diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h
index 8d76452cacdc..f1a7bc46f1c0 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h
@@ -345,7 +345,7 @@ struct mlx5e_tx_wqe_info {
 	u8  num_wqebbs;
 	u8  num_dma;
 #ifdef CONFIG_MLX5_EN_TLS
-	skb_frag_t *resync_dump_frag;
+	struct page *resync_dump_frag_page;
 #endif
 };
 
@@ -410,6 +410,7 @@ struct mlx5e_txqsq {
 	struct device             *pdev;
 	__be32                     mkey_be;
 	unsigned long              state;
+	unsigned int               hw_mtu;
 	struct hwtstamp_config    *tstamp;
 	struct mlx5_clock         *clock;
 
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h b/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h
index 87be96747902..7c8796d9743f 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en/txrx.h
@@ -15,15 +15,14 @@
 #else
 /* TLS offload requires additional stop_room for:
  *  - a resync SKB.
- * kTLS offload requires additional stop_room for:
- * - static params WQE,
- * - progress params WQE, and
- * - resync DUMP per frag.
+ * kTLS offload requires fixed additional stop_room for:
+ * - a static params WQE, and a progress params WQE.
+ * The additional MTU-depending room for the resync DUMP WQEs
+ * will be calculated and added in runtime.
  */
 #define MLX5E_SQ_TLS_ROOM  \
 	(MLX5_SEND_WQE_MAX_WQEBBS + \
-	 MLX5E_KTLS_STATIC_WQEBBS + MLX5E_KTLS_PROGRESS_WQEBBS + \
-	 MAX_SKB_FRAGS * MLX5E_KTLS_MAX_DUMP_WQEBBS)
+	 MLX5E_KTLS_STATIC_WQEBBS + MLX5E_KTLS_PROGRESS_WQEBBS)
 #endif
 
 #define INL_HDR_START_SZ (sizeof(((struct mlx5_wqe_eth_seg *)NULL)->inline_hdr.start))
@@ -92,7 +91,7 @@ mlx5e_fill_sq_frag_edge(struct mlx5e_txqsq *sq, struct mlx5_wq_cyc *wq,
 
 	/* fill sq frag edge with nops to avoid wqe wrapping two pages */
 	for (; wi < edge_wi; wi++) {
-		wi->skb        = NULL;
+		memset(wi, 0, sizeof(*wi));
 		wi->num_wqebbs = 1;
 		mlx5e_post_nop(wq, sq->sqn, &sq->pc);
 	}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.c
index d2ff74d52720..46725cd743a3 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.c
@@ -38,7 +38,7 @@ static int mlx5e_ktls_add(struct net_device *netdev, struct sock *sk,
 		return -ENOMEM;
 
 	tx_priv->expected_seq = start_offload_tcp_sn;
-	tx_priv->crypto_info  = crypto_info;
+	tx_priv->crypto_info  = *(struct tls12_crypto_info_aes_gcm_128 *)crypto_info;
 	mlx5e_set_ktls_tx_priv_ctx(tls_ctx, tx_priv);
 
 	/* tc and underlay_qpn values are not in use for tls tis */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
index b7298f9ee3d3..a3efa29a4629 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls.h
@@ -21,7 +21,14 @@
 	 MLX5_ST_SZ_BYTES(tls_progress_params))
 #define MLX5E_KTLS_PROGRESS_WQEBBS \
 	(DIV_ROUND_UP(MLX5E_KTLS_PROGRESS_WQE_SZ, MLX5_SEND_WQE_BB))
-#define MLX5E_KTLS_MAX_DUMP_WQEBBS 2
+
+struct mlx5e_dump_wqe {
+	struct mlx5_wqe_ctrl_seg ctrl;
+	struct mlx5_wqe_data_seg data;
+};
+
+#define MLX5E_KTLS_DUMP_WQEBBS \
+	(DIV_ROUND_UP(sizeof(struct mlx5e_dump_wqe), MLX5_SEND_WQE_BB))
 
