Merge branch 'mlx5-esw-qos-refactor-and-shampo-cleanup'

Tariq Toukan says:

====================
mlx5 esw qos refactor and SHAMPO cleanup

This patchset for the mlx5 core and Eth drivers consists of 3 parts.

First patch by Patrisious improves the E-switch mode change operation.

The following 6 patches by Carolina introduce further refactoring for
the QoS handling, to set the foundation for future extensions.

In the following 5 patches by Dragos, we enhance the SHAMPO datapath
flow by simplifying some logic, and cleaning up the implementation.
====================

Link: https://patch.msgid.link/20241107194357.683732-1-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

9 files changed, 303 insertions, 439 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 58f3df784ded..979fc56205e1 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -83,6 +83,7 @@ struct page_pool;
 #define MLX5E_SHAMPO_LOG_HEADER_ENTRY_SIZE (8)
 #define MLX5E_SHAMPO_LOG_MAX_HEADER_ENTRY_SIZE (9)
 #define MLX5E_SHAMPO_WQ_HEADER_PER_PAGE (PAGE_SIZE >> MLX5E_SHAMPO_LOG_MAX_HEADER_ENTRY_SIZE)
+#define MLX5E_SHAMPO_LOG_WQ_HEADER_PER_PAGE (PAGE_SHIFT - MLX5E_SHAMPO_LOG_MAX_HEADER_ENTRY_SIZE)
 #define MLX5E_SHAMPO_WQ_BASE_HEAD_ENTRY_SIZE (64)
 #define MLX5E_SHAMPO_WQ_RESRV_SIZE (64 * 1024)
 #define MLX5E_SHAMPO_WQ_BASE_RESRV_SIZE (4096)
@@ -624,16 +625,14 @@ struct mlx5e_dma_info {
 
 struct mlx5e_shampo_hd {
 	u32 mkey;
-	struct mlx5e_dma_info *info;
 	struct mlx5e_frag_page *pages;
-	u16 curr_page_index;
 	u32 hd_per_wq;
 	u16 hd_per_wqe;
+	u16 pages_per_wq;
 	unsigned long *bitmap;
 	u16 pi;
 	u16 ci;
 	__be32 key;
-	u64 last_addr;
 };
 
 struct mlx5e_hw_gro_data {
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 59d7a0e28f24..2e27e9d6b820 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -350,19 +350,15 @@ static int mlx5e_rq_shampo_hd_info_alloc(struct mlx5e_rq *rq, int node)
 
 	shampo->bitmap = bitmap_zalloc_node(shampo->hd_per_wq, GFP_KERNEL,
 					    node);
-	shampo->info = kvzalloc_node(array_size(shampo->hd_per_wq,
-						sizeof(*shampo->info)),
-				     GFP_KERNEL, node);
 	shampo->pages = kvzalloc_node(array_size(shampo->hd_per_wq,
 						 sizeof(*shampo->pages)),
 				     GFP_KERNEL, node);
-	if (!shampo->bitmap || !shampo->info || !shampo->pages)
+	if (!shampo->bitmap || !shampo->pages)
 		goto err_nomem;
 
 	return 0;
 
 err_nomem:
-	kvfree(shampo->info);
 	kvfree(shampo->bitmap);
 	kvfree(shampo->pages);
 
@@ -372,7 +368,6 @@ err_nomem:
 static void mlx5e_rq_shampo_hd_info_free(struct mlx5e_rq *rq)
 {
 	kvfree(rq->mpwqe.shampo->bitmap);
-	kvfree(rq->mpwqe.shampo->info);
 	kvfree(rq->mpwqe.shampo->pages);
 }
 
@@ -767,8 +762,6 @@ static int mlx5_rq_shampo_alloc(struct mlx5_core_dev *mdev,
 				u32 *pool_size,
 				int node)
 {
-	void *wqc = MLX5_ADDR_OF(rqc, rqp->rqc, wq);
-	int wq_size;
 	int err;
 
 	if (!test_bit(MLX5E_RQ_STATE_SHAMPO, &rq->state))
@@ -793,9 +786,9 @@ static int mlx5_rq_shampo_alloc(struct mlx5_core_dev *mdev,
 		cpu_to_be32(rq->mpwqe.shampo->mkey);
 	rq->mpwqe.shampo->hd_per_wqe =
 		mlx5e_shampo_hd_per_wqe(mdev, params, rqp);
-	wq_size = BIT(MLX5_GET(wq, wqc, log_wq_sz));
-	*pool_size += (rq->mpwqe.shampo->hd_per_wqe * wq_size) /
-		     MLX5E_SHAMPO_WQ_HEADER_PER_PAGE;
+	rq->mpwqe.shampo->pages_per_wq =
+		rq->mpwqe.shampo->hd_per_wq / MLX5E_SHAMPO_WQ_HEADER_PER_PAGE;
+	*pool_size += rq->mpwqe.shampo->pages_per_wq;
 	return 0;
 
 err_hw_gro_data:
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
index d81083f4f316..1963bc5adb18 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_rx.c
@@ -643,83 +643,82 @@ static void build_ksm_umr(struct mlx5e_icosq *sq, struct mlx5e_umr_wqe *umr_wqe,
 	umr_wqe->uctrl.mkey_mask     = cpu_to_be64(MLX5_MKEY_MASK_FREE);
 }
 
+static struct mlx5e_frag_page *mlx5e_shampo_hd_to_frag_page(struct mlx5e_rq *rq, int header_index)
+{
+	BUILD_BUG_ON(MLX5E_SHAMPO_LOG_MAX_HEADER_ENTRY_SIZE > PAGE_SHIFT);
+
+	return &rq->mpwqe.shampo->pages[header_index >> MLX5E_SHAMPO_LOG_WQ_HEADER_PER_PAGE];
+}
+
+static u64 mlx5e_shampo_hd_offset(int header_index)
+{
+	return (header_index & (MLX5E_SHAMPO_WQ_HEADER_PER_PAGE - 1)) <<
+		MLX5E_SHAMPO_LOG_MAX_HEADER_ENTRY_SIZE;
+}
+
+static void mlx5e_free_rx_shampo_hd_entry(struct mlx5e_rq *rq, u16 header_index);
+
 static int mlx5e_build_shampo_hd_umr(struct mlx5e_rq *rq,
 				     struct mlx5e_icosq *sq,
 				     u16 ksm_entries, u16 index)
 {
 	struct mlx5e_shampo_hd *shampo = rq->mpwqe.shampo;
-	u16 entries, pi, header_offset, err, wqe_bbs, new_entries;
+	u16 pi, header_offset, err, wqe_bbs;
 	u32 lkey = rq->mdev->mlx5e_res.hw_objs.mkey;
-	u16 page_index = shampo->curr_page_index;
-	struct mlx5e_frag_page *frag_page;
-	u64 addr = shampo->last_addr;
-	struct mlx5e_dma_info *dma_info;
 	struct mlx5e_umr_wqe *umr_wqe;
-	int headroom, i;
+	int headroom, i = 0;
 
 	headroom = rq->buff.headroom;
-	new_entries = ksm_entries - (shampo->pi & (MLX5_UMR_KSM_NUM_ENTRIES_ALIGNMENT - 1));
-	entries = ALIGN(ksm_entries, MLX5_UMR_KSM_NUM_ENTRIES_ALIGNMENT);
-	wqe_bbs = MLX5E_KSM_UMR_WQEBBS(entries);
+	wqe_bbs = MLX5E_KSM_UMR_WQEBBS(ksm_entries);
 	pi = mlx5e_icosq_get_next_pi(sq, wqe_bbs);
 	umr_wqe = mlx5_wq_cyc_get_wqe(&sq->wq, pi);
-	build_ksm_umr(sq, umr_wqe, shampo->key, index, entries);
+	build_ksm_umr(sq, umr_wqe, shampo->key, index, ksm_entries);
 
