sw_lldp(struct ice_vsi *vsi, bool tx, bool create);
 
 int ice_set_link(struct ice_vsi *vsi, bool ena);
 
-#ifdef CONFIG_DCB
+void ice_vsi_delete(struct ice_vsi *vsi);
+int ice_vsi_clear(struct ice_vsi *vsi);
+
 int ice_vsi_cfg_tc(struct ice_vsi *vsi, u8 ena_tc);
-#endif /* CONFIG_DCB */
+
+int ice_vsi_cfg_rss_lut_key(struct ice_vsi *vsi);
+
+void ice_vsi_cfg_netdev_tc(struct ice_vsi *vsi, u8 ena_tc);
 
 struct ice_vsi *
 ice_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi,
-	      enum ice_vsi_type vsi_type, u16 vf_id);
+	      enum ice_vsi_type vsi_type, u16 vf_id, struct ice_channel *ch);
 
 void ice_napi_del(struct ice_vsi *vsi);
 

drivers/net/ethernet/intel/ice/ice_main.c

@@ -3240,7 +3240,7 @@ void ice_fill_rss_lut(u8 *lut, u16 rss_table_size, u16 rss_size)
 static struct ice_vsi *
 ice_pf_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi)
 {
-	return ice_vsi_setup(pf, pi, ICE_VSI_PF, ICE_INVAL_VFID);
+	return ice_vsi_setup(pf, pi, ICE_VSI_PF, ICE_INVAL_VFID, NULL);
 }
 
 /**
@@ -3254,7 +3254,7 @@ ice_pf_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi)
 static struct ice_vsi *
 ice_ctrl_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi)
 {
-	return ice_vsi_setup(pf, pi, ICE_VSI_CTRL, ICE_INVAL_VFID);
+	return ice_vsi_setup(pf, pi, ICE_VSI_CTRL, ICE_INVAL_VFID, NULL);
 }
 
 /**
@@ -3268,7 +3268,7 @@ ice_ctrl_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi)
 struct ice_vsi *
 ice_lb_vsi_setup(struct ice_pf *pf, struct ice_port_info *pi)
 {
-	return ice_vsi_setup(pf, pi, ICE_VSI_LB, ICE_INVAL_VFID);
+	return ice_vsi_setup(pf, pi, ICE_VSI_LB, ICE_INVAL_VFID, NULL);
 }
 
 /**

drivers/net/ethernet/intel/ice/ice_sched.c

@@ -2998,6 +2998,43 @@ static void ice_set_clear_shared_bw(struct ice_bw_type_info *bw_t_info, u32 bw)
 	}
 }
 
+/**
+ * ice_sched_save_vsi_bw - save VSI node's BW information
+ * @pi: port information structure
+ * @vsi_handle: sw VSI handle
+ * @tc: traffic class
+ * @rl_type: rate limit type min, max, or shared
+ * @bw: bandwidth in Kbps - Kilo bits per sec
+ *
+ * Save BW information of VSI type node for post replay use.
+ */
+static int
+ice_sched_save_vsi_bw(struct ice_port_info *pi, u16 vsi_handle, u8 tc,
+		      enum ice_rl_type rl_type, u32 bw)
+{
+	struct ice_vsi_ctx *vsi_ctx;
+
+	if (!ice_is_vsi_valid(pi->hw, vsi_handle))
+		return -EINVAL;
+	vsi_ctx = ice_get_vsi_ctx(pi->hw, vsi_handle);
+	if (!vsi_ctx)
+		return -EINVAL;
+	switch (rl_type) {
+	case ICE_MIN_BW:
+		ice_set_clear_cir_bw(&vsi_ctx->sched.bw_t_info[tc], bw);
+		break;
+	case ICE_MAX_BW:
+		ice_set_clear_eir_bw(&vsi_ctx->sched.bw_t_info[tc], bw);
+		break;
+	case ICE_SHARED_BW:
+		ice_set_clear_shared_bw(&vsi_ctx->sched.bw_t_info[tc], bw);
+		break;
+	default:
+		return -EINVAL;
+	}
+	return 0;
+}
+
 /**
  * ice_sched_calc_wakeup - calculate RL profile wakeup parameter
  * @hw: pointer to the HW struct
@@ -3875,9 +3912,17 @@ enum ice_status
 ice_cfg_vsi_bw_lmt_per_tc(struct ice_port_info *pi, u16 vsi_handle, u8 tc,
 			  enum ice_rl_type rl_type, u32 bw)
 {
-	return ice_sched_set_node_bw_lmt_per_tc(pi, vsi_handle,
-						ICE_AGG_TYPE_VSI,
-						tc, rl_type, bw);
+	int status;
+
+	status = ice_sched_set_node_bw_lmt_per_tc(pi, vsi_handle,
+						  ICE_AGG_TYPE_VSI,
+						  tc, rl_type, bw);
+	if (!status) {
+		mutex_lock(&pi->sched_lock);
+		status = ice_sched_save_vsi_bw(pi, vsi_handle, tc, rl_type, bw);
+		mutex_unlock(&pi->sched_lock);
+	}
+	return status;
 }
 