 enum {
 	MLX5E_TLS_PROGRESS_PARAMS_AUTH_STATE_NO_OFFLOAD     = 0,
@@ -37,7 +44,7 @@ enum {
 
 struct mlx5e_ktls_offload_context_tx {
 	struct tls_offload_context_tx *tx_ctx;
-	struct tls_crypto_info *crypto_info;
+	struct tls12_crypto_info_aes_gcm_128 crypto_info;
 	u32 expected_seq;
 	u32 tisn;
 	u32 key_id;
@@ -86,14 +93,28 @@ struct sk_buff *mlx5e_ktls_handle_tx_skb(struct net_device *netdev,
 					 struct mlx5e_tx_wqe **wqe, u16 *pi);
 void mlx5e_ktls_tx_handle_resync_dump_comp(struct mlx5e_txqsq *sq,
 					   struct mlx5e_tx_wqe_info *wi,
-					   struct mlx5e_sq_dma *dma);
-
+					   u32 *dma_fifo_cc);
+static inline u8
+mlx5e_ktls_dumps_num_wqebbs(struct mlx5e_txqsq *sq, unsigned int nfrags,
+			    unsigned int sync_len)
+{
+	/* Given the MTU and sync_len, calculates an upper bound for the
+	 * number of WQEBBs needed for the TX resync DUMP WQEs of a record.
+	 */
+	return MLX5E_KTLS_DUMP_WQEBBS *
+		(nfrags + DIV_ROUND_UP(sync_len, sq->hw_mtu));
+}
 #else
 
 static inline void mlx5e_ktls_build_netdev(struct mlx5e_priv *priv)
 {
 }
 
+static inline void
+mlx5e_ktls_tx_handle_resync_dump_comp(struct mlx5e_txqsq *sq,
+				      struct mlx5e_tx_wqe_info *wi,
+				      u32 *dma_fifo_cc) {}
+
 #endif
 
 #endif /* __MLX5E_TLS_H__ */
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
index d195366461c9..778dab1af8fc 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_accel/ktls_tx.c
@@ -24,17 +24,12 @@ enum {
 static void
 fill_static_params_ctx(void *ctx, struct mlx5e_ktls_offload_context_tx *priv_tx)
 {
-	struct tls_crypto_info *crypto_info = priv_tx->crypto_info;
-	struct tls12_crypto_info_aes_gcm_128 *info;
+	struct tls12_crypto_info_aes_gcm_128 *info = &priv_tx->crypto_info;
 	char *initial_rn, *gcm_iv;
 	u16 salt_sz, rec_seq_sz;
 	char *salt, *rec_seq;
 	u8 tls_version;
 
-	if (WARN_ON(crypto_info->cipher_type != TLS_CIPHER_AES_GCM_128))
-		return;
-
-	info = (struct tls12_crypto_info_aes_gcm_128 *)crypto_info;
 	EXTRACT_INFO_FIELDS;
 
 	gcm_iv      = MLX5_ADDR_OF(tls_static_params, ctx, gcm_iv);
@@ -108,16 +103,15 @@ build_progress_params(struct mlx5e_tx_wqe *wqe, u16 pc, u32 sqn,
 }
 
 static void tx_fill_wi(struct mlx5e_txqsq *sq,
-		       u16 pi, u8 num_wqebbs,
-		       skb_frag_t *resync_dump_frag,
-		       u32 num_bytes)
+		       u16 pi, u8 num_wqebbs, u32 num_bytes,
+		       struct page *page)
 {
 	struct mlx5e_tx_wqe_info *wi = &sq->db.wqe_info[pi];
 
-	wi->skb              = NULL;
-	wi->num_wqebbs       = num_wqebbs;
-	wi->resync_dump_frag = resync_dump_frag;
-	wi->num_bytes        = num_bytes;
+	memset(wi, 0, sizeof(*wi));
+	wi->num_wqebbs = num_wqebbs;
+	wi->num_bytes  = num_bytes;
+	wi->resync_dump_frag_page = page;
 }
 
 void mlx5e_ktls_tx_offload_set_pending(struct mlx5e_ktls_offload_context_tx *priv_tx)
@@ -145,7 +139,7 @@ post_static_params(struct mlx5e_txqsq *sq,
 
 	umr_wqe = mlx5e_sq_fetch_wqe(sq, MLX5E_KTLS_STATIC_UMR_WQE_SZ, &pi);
 	build_static_params(umr_wqe, sq->pc, sq->sqn, priv_tx, fence);
-	tx_fill_wi(sq, pi, MLX5E_KTLS_STATIC_WQEBBS, NULL, 0);
+	tx_fill_wi(sq, pi, MLX5E_KTLS_STATIC_WQEBBS, 0, NULL);
 	sq->pc += MLX5E_KTLS_STATIC_WQEBBS;
 }
 