-	frag_page = &shampo->pages[page_index];
+	WARN_ON_ONCE(ksm_entries & (MLX5E_SHAMPO_WQ_HEADER_PER_PAGE - 1));
+	while (i < ksm_entries) {
+		struct mlx5e_frag_page *frag_page = mlx5e_shampo_hd_to_frag_page(rq, index);
+		u64 addr;
 
-	for (i = 0; i < entries; i++, index++) {
-		dma_info = &shampo->info[index];
-		if (i >= ksm_entries || (index < shampo->pi && shampo->pi - index <
-					 MLX5_UMR_KSM_NUM_ENTRIES_ALIGNMENT))
-			goto update_ksm;
-		header_offset = (index & (MLX5E_SHAMPO_WQ_HEADER_PER_PAGE - 1)) <<
-			MLX5E_SHAMPO_LOG_MAX_HEADER_ENTRY_SIZE;
-		if (!(header_offset & (PAGE_SIZE - 1))) {
-			page_index = (page_index + 1) & (shampo->hd_per_wq - 1);
-			frag_page = &shampo->pages[page_index];
+		err = mlx5e_page_alloc_fragmented(rq, frag_page);
+		if (unlikely(err))
+			goto err_unmap;
 
-			err = mlx5e_page_alloc_fragmented(rq, frag_page);
-			if (unlikely(err))
-				goto err_unmap;
 
-			addr = page_pool_get_dma_addr(frag_page->page);
+		addr = page_pool_get_dma_addr(frag_page->page);
 
-			dma_info->addr = addr;
-			dma_info->frag_page = frag_page;
-		} else {
-			dma_info->addr = addr + header_offset;
-			dma_info->frag_page = frag_page;
-		}
+		for (int j = 0; j < MLX5E_SHAMPO_WQ_HEADER_PER_PAGE; j++) {
+			header_offset = mlx5e_shampo_hd_offset(index++);
 
-update_ksm:
-		umr_wqe->inline_ksms[i] = (struct mlx5_ksm) {
-			.key = cpu_to_be32(lkey),
-			.va  = cpu_to_be64(dma_info->addr + headroom),
-		};
+			umr_wqe->inline_ksms[i++] = (struct mlx5_ksm) {
+				.key = cpu_to_be32(lkey),
+				.va  = cpu_to_be64(addr + header_offset + headroom),
+			};
+		}
 	}
 
 	sq->db.wqe_info[pi] = (struct mlx5e_icosq_wqe_info) {
 		.wqe_type	= MLX5E_ICOSQ_WQE_SHAMPO_HD_UMR,
 		.num_wqebbs	= wqe_bbs,
-		.shampo.len	= new_entries,
+		.shampo.len	= ksm_entries,
 	};
 
-	shampo->pi = (shampo->pi + new_entries) & (shampo->hd_per_wq - 1);
-	shampo->curr_page_index = page_index;
-	shampo->last_addr = addr;
+	shampo->pi = (shampo->pi + ksm_entries) & (shampo->hd_per_wq - 1);
 	sq->pc += wqe_bbs;
 	sq->doorbell_cseg = &umr_wqe->ctrl;
 
 	return 0;
 
 err_unmap:
-	while (--i >= 0) {
-		dma_info = &shampo->info[--index];
-		if (!(i & (MLX5E_SHAMPO_WQ_HEADER_PER_PAGE - 1))) {
-			dma_info->addr = ALIGN_DOWN(dma_info->addr, PAGE_SIZE);
-			mlx5e_page_release_fragmented(rq, dma_info->frag_page);
+	while (--i) {
+		--index;
+		header_offset = mlx5e_shampo_hd_offset(index);
+		if (!header_offset) {
+			struct mlx5e_frag_page *frag_page = mlx5e_shampo_hd_to_frag_page(rq, index);
+
+			mlx5e_page_release_fragmented(rq, frag_page);
 		}
 	}
+
 	rq->stats->buff_alloc_err++;
 	return err;
 }
@@ -731,7 +730,8 @@ static int mlx5e_alloc_rx_hd_mpwqe(struct mlx5e_rq *rq)
 	struct mlx5e_icosq *sq = rq->icosq;
 	int i, err, max_ksm_entries, len;
 
-	max_ksm_entries = MLX5E_MAX_KSM_PER_WQE(rq->mdev);
+	max_ksm_entries = ALIGN_DOWN(MLX5E_MAX_KSM_PER_WQE(rq->mdev),
+				     MLX5E_SHAMPO_WQ_HEADER_PER_PAGE);
 	ksm_entries = bitmap_find_window(shampo->bitmap,
 					 shampo->hd_per_wqe,
 					 shampo->hd_per_wq, shampo->pi);
@@ -739,8 +739,8 @@ static int mlx5e_alloc_rx_hd_mpwqe(struct mlx5e_rq *rq)
 	if (!ksm_entries)
 		return 0;
 
-	ksm_entries += (shampo->pi & (MLX5_UMR_KSM_NUM_ENTRIES_ALIGNMENT - 1));
-	index = ALIGN_DOWN(shampo->pi, MLX5_UMR_KSM_NUM_ENTRIES_ALIGNMENT);
+	/* pi is aligned to MLX5E_SHAMPO_WQ_HEADER_PER_PAGE */
+	index = shampo->pi;
 	entries_before = shampo->hd_per_wq - index;
 
 	if (unlikely(entries_before < ksm_entries))
@@ -851,13 +851,11 @@ static void
 mlx5e_free_rx_shampo_hd_entry(struct mlx5e_rq *rq, u16 header_index)
 {
 	struct mlx5e_shampo_hd *shampo = rq->mpwqe.shampo;
-	u64 addr = shampo->info[header_index].addr;
 
 	if (((header_index + 1) & (MLX5E_SHAMPO_WQ_HEADER_PER_PAGE - 1)) == 0) {
-		struct mlx5e_dma_info *dma_info = &shampo->info[header_index];
+		struct mlx5e_frag_page *frag_page = mlx5e_shampo_hd_to_frag_page(rq, header_index);
 
-		dma_info->addr = ALIGN_DOWN(addr, PAGE_SIZE);
-		mlx5e_page_release_fragmented(rq, dma_info->frag_page);
+		mlx5e_page_release_fragmented(rq, frag_page);
 	}
 	clear_bit(header_index, shampo->bitmap);
 }
@@ -1211,10 +1209,10 @@ static void mlx5e_lro_update_hdr(struct sk_buff *skb, struct mlx5_cqe64 *cqe,
 
 static void *mlx5e_shampo_get_packet_hd(struct mlx5e_rq *rq, u16 header_index)
 {
-	struct mlx5e_dma_info *last_head = &rq->mpwqe.shampo->info[header_index];
-	u16 head_offset = (last_head->addr & (PAGE_SIZE - 1)) + rq->buff.headroom;
+	struct mlx5e_frag_page *frag_page = mlx5e_shampo_hd_to_frag_page(rq, header_index);
+	u16 head_offset = mlx5e_shampo_hd_offset(header_index) + rq->buff.headroom;
 