 /**
@@ -3894,10 +3939,19 @@ enum ice_status
 ice_cfg_vsi_bw_dflt_lmt_per_tc(struct ice_port_info *pi, u16 vsi_handle, u8 tc,
 			       enum ice_rl_type rl_type)
 {
-	return ice_sched_set_node_bw_lmt_per_tc(pi, vsi_handle,
-						ICE_AGG_TYPE_VSI,
-						tc, rl_type,
-						ICE_SCHED_DFLT_BW);
+	int status;
+
+	status = ice_sched_set_node_bw_lmt_per_tc(pi, vsi_handle,
+						  ICE_AGG_TYPE_VSI,
+						  tc, rl_type,
+						  ICE_SCHED_DFLT_BW);
+	if (!status) {
+		mutex_lock(&pi->sched_lock);
+		status = ice_sched_save_vsi_bw(pi, vsi_handle, tc, rl_type,
+					       ICE_SCHED_DFLT_BW);
+		mutex_unlock(&pi->sched_lock);
+	}
+	return status;
 }
 
 /**

drivers/net/ethernet/intel/ice/ice_sched.h

@@ -58,6 +58,8 @@ struct ice_sched_agg_info {
 	DECLARE_BITMAP(tc_bitmap, ICE_MAX_TRAFFIC_CLASS);
 	u32 agg_id;
 	enum ice_agg_type agg_type;
+	/* bw_t_info saves aggregator BW information */
+	struct ice_bw_type_info bw_t_info[ICE_MAX_TRAFFIC_CLASS];
 	/* save aggregator TC bitmap */
 	DECLARE_BITMAP(replay_tc_bitmap, ICE_MAX_TRAFFIC_CLASS);
 };

drivers/net/ethernet/intel/ice/ice_txrx.h

@@ -287,6 +287,7 @@ struct ice_rx_ring {
 
 	struct rcu_head rcu;		/* to avoid race on free */
 	/* CL4 - 3rd cacheline starts here */
+	struct ice_channel *ch;
 	struct bpf_prog *xdp_prog;
 	struct ice_tx_ring *xdp_ring;
 	struct xsk_buff_pool *xsk_pool;
@@ -328,6 +329,7 @@ struct ice_tx_ring {
 	/* CL3 - 3rd cacheline starts here */
 	struct rcu_head rcu;		/* to avoid race on free */
 	DECLARE_BITMAP(xps_state, ICE_TX_NBITS);	/* XPS Config State */
+	struct ice_channel *ch;
 	struct ice_ptp_tx *tx_tstamps;
 	spinlock_t tx_lock;
 	u32 txq_teid;			/* Added Tx queue TEID */
@@ -352,6 +354,11 @@ static inline void ice_clear_ring_build_skb_ena(struct ice_rx_ring *ring)
 	ring->flags &= ~ICE_RX_FLAGS_RING_BUILD_SKB;
 }
 
+static inline bool ice_ring_ch_enabled(struct ice_tx_ring *ring)
+{
+	return !!ring->ch;
+}
+
 static inline bool ice_ring_is_xdp(struct ice_tx_ring *ring)
 {
 	return !!(ring->flags & ICE_TX_FLAGS_RING_XDP);

drivers/net/ethernet/intel/ice/ice_type.h

@@ -138,6 +138,7 @@ enum ice_vsi_type {
 	ICE_VSI_PF = 0,
 	ICE_VSI_VF = 1,
 	ICE_VSI_CTRL = 3,	/* equates to ICE_VSI_PF with 1 queue pair */
+	ICE_VSI_CHNL = 4,
 	ICE_VSI_LB = 6,
 	ICE_VSI_SWITCHDEV_CTRL = 7,
 };
@@ -570,6 +571,8 @@ struct ice_sched_vsi_info {
 	struct list_head list_entry;
 	u16 max_lanq[ICE_MAX_TRAFFIC_CLASS];
 	u16 max_rdmaq[ICE_MAX_TRAFFIC_CLASS];
+	/* bw_t_info saves VSI BW information */
+	struct ice_bw_type_info bw_t_info[ICE_MAX_TRAFFIC_CLASS];
 };
 
 /* driver defines the policy */

drivers/net/ethernet/intel/ice/ice_virtchnl_pf.c

@@ -832,7 +832,7 @@ static struct ice_vsi *ice_vf_vsi_setup(struct ice_vf *vf)
 	struct ice_pf *pf = vf->pf;
 	struct ice_vsi *vsi;
 
-	vsi = ice_vsi_setup(pf, pi, ICE_VSI_VF, vf->vf_id);
+	vsi = ice_vsi_setup(pf, pi, ICE_VSI_VF, vf->vf_id, NULL);
 
 	if (!vsi) {
 		dev_err(ice_pf_to_dev(pf), "Failed to create VF VSI\n");
@@ -859,7 +859,7 @@ struct ice_vsi *ice_vf_ctrl_vsi_setup(struct ice_vf *vf)
 	struct ice_pf *pf = vf->pf;
 	struct ice_vsi *vsi;
 
-	vsi = ice_vsi_setup(pf, pi, ICE_VSI_CTRL, vf->vf_id);
+	vsi = ice_vsi_setup(pf, pi, ICE_VSI_CTRL, vf->vf_id, NULL);
 	if (!vsi) {
 		dev_err(ice_pf_to_dev(pf), "Failed to create VF control VSI\n");
 		ice_vf_ctrl_invalidate_vsi(vf);