@@ -159,7 +153,7 @@ post_progress_params(struct mlx5e_txqsq *sq,
 
 	wqe = mlx5e_sq_fetch_wqe(sq, MLX5E_KTLS_PROGRESS_WQE_SZ, &pi);
 	build_progress_params(wqe, sq->pc, sq->sqn, priv_tx, fence);
-	tx_fill_wi(sq, pi, MLX5E_KTLS_PROGRESS_WQEBBS, NULL, 0);
+	tx_fill_wi(sq, pi, MLX5E_KTLS_PROGRESS_WQEBBS, 0, NULL);
 	sq->pc += MLX5E_KTLS_PROGRESS_WQEBBS;
 }
 
@@ -169,6 +163,14 @@ mlx5e_ktls_tx_post_param_wqes(struct mlx5e_txqsq *sq,
 			      bool skip_static_post, bool fence_first_post)
 {
 	bool progress_fence = skip_static_post || !fence_first_post;
+	struct mlx5_wq_cyc *wq = &sq->wq;
+	u16 contig_wqebbs_room, pi;
+
+	pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
+	contig_wqebbs_room = mlx5_wq_cyc_get_contig_wqebbs(wq, pi);
+	if (unlikely(contig_wqebbs_room <
+		     MLX5E_KTLS_STATIC_WQEBBS + MLX5E_KTLS_PROGRESS_WQEBBS))
+		mlx5e_fill_sq_frag_edge(sq, wq, pi, contig_wqebbs_room);
 
 	if (!skip_static_post)
 		post_static_params(sq, priv_tx, fence_first_post);
@@ -180,29 +182,36 @@ struct tx_sync_info {
 	u64 rcd_sn;
 	s32 sync_len;
 	int nr_frags;
-	skb_frag_t *frags[MAX_SKB_FRAGS];
+	skb_frag_t frags[MAX_SKB_FRAGS];
 };
 
-static bool tx_sync_info_get(struct mlx5e_ktls_offload_context_tx *priv_tx,
-			     u32 tcp_seq, struct tx_sync_info *info)
+enum mlx5e_ktls_sync_retval {
+	MLX5E_KTLS_SYNC_DONE,
+	MLX5E_KTLS_SYNC_FAIL,
+	MLX5E_KTLS_SYNC_SKIP_NO_DATA,
+};
+
+static enum mlx5e_ktls_sync_retval
+tx_sync_info_get(struct mlx5e_ktls_offload_context_tx *priv_tx,
+		 u32 tcp_seq, struct tx_sync_info *info)
 {
 	struct tls_offload_context_tx *tx_ctx = priv_tx->tx_ctx;
+	enum mlx5e_ktls_sync_retval ret = MLX5E_KTLS_SYNC_DONE;
 	struct tls_record_info *record;
 	int remaining, i = 0;
 	unsigned long flags;
-	bool ret = true;
 
 	spin_lock_irqsave(&tx_ctx->lock, flags);
 	record = tls_get_record(tx_ctx, tcp_seq, &info->rcd_sn);
 
 	if (unlikely(!record)) {
-		ret = false;
+		ret = MLX5E_KTLS_SYNC_FAIL;
 		goto out;
 	}
 
 	if (unlikely(tcp_seq < tls_record_start_seq(record))) {
-		if (!tls_record_is_start_marker(record))
-			ret = false;
+		ret = tls_record_is_start_marker(record) ?
+			MLX5E_KTLS_SYNC_SKIP_NO_DATA : MLX5E_KTLS_SYNC_FAIL;
 		goto out;
 	}
 
@@ -211,13 +220,13 @@ static bool tx_sync_info_get(struct mlx5e_ktls_offload_context_tx *priv_tx,
 	while (remaining > 0) {
 		skb_frag_t *frag = &record->frags[i];
 
-		__skb_frag_ref(frag);
+		get_page(skb_frag_page(frag));
 		remaining -= skb_frag_size(frag);
-		info->frags[i++] = frag;
+		info->frags[i++] = *frag;
 	}
 	/* reduce the part which will be sent with the original SKB */
 	if (remaining < 0)
-		skb_frag_size_add(info->frags[i - 1], remaining);
+		skb_frag_size_add(&info->frags[i - 1], remaining);
 	info->nr_frags = i;
 out:
 	spin_unlock_irqrestore(&tx_ctx->lock, flags);
@@ -229,17 +238,12 @@ tx_post_resync_params(struct mlx5e_txqsq *sq,
 		      struct mlx5e_ktls_offload_context_tx *priv_tx,
 		      u64 rcd_sn)
 {
-	struct tls_crypto_info *crypto_info = priv_tx->crypto_info;
-	struct tls12_crypto_info_aes_gcm_128 *info;
+	struct tls12_crypto_info_aes_gcm_128 *info = &priv_tx->crypto_info;
 	__be64 rn_be = cpu_to_be64(rcd_sn);
 	bool skip_static_post;
 	u16 rec_seq_sz;
 	char *rec_seq;
 