-	return page_address(last_head->frag_page->page) + head_offset;
+	return page_address(frag_page->page) + head_offset;
 }
 
 static void mlx5e_shampo_update_ipv4_udp_hdr(struct mlx5e_rq *rq, struct iphdr *ipv4)
@@ -2185,29 +2183,30 @@ static struct sk_buff *
 mlx5e_skb_from_cqe_shampo(struct mlx5e_rq *rq, struct mlx5e_mpw_info *wi,
 			  struct mlx5_cqe64 *cqe, u16 header_index)
 {
-	struct mlx5e_dma_info *head = &rq->mpwqe.shampo->info[header_index];
-	u16 head_offset = head->addr & (PAGE_SIZE - 1);
+	struct mlx5e_frag_page *frag_page = mlx5e_shampo_hd_to_frag_page(rq, header_index);
+	dma_addr_t page_dma_addr = page_pool_get_dma_addr(frag_page->page);
+	u16 head_offset = mlx5e_shampo_hd_offset(header_index);
+	dma_addr_t dma_addr = page_dma_addr + head_offset;
 	u16 head_size = cqe->shampo.header_size;
 	u16 rx_headroom = rq->buff.headroom;
 	struct sk_buff *skb = NULL;
 	void *hdr, *data;
 	u32 frag_size;
 
-	hdr		= page_address(head->frag_page->page) + head_offset;
+	hdr		= page_address(frag_page->page) + head_offset;
 	data		= hdr + rx_headroom;
 	frag_size	= MLX5_SKB_FRAG_SZ(rx_headroom + head_size);
 
 	if (likely(frag_size <= BIT(MLX5E_SHAMPO_LOG_MAX_HEADER_ENTRY_SIZE))) {
 		/* build SKB around header */
-		dma_sync_single_range_for_cpu(rq->pdev, head->addr, 0, frag_size, rq->buff.map_dir);
+		dma_sync_single_range_for_cpu(rq->pdev, dma_addr, 0, frag_size, rq->buff.map_dir);
 		net_prefetchw(hdr);
 		net_prefetch(data);
 		skb = mlx5e_build_linear_skb(rq, hdr, frag_size, rx_headroom, head_size, 0);
-
 		if (unlikely(!skb))
 			return NULL;
 
-		head->frag_page->frags++;
+		frag_page->frags++;
 	} else {
 		/* allocate SKB and copy header for large header */
 		rq->stats->gro_large_hds++;
@@ -2219,7 +2218,7 @@ mlx5e_skb_from_cqe_shampo(struct mlx5e_rq *rq, struct mlx5e_mpw_info *wi,
 		}
 
 		net_prefetchw(skb->data);
-		mlx5e_copy_skb_header(rq, skb, head->frag_page->page, head->addr,
+		mlx5e_copy_skb_header(rq, skb, frag_page->page, dma_addr,
 				      head_offset + rx_headroom,
 				      rx_headroom, head_size);
 		/* skb linear part was allocated with headlen and aligned to long */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c
index d0f38818363f..982fe3714683 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/devlink_port.c
@@ -195,7 +195,7 @@ void mlx5_esw_offloads_devlink_port_unregister(struct mlx5_vport *vport)
 		return;
 	dl_port = vport->dl_port;
 
-	mlx5_esw_qos_vport_update_node(vport, NULL, NULL);
+	mlx5_esw_qos_vport_update_parent(vport, NULL, NULL);
 	devl_rate_leaf_destroy(&dl_port->dl_port);
 
 	devl_port_unregister(&dl_port->dl_port);
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c
index 940e1c2d1e39..8b7c843446e1 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.c
@@ -101,6 +101,12 @@ esw_qos_node_set_parent(struct mlx5_esw_sched_node *node, struct mlx5_esw_sched_
 	node->esw = parent->esw;
 }
 
+void mlx5_esw_qos_vport_qos_free(struct mlx5_vport *vport)
+{
+	kfree(vport->qos.sched_node);
+	memset(&vport->qos, 0, sizeof(vport->qos));
+}
+
 u32 mlx5_esw_qos_vport_get_sched_elem_ix(const struct mlx5_vport *vport)
 {
 	if (!vport->qos.sched_node)
@@ -118,18 +124,49 @@ mlx5_esw_qos_vport_get_parent(const struct mlx5_vport *vport)
 	return vport->qos.sched_node->parent;
 }
 
-static void esw_qos_sched_elem_config_warn(struct mlx5_esw_sched_node *node, int err)
+static void esw_qos_sched_elem_warn(struct mlx5_esw_sched_node *node, int err, const char *op)
 {
 	if (node->vport) {
 		esw_warn(node->esw->dev,
-			 "E-Switch modify %s scheduling element failed (vport=%d,err=%d)\n",
-			 sched_node_type_str[node->type], node->vport->vport, err);
+			 "E-Switch %s %s scheduling element failed (vport=%d,err=%d)\n",
+			 op, sched_node_type_str[node->type], node->vport->vport, err);
 		return;
 	}
 
 	esw_warn(node->esw->dev,
-		 "E-Switch modify %s scheduling element failed (err=%d)\n",
-		 sched_node_type_str[node->type], err);
+		 "E-Switch %s %s scheduling element failed (err=%d)\n",
+		 op, sched_node_type_str[node->type], err);
+}
+
+static int esw_qos_node_create_sched_element(struct mlx5_esw_sched_node *node, void *ctx,
+					     struct netlink_ext_ack *extack)
+{
+	int err;
+
+	err = mlx5_create_scheduling_element_cmd(node->esw->dev, SCHEDULING_HIERARCHY_E_SWITCH, ctx,
+						 &node->ix);
+	if (err) {
+		esw_qos_sched_elem_warn(node, err, "create");
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch create scheduling element failed");
+	}
+
+	return err;
+}
+
+static int esw_qos_node_destroy_sched_element(struct mlx5_esw_sched_node *node,
+					      struct netlink_ext_ack *extack)
+{
+	int err;
+
+	err = mlx5_destroy_scheduling_element_cmd(node->esw->dev,
+						  SCHEDULING_HIERARCHY_E_SWITCH,
+						  node->ix);
+	if (err) {
+		esw_qos_sched_elem_warn(node, err, "destroy");
+		NL_SET_ERR_MSG_MOD(extack, "E-Switch destroying scheduling element failed.");
+	}
+
+	return err;
 }
 
 static int esw_qos_sched_elem_config(struct mlx5_esw_sched_node *node, u32 max_rate, u32 bw_share,
@@ -143,10 +180,21 @@ static int esw_qos_sched_elem_config(struct mlx5_esw_sched_node *node, u32 max_r
 	if (!MLX5_CAP_GEN(dev, qos) || !MLX5_CAP_QOS(dev, esw_scheduling))
 		return -EOPNOTSUPP;
 