-	if (WARN_ON(crypto_info->cipher_type != TLS_CIPHER_AES_GCM_128))
-		return;
-
-	info = (struct tls12_crypto_info_aes_gcm_128 *)crypto_info;
 	rec_seq = info->rec_seq;
 	rec_seq_sz = sizeof(info->rec_seq);
 
@@ -250,11 +254,6 @@ tx_post_resync_params(struct mlx5e_txqsq *sq,
 	mlx5e_ktls_tx_post_param_wqes(sq, priv_tx, skip_static_post, true);
 }
 
-struct mlx5e_dump_wqe {
-	struct mlx5_wqe_ctrl_seg ctrl;
-	struct mlx5_wqe_data_seg data;
-};
-
 static int
 tx_post_resync_dump(struct mlx5e_txqsq *sq, skb_frag_t *frag, u32 tisn, bool first)
 {
@@ -262,7 +261,6 @@ tx_post_resync_dump(struct mlx5e_txqsq *sq, skb_frag_t *frag, u32 tisn, bool fir
 	struct mlx5_wqe_data_seg *dseg;
 	struct mlx5e_dump_wqe *wqe;
 	dma_addr_t dma_addr = 0;
-	u8  num_wqebbs;
 	u16 ds_cnt;
 	int fsz;
 	u16 pi;
@@ -270,7 +268,6 @@ tx_post_resync_dump(struct mlx5e_txqsq *sq, skb_frag_t *frag, u32 tisn, bool fir
 	wqe = mlx5e_sq_fetch_wqe(sq, sizeof(*wqe), &pi);
 
 	ds_cnt = sizeof(*wqe) / MLX5_SEND_WQE_DS;
-	num_wqebbs = DIV_ROUND_UP(ds_cnt, MLX5_SEND_WQEBB_NUM_DS);
 
 	cseg = &wqe->ctrl;
 	dseg = &wqe->data;
@@ -291,24 +288,27 @@ tx_post_resync_dump(struct mlx5e_txqsq *sq, skb_frag_t *frag, u32 tisn, bool fir
 	dseg->byte_count = cpu_to_be32(fsz);
 	mlx5e_dma_push(sq, dma_addr, fsz, MLX5E_DMA_MAP_PAGE);
 
-	tx_fill_wi(sq, pi, num_wqebbs, frag, fsz);
-	sq->pc += num_wqebbs;
-
-	WARN(num_wqebbs > MLX5E_KTLS_MAX_DUMP_WQEBBS,
-	     "unexpected DUMP num_wqebbs, %d > %d",
-	     num_wqebbs, MLX5E_KTLS_MAX_DUMP_WQEBBS);
+	tx_fill_wi(sq, pi, MLX5E_KTLS_DUMP_WQEBBS, fsz, skb_frag_page(frag));
+	sq->pc += MLX5E_KTLS_DUMP_WQEBBS;
 
 	return 0;
 }
 
 void mlx5e_ktls_tx_handle_resync_dump_comp(struct mlx5e_txqsq *sq,
 					   struct mlx5e_tx_wqe_info *wi,
-					   struct mlx5e_sq_dma *dma)
+					   u32 *dma_fifo_cc)
 {
-	struct mlx5e_sq_stats *stats = sq->stats;
+	struct mlx5e_sq_stats *stats;
+	struct mlx5e_sq_dma *dma;
+
+	if (!wi->resync_dump_frag_page)
+		return;
+
+	dma = mlx5e_dma_get(sq, (*dma_fifo_cc)++);
+	stats = sq->stats;
 
 	mlx5e_tx_dma_unmap(sq->pdev, dma);
-	__skb_frag_unref(wi->resync_dump_frag);
+	put_page(wi->resync_dump_frag_page);
 	stats->tls_dump_packets++;
 	stats->tls_dump_bytes += wi->num_bytes;
 }
@@ -318,25 +318,31 @@ static void tx_post_fence_nop(struct mlx5e_txqsq *sq)
 	struct mlx5_wq_cyc *wq = &sq->wq;
 	u16 pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
 