-	MLX5_SET(scheduling_context, sched_ctx, max_average_bw, max_rate);
-	MLX5_SET(scheduling_context, sched_ctx, bw_share, bw_share);
-	bitmask |= MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_MAX_AVERAGE_BW;
-	bitmask |= MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_BW_SHARE;
+	if (bw_share && (!MLX5_CAP_QOS(dev, esw_bw_share) ||
+			 MLX5_CAP_QOS(dev, max_tsar_bw_share) < MLX5_MIN_BW_SHARE))
+		return -EOPNOTSUPP;
+
+	if (node->max_rate == max_rate && node->bw_share == bw_share)
+		return 0;
+
+	if (node->max_rate != max_rate) {
+		MLX5_SET(scheduling_context, sched_ctx, max_average_bw, max_rate);
+		bitmask |= MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_MAX_AVERAGE_BW;
+	}
+	if (node->bw_share != bw_share) {
+		MLX5_SET(scheduling_context, sched_ctx, bw_share, bw_share);
+		bitmask |= MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_BW_SHARE;
+	}
 
 	err = mlx5_modify_scheduling_element_cmd(dev,
 						 SCHEDULING_HIERARCHY_E_SWITCH,
@@ -154,12 +202,14 @@ static int esw_qos_sched_elem_config(struct mlx5_esw_sched_node *node, u32 max_r
 						 node->ix,
 						 bitmask);
 	if (err) {
-		esw_qos_sched_elem_config_warn(node, err);
+		esw_qos_sched_elem_warn(node, err, "modify");
 		NL_SET_ERR_MSG_MOD(extack, "E-Switch modify scheduling element failed");
 
 		return err;
 	}
 
+	node->max_rate = max_rate;
+	node->bw_share = bw_share;
 	if (node->type == SCHED_NODE_TYPE_VPORTS_TSAR)
 		trace_mlx5_esw_node_qos_config(dev, node, node->ix, bw_share, max_rate);
 	else if (node->type == SCHED_NODE_TYPE_VPORT)
@@ -208,166 +258,51 @@ static u32 esw_qos_calc_bw_share(u32 min_rate, u32 divider, u32 fw_max)
 	return min_t(u32, max_t(u32, DIV_ROUND_UP(min_rate, divider), MLX5_MIN_BW_SHARE), fw_max);
 }
 
-static int esw_qos_update_sched_node_bw_share(struct mlx5_esw_sched_node *node,
-					      u32 divider,
-					      struct netlink_ext_ack *extack)
+static void esw_qos_update_sched_node_bw_share(struct mlx5_esw_sched_node *node,
+					       u32 divider,
+					       struct netlink_ext_ack *extack)
 {
 	u32 fw_max_bw_share = MLX5_CAP_QOS(node->esw->dev, max_tsar_bw_share);
 	u32 bw_share;
-	int err;
 
 	bw_share = esw_qos_calc_bw_share(node->min_rate, divider, fw_max_bw_share);
 
-	if (bw_share == node->bw_share)
-		return 0;
-
-	err = esw_qos_sched_elem_config(node, node->max_rate, bw_share, extack);
-	if (err)
-		return err;
-
-	node->bw_share = bw_share;
-
-	return err;
+	esw_qos_sched_elem_config(node, node->max_rate, bw_share, extack);
 }
 
-static int esw_qos_normalize_min_rate(struct mlx5_eswitch *esw,
-				      struct mlx5_esw_sched_node *parent,
-				      struct netlink_ext_ack *extack)
+static void esw_qos_normalize_min_rate(struct mlx5_eswitch *esw,
+				       struct mlx5_esw_sched_node *parent,
+				       struct netlink_ext_ack *extack)
 {
 	struct list_head *nodes = parent ? &parent->children : &esw->qos.domain->nodes;
 	u32 divider = esw_qos_calculate_min_rate_divider(esw, parent);
 	struct mlx5_esw_sched_node *node;
 
 	list_for_each_entry(node, nodes, entry) {
-		int err;
-
 		if (node->esw != esw || node->ix == esw->qos.root_tsar_ix)
 			continue;
 
-		err = esw_qos_update_sched_node_bw_share(node, divider, extack);
-		if (err)
-			return err;
+		esw_qos_update_sched_node_bw_share(node, divider, extack);
 
 		if (list_empty(&node->children))
 			continue;
 
-		err = esw_qos_normalize_min_rate(node->esw, node, extack);
-		if (err)
-			return err;
+		esw_qos_normalize_min_rate(node->esw, node, extack);
 	}
-
-	return 0;
-}
-
-static int esw_qos_set_vport_min_rate(struct mlx5_vport *vport,
-				      u32 min_rate, struct netlink_ext_ack *extack)
-{
-	struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node;
-	u32 fw_max_bw_share, previous_min_rate;
-	bool min_rate_supported;
-	int err;
-
-	esw_assert_qos_lock_held(vport_node->esw);
-	fw_max_bw_share = MLX5_CAP_QOS(vport->dev, max_tsar_bw_share);
-	min_rate_supported = MLX5_CAP_QOS(vport->dev, esw_bw_share) &&
-				fw_max_bw_share >= MLX5_MIN_BW_SHARE;
-	if (min_rate && !min_rate_supported)
-		return -EOPNOTSUPP;
-	if (min_rate == vport_node->min_rate)
-		return 0;
-
-	previous_min_rate = vport_node->min_rate;
-	vport_node->min_rate = min_rate;
-	err = esw_qos_normalize_min_rate(vport_node->parent->esw, vport_node->parent, extack);
-	if (err)
-		vport_node->min_rate = previous_min_rate;
-
-	return err;
-}
-
-static int esw_qos_set_vport_max_rate(struct mlx5_vport *vport,
-				      u32 max_rate, struct netlink_ext_ack *extack)
-{
-	struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node;
-	u32 act_max_rate = max_rate;
-	bool max_rate_supported;
-	int err;
-
-	esw_assert_qos_lock_held(vport_node->esw);
-	max_rate_supported = MLX5_CAP_QOS(vport->dev, esw_rate_limit);
-
-	if (max_rate && !max_rate_supported)
-		return -EOPNOTSUPP;
-	if (max_rate == vport_node->max_rate)
-		return 0;
-
-	/* Use parent node limit if new max rate is 0. */
-	if (!max_rate)
-		act_max_rate = vport_node->parent->max_rate;
-
-	err = esw_qos_sched_elem_config(vport_node, act_max_rate, vport_node->bw_share, extack);
-	if (!err)
-		vport_node->max_rate = max_rate;
-
-	return err;
 }
 
 static int esw_qos_set_node_min_rate(struct mlx5_esw_sched_node *node,
 				     u32 min_rate, struct netlink_ext_ack *extack)
 {
 	struct mlx5_eswitch *esw = node->esw;
-	u32 previous_min_rate;
-	int err;
-
-	if (!MLX5_CAP_QOS(esw->dev, esw_bw_share) ||
-	    MLX5_CAP_QOS(esw->dev, max_tsar_bw_share) < MLX5_MIN_BW_SHARE)
-		return -EOPNOTSUPP;
 
 	if (min_rate == node->min_rate)
 		return 0;
 
-	previous_min_rate = node->min_rate;
 	node->min_rate = min_rate;
-	err = esw_qos_normalize_min_rate(esw, NULL, extack);
-	if (err) {
-		NL_SET_ERR_MSG_MOD(extack, "E-Switch node min rate setting failed");
-
-		/* Attempt restoring previous configuration */
-		node->min_rate = previous_min_rate;
-		if (esw_qos_normalize_min_rate(esw, NULL, extack))
-			NL_SET_ERR_MSG_MOD(extack, "E-Switch BW share restore failed");
-	}
-
-	return err;
-}
-
-static int esw_qos_set_node_max_rate(struct mlx5_esw_sched_node *node,
-				     u32 max_rate, struct netlink_ext_ack *extack)
-{
-	struct mlx5_esw_sched_node *vport_node;
-	int err;
-
-	if (node->max_rate == max_rate)
-		return 0;
-
-	err = esw_qos_sched_elem_config(node, max_rate, node->bw_share, extack);
-	if (err)
-		return err;
-
-	node->max_rate = max_rate;
-
-	/* Any unlimited vports in the node should be set with the value of the node. */
-	list_for_each_entry(vport_node, &node->children, entry) {
-		if (vport_node->max_rate)
-			continue;
+	esw_qos_normalize_min_rate(esw, node->parent, extack);
 