-	tx_fill_wi(sq, pi, 1, NULL, 0);
+	tx_fill_wi(sq, pi, 1, 0, NULL);
 
 	mlx5e_post_nop_fence(wq, sq->sqn, &sq->pc);
 }
 
-static struct sk_buff *
+static enum mlx5e_ktls_sync_retval
 mlx5e_ktls_tx_handle_ooo(struct mlx5e_ktls_offload_context_tx *priv_tx,
 			 struct mlx5e_txqsq *sq,
-			 struct sk_buff *skb,
+			 int datalen,
 			 u32 seq)
 {
 	struct mlx5e_sq_stats *stats = sq->stats;
 	struct mlx5_wq_cyc *wq = &sq->wq;
+	enum mlx5e_ktls_sync_retval ret;
 	struct tx_sync_info info = {};
 	u16 contig_wqebbs_room, pi;
 	u8 num_wqebbs;
-	int i;
+	int i = 0;
 
-	if (!tx_sync_info_get(priv_tx, seq, &info)) {
+	ret = tx_sync_info_get(priv_tx, seq, &info);
+	if (unlikely(ret != MLX5E_KTLS_SYNC_DONE)) {
+		if (ret == MLX5E_KTLS_SYNC_SKIP_NO_DATA) {
+			stats->tls_skip_no_sync_data++;
+			return MLX5E_KTLS_SYNC_SKIP_NO_DATA;
+		}
 		/* We might get here if a retransmission reaches the driver
 		 * after the relevant record is acked.
 		 * It should be safe to drop the packet in this case
@@ -346,13 +352,8 @@ mlx5e_ktls_tx_handle_ooo(struct mlx5e_ktls_offload_context_tx *priv_tx,
 	}
 
 	if (unlikely(info.sync_len < 0)) {
-		u32 payload;
-		int headln;
-
-		headln = skb_transport_offset(skb) + tcp_hdrlen(skb);
-		payload = skb->len - headln;
-		if (likely(payload <= -info.sync_len))
-			return skb;
+		if (likely(datalen <= -info.sync_len))
+			return MLX5E_KTLS_SYNC_DONE;
 
 		stats->tls_drop_bypass_req++;
 		goto err_out;
@@ -360,30 +361,62 @@ mlx5e_ktls_tx_handle_ooo(struct mlx5e_ktls_offload_context_tx *priv_tx,
 
 	stats->tls_ooo++;
 
-	num_wqebbs = MLX5E_KTLS_STATIC_WQEBBS + MLX5E_KTLS_PROGRESS_WQEBBS +
-		(info.nr_frags ? info.nr_frags * MLX5E_KTLS_MAX_DUMP_WQEBBS : 1);
+	tx_post_resync_params(sq, priv_tx, info.rcd_sn);
+
+	/* If no dump WQE was sent, we need to have a fence NOP WQE before the
+	 * actual data xmit.
+	 */
+	if (!info.nr_frags) {
+		tx_post_fence_nop(sq);
+		return MLX5E_KTLS_SYNC_DONE;
+	}
+
+	num_wqebbs = mlx5e_ktls_dumps_num_wqebbs(sq, info.nr_frags, info.sync_len);
 	pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
 	contig_wqebbs_room = mlx5_wq_cyc_get_contig_wqebbs(wq, pi);
+
 	if (unlikely(contig_wqebbs_room < num_wqebbs))
 		mlx5e_fill_sq_frag_edge(sq, wq, pi, contig_wqebbs_room);
 
 	tx_post_resync_params(sq, priv_tx, info.rcd_sn);
 
-	for (i = 0; i < info.nr_frags; i++)
-		if (tx_post_resync_dump(sq, info.frags[i], priv_tx->tisn, !i))
-			goto err_out;
+	for (; i < info.nr_frags; i++) {
+		unsigned int orig_fsz, frag_offset = 0, n = 0;
+		skb_frag_t *f = &info.frags[i];
 
-	/* If no dump WQE was sent, we need to have a fence NOP WQE before the
-	 * actual data xmit.
-	 */
-	if (!info.nr_frags)
-		tx_post_fence_nop(sq);
+		orig_fsz = skb_frag_size(f);
 
-	return skb;
+		do {
+			bool fence = !(i || frag_offset);
+			unsigned int fsz;
+
+			n++;
+			fsz = min_t(unsigned int, sq->hw_mtu, orig_fsz - frag_offset);
+			skb_frag_size_set(f, fsz);
+			if (tx_post_resync_dump(sq, f, priv_tx->tisn, fence)) {
+				page_ref_add(skb_frag_page(f), n - 1);
+				goto err_out;
+			}
+
+			skb_frag_off_add(f, fsz);
+			frag_offset += fsz;
+		} while (frag_offset < orig_fsz);
+
+		page_ref_add(skb_frag_page(f), n - 1);
+	}
+
+	return MLX5E_KTLS_SYNC_DONE;
 
 err_out:
-	dev_kfree_skb_any(skb);
-	return NULL;
+	for (; i < info.nr_frags; i++)
+		/* The put_page() here undoes the page ref obtained in tx_sync_info_get().
+		 * Page refs obtained for the DUMP WQEs above (by page_ref_add) will be
+		 * released only upon their completions (or in mlx5e_free_txqsq_descs,
+		 * if channel closes).
+		 */
+		put_page(skb_frag_page(&info.frags[i]));
+
+	return MLX5E_KTLS_SYNC_FAIL;
 }
 
 struct sk_buff *mlx5e_ktls_handle_tx_skb(struct net_device *netdev,
@@ -419,10 +452,15 @@ struct sk_buff *mlx5e_ktls_handle_tx_skb(struct net_device *netdev,
 
 	seq = ntohl(tcp_hdr(skb)->seq);
 	if (unlikely(priv_tx->expected_seq != seq)) {
-		skb = mlx5e_ktls_tx_handle_ooo(priv_tx, sq, skb, seq);
-		if (unlikely(!skb))
+		enum mlx5e_ktls_sync_retval ret =
+			mlx5e_ktls_tx_handle_ooo(priv_tx, sq, datalen, seq);
+
+		if (likely(ret == MLX5E_KTLS_SYNC_DONE))
+			*wqe = mlx5e_sq_fetch_wqe(sq, sizeof(**wqe), pi);
+		else if (ret == MLX5E_KTLS_SYNC_FAIL)
+			goto err_out;
+		else /* ret == MLX5E_KTLS_SYNC_SKIP_NO_DATA */
 			goto out;
-		*wqe = mlx5e_sq_fetch_wqe(sq, sizeof(**wqe), pi);
 	}
 
 	priv_tx->expected_seq = seq + datalen;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
index 7569287f8f3c..772bfdbdeb9c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c
@@ -1128,6 +1128,7 @@ static int mlx5e_alloc_txqsq(struct mlx5e_channel *c,
 	sq->txq_ix    = txq_ix;
 	sq->uar_map   = mdev->mlx5e_res.bfreg.map;
 	sq->min_inline_mode = params->tx_min_inline_mode;
+	sq->hw_mtu    = MLX5E_SW2HW_MTU(params, params->sw_mtu);
 	sq->stats     = &c->priv->channel_stats[c->ix].sq[tc];
 	sq->stop_room = MLX5E_SQ_STOP_ROOM;
 	INIT_WORK(&sq->recover_work, mlx5e_tx_err_cqe_work);
@@ -1135,10 +1136,14 @@ static int mlx5e_alloc_txqsq(struct mlx5e_channel *c,
 		set_bit(MLX5E_SQ_STATE_VLAN_NEED_L2_INLINE, &sq->state);
 	if (MLX5_IPSEC_DEV(c->priv->mdev))
 		set_bit(MLX5E_SQ_STATE_IPSEC, &sq->state);
+#ifdef CONFIG_MLX5_EN_TLS
 	if (mlx5_accel_is_tls_device(c->priv->mdev)) {
 		set_bit(MLX5E_SQ_STATE_TLS, &sq->state);
-		sq->stop_room += MLX5E_SQ_TLS_ROOM;
+		sq->stop_room += MLX5E_SQ_TLS_ROOM +
+			mlx5e_ktls_dumps_num_wqebbs(sq, MAX_SKB_FRAGS,
+						    TLS_MAX_PAYLOAD_SIZE);
 	}
+#endif
 
 	param->wq.db_numa_node = cpu_to_node(c->cpu);
 	err = mlx5_wq_cyc_create(mdev, &param->wq, sqc_wq, wq, &sq->wq_ctrl);
@@ -1349,9 +1354,13 @@ static void mlx5e_deactivate_txqsq(struct mlx5e_txqsq *sq)
 	/* last doorbell out, godspeed .. */
 	if (mlx5e_wqc_has_room_for(wq, sq->cc, sq->pc, 1)) {
 		u16 pi = mlx5_wq_cyc_ctr2ix(wq, sq->pc);
+		struct mlx5e_tx_wqe_info *wi;
 		struct mlx5e_tx_wqe *nop;
 