-		err = esw_qos_sched_elem_config(vport_node, max_rate, vport_node->bw_share, extack);
-		if (err)
-			NL_SET_ERR_MSG_MOD(extack,
-					   "E-Switch vport implicit rate limit setting failed");
-	}
-
-	return err;
+	return 0;
 }
 
 static int esw_qos_create_node_sched_elem(struct mlx5_core_dev *dev, u32 parent_element_id,
@@ -397,14 +332,12 @@ static int esw_qos_create_node_sched_elem(struct mlx5_core_dev *dev, u32 parent_
 						  tsar_ix);
 }
 
-static int
-esw_qos_vport_create_sched_element(struct mlx5_vport *vport, struct mlx5_esw_sched_node *parent,
-				   u32 max_rate, u32 bw_share, u32 *sched_elem_ix)
+static int esw_qos_vport_create_sched_element(struct mlx5_esw_sched_node *vport_node,
+					      struct netlink_ext_ack *extack)
 {
 	u32 sched_ctx[MLX5_ST_SZ_DW(scheduling_context)] = {};
-	struct mlx5_core_dev *dev = parent->esw->dev;
+	struct mlx5_core_dev *dev = vport_node->esw->dev;
 	void *attr;
-	int err;
 
 	if (!mlx5_qos_element_type_supported(dev,
 					     SCHEDULING_CONTEXT_ELEMENT_TYPE_VPORT,
@@ -414,93 +347,11 @@ esw_qos_vport_create_sched_element(struct mlx5_vport *vport, struct mlx5_esw_sch
 	MLX5_SET(scheduling_context, sched_ctx, element_type,
 		 SCHEDULING_CONTEXT_ELEMENT_TYPE_VPORT);
 	attr = MLX5_ADDR_OF(scheduling_context, sched_ctx, element_attributes);
-	MLX5_SET(vport_element, attr, vport_number, vport->vport);
-	MLX5_SET(scheduling_context, sched_ctx, parent_element_id, parent->ix);
-	MLX5_SET(scheduling_context, sched_ctx, max_average_bw, max_rate);
-	MLX5_SET(scheduling_context, sched_ctx, bw_share, bw_share);
-
-	err = mlx5_create_scheduling_element_cmd(dev,
-						 SCHEDULING_HIERARCHY_E_SWITCH,
-						 sched_ctx,
-						 sched_elem_ix);
-	if (err) {
-		esw_warn(dev,
-			 "E-Switch create vport scheduling element failed (vport=%d,err=%d)\n",
-			 vport->vport, err);
-		return err;
-	}
-
-	return 0;
-}
-
-static int esw_qos_update_node_scheduling_element(struct mlx5_vport *vport,
-						  struct mlx5_esw_sched_node *curr_node,
-						  struct mlx5_esw_sched_node *new_node,
-						  struct netlink_ext_ack *extack)
-{
-	struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node;
-	u32 max_rate;
-	int err;
-
-	err = mlx5_destroy_scheduling_element_cmd(curr_node->esw->dev,
-						  SCHEDULING_HIERARCHY_E_SWITCH,
-						  vport_node->ix);
-	if (err) {
-		NL_SET_ERR_MSG_MOD(extack, "E-Switch destroy vport scheduling element failed");
-		return err;
-	}
-
-	/* Use new node max rate if vport max rate is unlimited. */
-	max_rate = vport_node->max_rate ? vport_node->max_rate : new_node->max_rate;
-	err = esw_qos_vport_create_sched_element(vport, new_node, max_rate,
-						 vport_node->bw_share,
-						 &vport_node->ix);
-	if (err) {
-		NL_SET_ERR_MSG_MOD(extack, "E-Switch vport node set failed.");
-		goto err_sched;
-	}
-
-	esw_qos_node_set_parent(vport->qos.sched_node, new_node);
-
-	return 0;
+	MLX5_SET(vport_element, attr, vport_number, vport_node->vport->vport);
+	MLX5_SET(scheduling_context, sched_ctx, parent_element_id, vport_node->parent->ix);
+	MLX5_SET(scheduling_context, sched_ctx, max_average_bw, vport_node->max_rate);
 
-err_sched:
-	max_rate = vport_node->max_rate ? vport_node->max_rate : curr_node->max_rate;
-	if (esw_qos_vport_create_sched_element(vport, curr_node, max_rate,
-					       vport_node->bw_share,
-					       &vport_node->ix))
-		esw_warn(curr_node->esw->dev, "E-Switch vport node restore failed (vport=%d)\n",
-			 vport->vport);
-
-	return err;
-}
-
-static int esw_qos_vport_update_node(struct mlx5_vport *vport,
-				     struct mlx5_esw_sched_node *node,
-				     struct netlink_ext_ack *extack)
-{
-	struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node;
-	struct mlx5_eswitch *esw = vport->dev->priv.eswitch;
-	struct mlx5_esw_sched_node *new_node, *curr_node;
-	int err;
-
-	esw_assert_qos_lock_held(esw);
-	curr_node = vport_node->parent;
-	new_node = node ?: esw->qos.node0;
-	if (curr_node == new_node)
-		return 0;
-
-	err = esw_qos_update_node_scheduling_element(vport, curr_node, new_node, extack);
-	if (err)
-		return err;
-
-	/* Recalculate bw share weights of old and new nodes */
-	if (vport_node->bw_share || new_node->bw_share) {
-		esw_qos_normalize_min_rate(curr_node->esw, curr_node, extack);
-		esw_qos_normalize_min_rate(new_node->esw, new_node, extack);
-	}
-
-	return 0;
+	return esw_qos_node_create_sched_element(vport_node, sched_ctx, extack);
 }
 
 static struct mlx5_esw_sched_node *
@@ -531,6 +382,12 @@ static void __esw_qos_free_node(struct mlx5_esw_sched_node *node)
 	kfree(node);
 }
 
+static void esw_qos_destroy_node(struct mlx5_esw_sched_node *node, struct netlink_ext_ack *extack)
+{
+	esw_qos_node_destroy_sched_element(node, extack);
+	__esw_qos_free_node(node);
+}
+
 static struct mlx5_esw_sched_node *
 __esw_qos_create_vports_sched_node(struct mlx5_eswitch *esw, struct mlx5_esw_sched_node *parent,
 				   struct netlink_ext_ack *extack)
@@ -552,17 +409,11 @@ __esw_qos_create_vports_sched_node(struct mlx5_eswitch *esw, struct mlx5_esw_sch
 		goto err_alloc_node;
 	}
 
-	err = esw_qos_normalize_min_rate(esw, NULL, extack);
-	if (err) {
-		NL_SET_ERR_MSG_MOD(extack, "E-Switch nodes normalization failed");
-		goto err_min_rate;
-	}
+	esw_qos_normalize_min_rate(esw, NULL, extack);
 	trace_mlx5_esw_node_qos_create(esw->dev, node, node->ix);
 
 	return node;
 