-		sq->db.wqe_info[pi].skb = NULL;
+		wi = &sq->db.wqe_info[pi];
+
+		memset(wi, 0, sizeof(*wi));
+		wi->num_wqebbs = 1;
 		nop = mlx5e_post_nop(wq, sq->sqn, &sq->pc);
 		mlx5e_notify_hw(wq, sq->pc, sq->uar_map, &nop->ctrl);
 	}
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c
index ac6fdcda7019..7e6ebd0505cc 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c
@@ -52,11 +52,12 @@ static const struct counter_desc sw_stats_desc[] = {
 	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_encrypted_bytes) },
 	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_ctx) },
 	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_ooo) },
-	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_resync_bytes) },
-	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_drop_no_sync_data) },
-	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_drop_bypass_req) },
 	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_dump_packets) },
 	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_dump_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_resync_bytes) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_skip_no_sync_data) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_drop_no_sync_data) },
+	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, tx_tls_drop_bypass_req) },
 #endif
 
 	{ MLX5E_DECLARE_STAT(struct mlx5e_sw_stats, rx_lro_packets) },
@@ -288,11 +289,12 @@ static void mlx5e_grp_sw_update_stats(struct mlx5e_priv *priv)
 			s->tx_tls_encrypted_bytes   += sq_stats->tls_encrypted_bytes;
 			s->tx_tls_ctx               += sq_stats->tls_ctx;
 			s->tx_tls_ooo               += sq_stats->tls_ooo;
-			s->tx_tls_resync_bytes      += sq_stats->tls_resync_bytes;
-			s->tx_tls_drop_no_sync_data += sq_stats->tls_drop_no_sync_data;
-			s->tx_tls_drop_bypass_req   += sq_stats->tls_drop_bypass_req;
 			s->tx_tls_dump_bytes        += sq_stats->tls_dump_bytes;
 			s->tx_tls_dump_packets      += sq_stats->tls_dump_packets;
+			s->tx_tls_resync_bytes      += sq_stats->tls_resync_bytes;
+			s->tx_tls_skip_no_sync_data += sq_stats->tls_skip_no_sync_data;
+			s->tx_tls_drop_no_sync_data += sq_stats->tls_drop_no_sync_data;
+			s->tx_tls_drop_bypass_req   += sq_stats->tls_drop_bypass_req;
 #endif
 			s->tx_cqes		+= sq_stats->cqes;
 		}
@@ -1472,10 +1474,12 @@ static const struct counter_desc sq_stats_desc[] = {
 	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_encrypted_bytes) },
 	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_ctx) },
 	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_ooo) },
-	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_drop_no_sync_data) },
-	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_drop_bypass_req) },
 	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_dump_packets) },
 	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_dump_bytes) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_resync_bytes) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_skip_no_sync_data) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_drop_no_sync_data) },
+	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, tls_drop_bypass_req) },
 #endif
 	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, csum_none) },
 	{ MLX5E_DECLARE_TX_STAT(struct mlx5e_sq_stats, stopped) },
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.h b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.h
index 79f261bf86ac..869f3502f631 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.h
@@ -129,11 +129,12 @@ struct mlx5e_sw_stats {
 	u64 tx_tls_encrypted_bytes;
 	u64 tx_tls_ctx;
 	u64 tx_tls_ooo;
-	u64 tx_tls_resync_bytes;
-	u64 tx_tls_drop_no_sync_data;
-	u64 tx_tls_drop_bypass_req;
 	u64 tx_tls_dump_packets;
 	u64 tx_tls_dump_bytes;
+	u64 tx_tls_resync_bytes;
+	u64 tx_tls_skip_no_sync_data;
+	u64 tx_tls_drop_no_sync_data;
+	u64 tx_tls_drop_bypass_req;
 #endif
 
 	u64 rx_xsk_packets;
@@ -273,11 +274,12 @@ struct mlx5e_sq_stats {
 	u64 tls_encrypted_bytes;
 	u64 tls_ctx;
 	u64 tls_ooo;
-	u64 tls_resync_bytes;
-	u64 tls_drop_no_sync_data;
-	u64 tls_drop_bypass_req;
 	u64 tls_dump_packets;
 	u64 tls_dump_bytes;
+	u64 tls_resync_bytes;
+	u64 tls_skip_no_sync_data;
+	u64 tls_drop_no_sync_data;
+	u64 tls_drop_bypass_req;
 #endif
 	/* less likely accessed in data path */
 	u64 csum_none;
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
index d3a67a9b4eba..67dc4f0921b6 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/en_tx.c
@@ -403,7 +403,10 @@ netdev_tx_t mlx5e_xmit(struct sk_buff *skb, struct net_device *dev)
 static void mlx5e_dump_error_cqe(struct mlx5e_txqsq *sq,
 				 struct mlx5_err_cqe *err_cqe)
 {
-	u32 ci = mlx5_cqwq_get_ci(&sq->cq.wq);
+	struct mlx5_cqwq *wq = &sq->cq.wq;
+	u32 ci;
+
+	ci = mlx5_cqwq_ctr2ix(wq, wq->cc - 1);
 