-err_min_rate:
-	__esw_qos_free_node(node);
 err_alloc_node:
 	if (mlx5_destroy_scheduling_element_cmd(esw->dev,
 						SCHEDULING_HIERARCHY_E_SWITCH,
@@ -595,26 +446,13 @@ esw_qos_create_vports_sched_node(struct mlx5_eswitch *esw, struct netlink_ext_ac
 	return node;
 }
 
-static int __esw_qos_destroy_node(struct mlx5_esw_sched_node *node, struct netlink_ext_ack *extack)
+static void __esw_qos_destroy_node(struct mlx5_esw_sched_node *node, struct netlink_ext_ack *extack)
 {
 	struct mlx5_eswitch *esw = node->esw;
-	int err;
 
 	trace_mlx5_esw_node_qos_destroy(esw->dev, node, node->ix);
-
-	err = mlx5_destroy_scheduling_element_cmd(esw->dev,
-						  SCHEDULING_HIERARCHY_E_SWITCH,
-						  node->ix);
-	if (err)
-		NL_SET_ERR_MSG_MOD(extack, "E-Switch destroy TSAR_ID failed");
-	__esw_qos_free_node(node);
-
-	err = esw_qos_normalize_min_rate(esw, NULL, extack);
-	if (err)
-		NL_SET_ERR_MSG_MOD(extack, "E-Switch nodes normalization failed");
-
-
-	return err;
+	esw_qos_destroy_node(node, extack);
+	esw_qos_normalize_min_rate(esw, NULL, extack);
 }
 
 static int esw_qos_create(struct mlx5_eswitch *esw, struct netlink_ext_ack *extack)
@@ -699,45 +537,62 @@ static void esw_qos_put(struct mlx5_eswitch *esw)
 		esw_qos_destroy(esw);
 }
 
-static int esw_qos_vport_enable(struct mlx5_vport *vport,
-				u32 max_rate, u32 bw_share, struct netlink_ext_ack *extack)
+static void esw_qos_vport_disable(struct mlx5_vport *vport, struct netlink_ext_ack *extack)
+{
+	struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node;
+	struct mlx5_esw_sched_node *parent = vport_node->parent;
+
+	esw_qos_node_destroy_sched_element(vport_node, extack);
+
+	vport_node->bw_share = 0;
+	list_del_init(&vport_node->entry);
+	esw_qos_normalize_min_rate(parent->esw, parent, extack);
+
+	trace_mlx5_esw_vport_qos_destroy(vport_node->esw->dev, vport);
+}
+
+static int esw_qos_vport_enable(struct mlx5_vport *vport, struct mlx5_esw_sched_node *parent,
+				struct netlink_ext_ack *extack)
 {
-	struct mlx5_eswitch *esw = vport->dev->priv.eswitch;
-	u32 sched_elem_ix;
 	int err;
 
-	esw_assert_qos_lock_held(esw);
-	if (vport->qos.sched_node)
-		return 0;
+	esw_assert_qos_lock_held(vport->dev->priv.eswitch);
 
-	err = esw_qos_get(esw, extack);
+	esw_qos_node_set_parent(vport->qos.sched_node, parent);
+	err = esw_qos_vport_create_sched_element(vport->qos.sched_node, extack);
 	if (err)
 		return err;
 
-	err = esw_qos_vport_create_sched_element(vport, esw->qos.node0, max_rate, bw_share,
-						 &sched_elem_ix);
-	if (err)
-		goto err_out;
+	esw_qos_normalize_min_rate(parent->esw, parent, extack);
 
-	vport->qos.sched_node = __esw_qos_alloc_node(esw, sched_elem_ix, SCHED_NODE_TYPE_VPORT,
-						     esw->qos.node0);
-	if (!vport->qos.sched_node) {
-		err = -ENOMEM;
-		goto err_alloc;
-	}
+	return 0;
+}
 
-	vport->qos.sched_node->vport = vport;
+static int mlx5_esw_qos_vport_enable(struct mlx5_vport *vport, enum sched_node_type type,
+				     struct mlx5_esw_sched_node *parent, u32 max_rate,
+				     u32 min_rate, struct netlink_ext_ack *extack)
+{
+	struct mlx5_eswitch *esw = vport->dev->priv.eswitch;
+	struct mlx5_esw_sched_node *sched_node;
+	int err;
 
-	trace_mlx5_esw_vport_qos_create(vport->dev, vport, bw_share, max_rate);
+	esw_assert_qos_lock_held(esw);
+	err = esw_qos_get(esw, extack);
+	if (err)
+		return err;
 
-	return 0;
+	parent = parent ?: esw->qos.node0;
+	sched_node = __esw_qos_alloc_node(parent->esw, 0, type, parent);
+	if (!sched_node)
+		return -ENOMEM;
 
-err_alloc:
-	if (mlx5_destroy_scheduling_element_cmd(esw->dev,
-						SCHEDULING_HIERARCHY_E_SWITCH, sched_elem_ix))
-		esw_warn(esw->dev, "E-Switch destroy vport scheduling element failed.\n");
-err_out:
-	esw_qos_put(esw);
+	sched_node->max_rate = max_rate;
+	sched_node->min_rate = min_rate;
+	sched_node->vport = vport;
+	vport->qos.sched_node = sched_node;
+	err = esw_qos_vport_enable(vport, parent, extack);
+	if (err)
+		esw_qos_put(esw);
 
 	return err;
 }
@@ -745,51 +600,61 @@ err_out:
 void mlx5_esw_qos_vport_disable(struct mlx5_vport *vport)
 {
 	struct mlx5_eswitch *esw = vport->dev->priv.eswitch;
-	struct mlx5_esw_sched_node *vport_node;
-	struct mlx5_core_dev *dev;
-	int err;
+	struct mlx5_esw_sched_node *parent;
 
 	lockdep_assert_held(&esw->state_lock);
 	esw_qos_lock(esw);
-	vport_node = vport->qos.sched_node;
-	if (!vport_node)
+	if (!vport->qos.sched_node)
 		goto unlock;
-	WARN(vport_node->parent != esw->qos.node0,
-	     "Disabling QoS on port before detaching it from node");
-
-	dev = vport_node->esw->dev;
-	trace_mlx5_esw_vport_qos_destroy(dev, vport);
-
-	err = mlx5_destroy_scheduling_element_cmd(dev,
-						  SCHEDULING_HIERARCHY_E_SWITCH,
-						  vport_node->ix);
-	if (err)
-		esw_warn(dev,
-			 "E-Switch destroy vport scheduling element failed (vport=%d,err=%d)\n",
-			 vport->vport, err);
 
-	__esw_qos_free_node(vport_node);
-	memset(&vport->qos, 0, sizeof(vport->qos));
+	parent = vport->qos.sched_node->parent;
+	WARN(parent != esw->qos.node0, "Disabling QoS on port before detaching it from node");
 
+	esw_qos_vport_disable(vport, NULL);
+	mlx5_esw_qos_vport_qos_free(vport);
 	esw_qos_put(esw);
 unlock:
 	esw_qos_unlock(esw);
 }
 
+static int mlx5_esw_qos_set_vport_max_rate(struct mlx5_vport *vport, u32 max_rate,
+					   struct netlink_ext_ack *extack)
+{
+	struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node;
+
+	esw_assert_qos_lock_held(vport->dev->priv.eswitch);
+
+	if (!vport_node)
+		return mlx5_esw_qos_vport_enable(vport, SCHED_NODE_TYPE_VPORT, NULL, max_rate, 0,
+						 extack);
+	else
+		return esw_qos_sched_elem_config(vport_node, max_rate, vport_node->bw_share,
+						 extack);
+}
+
+static int mlx5_esw_qos_set_vport_min_rate(struct mlx5_vport *vport, u32 min_rate,
+					   struct netlink_ext_ack *extack)
+{
+	struct mlx5_esw_sched_node *vport_node = vport->qos.sched_node;
+
+	esw_assert_qos_lock_held(vport->dev->priv.eswitch);
+
+	if (!vport_node)
+		return mlx5_esw_qos_vport_enable(vport, SCHED_NODE_TYPE_VPORT, NULL, 0, min_rate,
+						 extack);
+	else
+		return esw_qos_set_node_min_rate(vport_node, min_rate, extack);
+}
+
 int mlx5_esw_qos_set_vport_rate(struct mlx5_vport *vport, u32 max_rate, u32 min_rate)
 {
 	struct mlx5_eswitch *esw = vport->dev->priv.eswitch;
 	int err;
 
 	esw_qos_lock(esw);
-	err = esw_qos_vport_enable(vport, 0, 0, NULL);
-	if (err)
-		goto unlock;
-
-	err = esw_qos_set_vport_min_rate(vport, min_rate, NULL);
+	err = mlx5_esw_qos_set_vport_min_rate(vport, min_rate, NULL);
 	if (!err)
-		err = esw_qos_set_vport_max_rate(vport, max_rate, NULL);
-unlock:
+		err = mlx5_esw_qos_set_vport_max_rate(vport, max_rate, NULL);
 	esw_qos_unlock(esw);
 	return err;
 }
@@ -809,6 +674,31 @@ bool mlx5_esw_qos_get_vport_rate(struct mlx5_vport *vport, u32 *max_rate, u32 *m
 	return enabled;
 }
 
+static int esw_qos_vport_update_parent(struct mlx5_vport *vport, struct mlx5_esw_sched_node *parent,
+				       struct netlink_ext_ack *extack)
+{
+	struct mlx5_eswitch *esw = vport->dev->priv.eswitch;
+	struct mlx5_esw_sched_node *curr_parent;
+	int err;
+
+	esw_assert_qos_lock_held(esw);
+	curr_parent = vport->qos.sched_node->parent;
+	parent = parent ?: esw->qos.node0;
+	if (curr_parent == parent)
+		return 0;
+
+	esw_qos_vport_disable(vport, extack);
+
+	err = esw_qos_vport_enable(vport, parent, extack);
+	if (err) {
+		if (esw_qos_vport_enable(vport, curr_parent, NULL))
+			esw_warn(parent->esw->dev, "vport restore QoS failed (vport=%d)\n",
+				 vport->vport);
+	}
+
+	return err;
+}
+
 static u32 mlx5_esw_qos_lag_link_speed_get_locked(struct mlx5_core_dev *mdev)
 {
 	struct ethtool_link_ksettings lksettings;
@@ -875,10 +765,8 @@ static int mlx5_esw_qos_link_speed_verify(struct mlx5_core_dev *mdev,
 
 int mlx5_esw_qos_modify_vport_rate(struct mlx5_eswitch *esw, u16 vport_num, u32 rate_mbps)
 {
-	u32 ctx[MLX5_ST_SZ_DW(scheduling_context)] = {};
 	struct mlx5_vport *vport;
 	u32 link_speed_max;
-	u32 bitmask;
 	int err;
 
 	vport = mlx5_eswitch_get_vport(esw, vport_num);
@@ -897,20 +785,7 @@ int mlx5_esw_qos_modify_vport_rate(struct mlx5_eswitch *esw, u16 vport_num, u32
 	}
 
 	esw_qos_lock(esw);
-	if (!vport->qos.sched_node) {
-		/* Eswitch QoS wasn't enabled yet. Enable it and vport QoS. */
-		err = esw_qos_vport_enable(vport, rate_mbps, 0, NULL);
-	} else {
-		struct mlx5_core_dev *dev = vport->qos.sched_node->parent->esw->dev;
-
-		MLX5_SET(scheduling_context, ctx, max_average_bw, rate_mbps);
-		bitmask = MODIFY_SCHEDULING_ELEMENT_IN_MODIFY_BITMASK_MAX_AVERAGE_BW;
-		err = mlx5_modify_scheduling_element_cmd(dev,
-							 SCHEDULING_HIERARCHY_E_SWITCH,
-							 ctx,
-							 vport->qos.sched_node->ix,
-							 bitmask);
-	}
+	err = mlx5_esw_qos_set_vport_max_rate(vport, rate_mbps, NULL);
 	esw_qos_unlock(esw);
 
 	return err;
@@ -981,12 +856,7 @@ int mlx5_esw_devlink_rate_leaf_tx_share_set(struct devlink_rate *rate_leaf, void
 		return err;
 
 	esw_qos_lock(esw);
-	err = esw_qos_vport_enable(vport, 0, 0, extack);
-	if (err)
-		goto unlock;
-
-	err = esw_qos_set_vport_min_rate(vport, tx_share, extack);
-unlock:
+	err = mlx5_esw_qos_set_vport_min_rate(vport, tx_share, extack);
 	esw_qos_unlock(esw);
 	return err;
 }
@@ -1007,12 +877,7 @@ int mlx5_esw_devlink_rate_leaf_tx_max_set(struct devlink_rate *rate_leaf, void *
 		return err;
 
 	esw_qos_lock(esw);
-	err = esw_qos_vport_enable(vport, 0, 0, extack);
-	if (err)
-		goto unlock;
-
-	err = esw_qos_set_vport_max_rate(vport, tx_max, extack);
-unlock:
+	err = mlx5_esw_qos_set_vport_max_rate(vport, tx_max, extack);
 	esw_qos_unlock(esw);
 	return err;
 }
@@ -1046,7 +911,7 @@ int mlx5_esw_devlink_rate_node_tx_max_set(struct devlink_rate *rate_node, void *
 		return err;
 
 	esw_qos_lock(esw);
-	err = esw_qos_set_node_max_rate(node, tx_max, extack);
+	err = esw_qos_sched_elem_config(node, tx_max, node->bw_share, extack);
 	esw_qos_unlock(esw);
 	return err;
 }
@@ -1087,35 +952,30 @@ int mlx5_esw_devlink_rate_node_del(struct devlink_rate *rate_node, void *priv,
 {
 	struct mlx5_esw_sched_node *node = priv;
 	struct mlx5_eswitch *esw = node->esw;
-	int err;
 
 	esw_qos_lock(esw);
-	err = __esw_qos_destroy_node(node, extack);
+	__esw_qos_destroy_node(node, extack);
 	esw_qos_put(esw);
 	esw_qos_unlock(esw);
-	return err;
+	return 0;
 }
 
-int mlx5_esw_qos_vport_update_node(struct mlx5_vport *vport,
-				   struct mlx5_esw_sched_node *node,
-				   struct netlink_ext_ack *extack)
+int mlx5_esw_qos_vport_update_parent(struct mlx5_vport *vport, struct mlx5_esw_sched_node *parent,
+				     struct netlink_ext_ack *extack)
 {
 	struct mlx5_eswitch *esw = vport->dev->priv.eswitch;
 	int err = 0;
 