 	netdev_err(sq->channel->netdev,
 		   "Error cqe on cqn 0x%x, ci 0x%x, sqn 0x%x, opcode 0x%x, syndrome 0x%x, vendor syndrome 0x%x\n",
@@ -479,14 +482,7 @@ bool mlx5e_poll_tx_cq(struct mlx5e_cq *cq, int napi_budget)
 			skb = wi->skb;
 
 			if (unlikely(!skb)) {
-#ifdef CONFIG_MLX5_EN_TLS
-				if (wi->resync_dump_frag) {
-					struct mlx5e_sq_dma *dma =
-						mlx5e_dma_get(sq, dma_fifo_cc++);
-
-					mlx5e_ktls_tx_handle_resync_dump_comp(sq, wi, dma);
-				}
-#endif
+				mlx5e_ktls_tx_handle_resync_dump_comp(sq, wi, &dma_fifo_cc);
 				sqcc += wi->num_wqebbs;
 				continue;
 			}
@@ -542,29 +538,38 @@ void mlx5e_free_txqsq_descs(struct mlx5e_txqsq *sq)
 {
 	struct mlx5e_tx_wqe_info *wi;
 	struct sk_buff *skb;
+	u32 dma_fifo_cc;
+	u16 sqcc;
 	u16 ci;
 	int i;
 
-	while (sq->cc != sq->pc) {
-		ci = mlx5_wq_cyc_ctr2ix(&sq->wq, sq->cc);
+	sqcc = sq->cc;
+	dma_fifo_cc = sq->dma_fifo_cc;
+
+	while (sqcc != sq->pc) {
+		ci = mlx5_wq_cyc_ctr2ix(&sq->wq, sqcc);
 		wi = &sq->db.wqe_info[ci];
 		skb = wi->skb;
 
-		if (!skb) { /* nop */
-			sq->cc++;
+		if (!skb) {
+			mlx5e_ktls_tx_handle_resync_dump_comp(sq, wi, &dma_fifo_cc);
+			sqcc += wi->num_wqebbs;
 			continue;
 		}
 
 		for (i = 0; i < wi->num_dma; i++) {
 			struct mlx5e_sq_dma *dma =
-				mlx5e_dma_get(sq, sq->dma_fifo_cc++);
+				mlx5e_dma_get(sq, dma_fifo_cc++);
 
 			mlx5e_tx_dma_unmap(sq->pdev, dma);
 		}
 
 		dev_kfree_skb_any(skb);
-		sq->cc += wi->num_wqebbs;
+		sqcc += wi->num_wqebbs;
 	}
+
+	sq->dma_fifo_cc = dma_fifo_cc;
+	sq->cc = sqcc;
 }
 
 #ifdef CONFIG_MLX5_CORE_IPOIB
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/conn.c b/drivers/net/ethernet/mellanox/mlx5/core/fpga/conn.c
index 4c50efe4e7f1..61021133029e 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/fpga/conn.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/conn.c
@@ -464,8 +464,10 @@ static int mlx5_fpga_conn_create_cq(struct mlx5_fpga_conn *conn, int cq_size)
 	}
 
 	err = mlx5_vector2eqn(mdev, smp_processor_id(), &eqn, &irqn);
-	if (err)
+	if (err) {
+		kvfree(in);
 		goto err_cqwq;
+	}
 
 	cqc = MLX5_ADDR_OF(create_cq_in, in, cq_context);
 	MLX5_SET(cqc, cqc, log_cq_size, ilog2(cq_size));
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/health.c b/drivers/net/ethernet/mellanox/mlx5/core/health.c
index d685122d9ff7..c07f3154437c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/health.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/health.c
@@ -572,7 +572,7 @@ mlx5_fw_fatal_reporter_dump(struct devlink_health_reporter *reporter,
 		return -ENOMEM;
 	err = mlx5_crdump_collect(dev, cr_data);
 	if (err)
-		return err;
+		goto free_data;
 
 	if (priv_ctx) {
 		struct mlx5_fw_reporter_ctx *fw_reporter_ctx = priv_ctx;