-	if (node && node->esw != esw) {
+	if (parent && parent->esw != esw) {
 		NL_SET_ERR_MSG_MOD(extack, "Cross E-Switch scheduling is not supported");
 		return -EOPNOTSUPP;
 	}
 
 	esw_qos_lock(esw);
-	if (!vport->qos.sched_node && !node)
-		goto unlock;
-
-	err = esw_qos_vport_enable(vport, 0, 0, extack);
-	if (!err)
-		err = esw_qos_vport_update_node(vport, node, extack);
-unlock:
+	if (!vport->qos.sched_node && parent)
+		err = mlx5_esw_qos_vport_enable(vport, SCHED_NODE_TYPE_VPORT, parent, 0, 0, extack);
+	else if (vport->qos.sched_node)
+		err = esw_qos_vport_update_parent(vport, parent, extack);
 	esw_qos_unlock(esw);
 	return err;
 }
@@ -1129,8 +989,8 @@ int mlx5_esw_devlink_rate_parent_set(struct devlink_rate *devlink_rate,
 	struct mlx5_vport *vport = priv;
 
 	if (!parent)
-		return mlx5_esw_qos_vport_update_node(vport, NULL, extack);
+		return mlx5_esw_qos_vport_update_parent(vport, NULL, extack);
 
 	node = parent_priv;
-	return mlx5_esw_qos_vport_update_node(vport, node, extack);
+	return mlx5_esw_qos_vport_update_parent(vport, node, extack);
 }
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.h b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.h
index 61a6fdd5c267..6eb8f6a648c8 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/esw/qos.h
@@ -13,6 +13,7 @@ int mlx5_esw_qos_set_vport_rate(struct mlx5_vport *evport, u32 max_rate, u32 min
 bool mlx5_esw_qos_get_vport_rate(struct mlx5_vport *vport, u32 *max_rate, u32 *min_rate);
 void mlx5_esw_qos_vport_disable(struct mlx5_vport *vport);
 
+void mlx5_esw_qos_vport_qos_free(struct mlx5_vport *vport);
 u32 mlx5_esw_qos_vport_get_sched_elem_ix(const struct mlx5_vport *vport);
 struct mlx5_esw_sched_node *mlx5_esw_qos_vport_get_parent(const struct mlx5_vport *vport);
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
index cead41ddbc38..7fb8a3381f84 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c
@@ -1061,8 +1061,7 @@ static void mlx5_eswitch_clear_vf_vports_info(struct mlx5_eswitch *esw)
 	unsigned long i;
 
 	mlx5_esw_for_each_vf_vport(esw, i, vport, esw->esw_funcs.num_vfs) {
-		kfree(vport->qos.sched_node);
-		memset(&vport->qos, 0, sizeof(vport->qos));
+		mlx5_esw_qos_vport_qos_free(vport);
 		memset(&vport->info, 0, sizeof(vport->info));
 		vport->info.link_state = MLX5_VPORT_ADMIN_STATE_AUTO;
 	}
@@ -1074,8 +1073,7 @@ static void mlx5_eswitch_clear_ec_vf_vports_info(struct mlx5_eswitch *esw)
 	unsigned long i;
 
 	mlx5_esw_for_each_ec_vf_vport(esw, i, vport, esw->esw_funcs.num_ec_vfs) {
-		kfree(vport->qos.sched_node);
-		memset(&vport->qos, 0, sizeof(vport->qos));
+		mlx5_esw_qos_vport_qos_free(vport);
 		memset(&vport->info, 0, sizeof(vport->info));
 		vport->info.link_state = MLX5_VPORT_ADMIN_STATE_AUTO;
 	}
@@ -1490,7 +1488,6 @@ int mlx5_eswitch_enable_locked(struct mlx5_eswitch *esw, int num_vfs)
 	if (esw->mode == MLX5_ESWITCH_LEGACY) {
 		err = esw_legacy_enable(esw);
 	} else {
-		mlx5_rescan_drivers(esw->dev);
 		err = esw_offloads_enable(esw);
 	}
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
index 14dd42d44e6f..a83d41121db6 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h
@@ -427,9 +427,8 @@ int mlx5_eswitch_set_vport_trust(struct mlx5_eswitch *esw,
 				 u16 vport_num, bool setting);
 int mlx5_eswitch_set_vport_rate(struct mlx5_eswitch *esw, u16 vport,
 				u32 max_rate, u32 min_rate);
-int mlx5_esw_qos_vport_update_node(struct mlx5_vport *vport,
-				   struct mlx5_esw_sched_node *node,
-				   struct netlink_ext_ack *extack);
+int mlx5_esw_qos_vport_update_parent(struct mlx5_vport *vport, struct mlx5_esw_sched_node *node,
+				     struct netlink_ext_ack *extack);
 int mlx5_eswitch_set_vepa(struct mlx5_eswitch *esw, u8 setting);
 int mlx5_eswitch_get_vepa(struct mlx5_eswitch *esw, u8 *setting);
 int mlx5_eswitch_get_vport_config(struct mlx5_eswitch *esw,
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
index fd34f43d18d5..5f1adebd9669 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch_offloads.c
@@ -2332,18 +2332,35 @@ out_free:
 	return err;
 }
 
+static void esw_mode_change(struct mlx5_eswitch *esw, u16 mode)
+{
+	mlx5_devcom_comp_lock(esw->dev->priv.hca_devcom_comp);
+
+	if (esw->dev->priv.flags & MLX5_PRIV_FLAGS_DISABLE_IB_ADEV) {
+		esw->mode = mode;
+		mlx5_devcom_comp_unlock(esw->dev->priv.hca_devcom_comp);
+		return;
+	}
+
+	esw->dev->priv.flags |= MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
+	mlx5_rescan_drivers_locked(esw->dev);
+	esw->mode = mode;
+	esw->dev->priv.flags &= ~MLX5_PRIV_FLAGS_DISABLE_IB_ADEV;
+	mlx5_rescan_drivers_locked(esw->dev);
+	mlx5_devcom_comp_unlock(esw->dev->priv.hca_devcom_comp);
+}
+
 static int esw_offloads_start(struct mlx5_eswitch *esw,
 			      struct netlink_ext_ack *extack)
 {
 	int err;
 
-	esw->mode = MLX5_ESWITCH_OFFLOADS;
+	esw_mode_change(esw, MLX5_ESWITCH_OFFLOADS);
 	err = mlx5_eswitch_enable_locked(esw, esw->dev->priv.sriov.num_vfs);
 	if (err) {
 		NL_SET_ERR_MSG_MOD(extack,
 				   "Failed setting eswitch to offloads");
-		esw->mode = MLX5_ESWITCH_LEGACY;
-		mlx5_rescan_drivers(esw->dev);
+		esw_mode_change(esw, MLX5_ESWITCH_LEGACY);
 		return err;
 	}
 	if (esw->offloads.inline_mode == MLX5_INLINE_MODE_NONE) {
@@ -3584,7 +3601,7 @@ static int esw_offloads_stop(struct mlx5_eswitch *esw,
 {
 	int err;
 
-	esw->mode = MLX5_ESWITCH_LEGACY;
+	esw_mode_change(esw, MLX5_ESWITCH_LEGACY);
 
 	/* If changing from switchdev to legacy mode without sriov enabled,
 	 * no need to create legacy fdb.
@@ -3770,7 +3787,6 @@ int mlx5_devlink_eswitch_mode_set(struct devlink *devlink, u16 mode,
 		err = esw_offloads_start(esw, extack);
 	} else if (mode == DEVLINK_ESWITCH_MODE_LEGACY) {
 		err = esw_offloads_stop(esw, extack);
-		mlx5_rescan_drivers(esw->dev);
 	} else {
 		err = -EINVAL;
 	}