Merge branch '200GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue

Tony Nguyen says:

====================
idpf: replace Tx flow scheduling buffer ring with buffer pool

Joshua Hay says:

This series fixes a stability issue in the flow scheduling Tx send/clean
path that results in a Tx timeout.

The existing guardrails in the Tx path were not sufficient to prevent
the driver from reusing completion tags that were still in flight (held
by the HW).  This collision would cause the driver to erroneously clean
the wrong packet thus leaving the descriptor ring in a bad state.

The main point of this fix is to replace the flow scheduling buffer ring
with a large pool/array of buffers.  The completion tag then simply is
the index into this array.  The driver tracks the free tags and pulls
the next free one from a refillq.  The cleaning routines simply use the
completion tag from the completion descriptor to index into the array to
quickly find the buffers to clean.

All of the code to support this is added first to ensure traffic still
passes with each patch.  The final patch then removes all of the
obsolete stashing code.

* '200GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
  idpf: remove obsolete stashing code
  idpf: stop Tx if there are insufficient buffer resources
  idpf: replace flow scheduling buffer ring with buffer pool
  idpf: simplify and fix splitq Tx packet rollback error path
  idpf: improve when to set RE bit logic
  idpf: add support for Tx refillqs in flow scheduling mode
====================

Link: https://patch.msgid.link/20250821180100.401955-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

3 files changed, 356 insertions, 515 deletions

diff --git a/drivers/net/ethernet/intel/idpf/idpf_singleq_txrx.c b/drivers/net/ethernet/intel/idpf/idpf_singleq_txrx.c
index 555879b1248d..b19b462e0bb6 100644
--- a/drivers/net/ethernet/intel/idpf/idpf_singleq_txrx.c
+++ b/drivers/net/ethernet/intel/idpf/idpf_singleq_txrx.c
@@ -180,6 +180,58 @@ static int idpf_tx_singleq_csum(struct sk_buff *skb,
 }
 
 /**
+ * idpf_tx_singleq_dma_map_error - handle TX DMA map errors
+ * @txq: queue to send buffer on
+ * @skb: send buffer
+ * @first: original first buffer info buffer for packet
+ * @idx: starting point on ring to unwind
+ */
+static void idpf_tx_singleq_dma_map_error(struct idpf_tx_queue *txq,
+					  struct sk_buff *skb,
+					  struct idpf_tx_buf *first, u16 idx)
+{
+	struct libeth_sq_napi_stats ss = { };
+	struct libeth_cq_pp cp = {
+		.dev	= txq->dev,
+		.ss	= &ss,
+	};
+
+	u64_stats_update_begin(&txq->stats_sync);
+	u64_stats_inc(&txq->q_stats.dma_map_errs);
+	u64_stats_update_end(&txq->stats_sync);
+
+	/* clear dma mappings for failed tx_buf map */
+	for (;;) {
+		struct idpf_tx_buf *tx_buf;
+
+		tx_buf = &txq->tx_buf[idx];
+		libeth_tx_complete(tx_buf, &cp);
+		if (tx_buf == first)
+			break;
+		if (idx == 0)
+			idx = txq->desc_count;
+		idx--;
+	}
+
+	if (skb_is_gso(skb)) {
+		union idpf_tx_flex_desc *tx_desc;
+
+		/* If we failed a DMA mapping for a TSO packet, we will have
+		 * used one additional descriptor for a context
+		 * descriptor. Reset that here.
+		 */
+		tx_desc = &txq->flex_tx[idx];
+		memset(tx_desc, 0, sizeof(*tx_desc));
+		if (idx == 0)
+			idx = txq->desc_count;
+		idx--;
+	}
+
+	/* Update tail in case netdev_xmit_more was previously true */
+	idpf_tx_buf_hw_update(txq, idx, false);
+}
+
+/**
  * idpf_tx_singleq_map - Build the Tx base descriptor
  * @tx_q: queue to send buffer on
  * @first: first buffer info buffer to use
@@ -219,8 +271,9 @@ static void idpf_tx_singleq_map(struct idpf_tx_queue *tx_q,
 	for (frag = &skb_shinfo(skb)->frags[0];; frag++) {
 		unsigned int max_data = IDPF_TX_MAX_DESC_DATA_ALIGNED;
 
-		if (dma_mapping_error(tx_q->dev, dma))
-			return idpf_tx_dma_map_error(tx_q, skb, first, i);
+		if (unlikely(dma_mapping_error(tx_q->dev, dma)))
+			return idpf_tx_singleq_dma_map_error(tx_q, skb,
+							     first, i);
 
 		/* record length, and DMA address */
 		dma_unmap_len_set(tx_buf, len, size);
@@ -362,11 +415,11 @@ netdev_tx_t idpf_tx_singleq_frame(struct sk_buff *skb,
 {
 	struct idpf_tx_offload_params offload = { };
 	struct idpf_tx_buf *first;
+	u32 count, buf_count = 1;
 	int csum, tso, needed;
-	unsigned int count;
 	__be16 protocol;
 
-	count = idpf_tx_desc_count_required(tx_q, skb);
+	count = idpf_tx_res_count_required(tx_q, skb, &buf_count);
 	if (unlikely(!count))
 		return idpf_tx_drop_skb(tx_q, skb);
 
diff --git a/drivers/net/ethernet/intel/idpf/idpf_txrx.c b/drivers/net/ethernet/intel/idpf/idpf_txrx.c
index 66a1b040639d..eaad52a83b04 100644
--- a/drivers/net/ethernet/intel/idpf/idpf_txrx.c
+++ b/drivers/net/ethernet/intel/idpf/idpf_txrx.c
@@ -8,48 +8,13 @@
 #include "idpf_ptp.h"
 #include "idpf_virtchnl.h"
 
-struct idpf_tx_stash {
-	struct hlist_node hlist;
-	struct libeth_sqe buf;
-};
-
-#define idpf_tx_buf_compl_tag(buf)	(*(u32 *)&(buf)->priv)
+#define idpf_tx_buf_next(buf)		(*(u32 *)&(buf)->priv)
 LIBETH_SQE_CHECK_PRIV(u32);
 
 static bool idpf_chk_linearize(struct sk_buff *skb, unsigned int max_bufs,
 			       unsigned int count);
 
 /**
- * idpf_buf_lifo_push - push a buffer pointer onto stack
- * @stack: pointer to stack struct
- * @buf: pointer to buf to push
- *
- * Returns 0 on success, negative on failure
- **/
-static int idpf_buf_lifo_push(struct idpf_buf_lifo *stack,
-			      struct idpf_tx_stash *buf)
-{
-	if (unlikely(stack->top == stack->size))
-		return -ENOSPC;
-
-	stack->bufs[stack->top++] = buf;
-
-	return 0;
-}
-
-/**
- * idpf_buf_lifo_pop - pop a buffer pointer from stack
- * @stack: pointer to stack struct
- **/
-static struct idpf_tx_stash *idpf_buf_lifo_pop(struct idpf_buf_lifo *stack)
-{
-	if (unlikely(!stack->top))
-		return NULL;
-
-	return stack->bufs[--stack->top];
-}
-
-/**
  * idpf_tx_timeout - Respond to a Tx Hang
  * @netdev: network interface device structure
  * @txqueue: TX queue
@@ -77,52 +42,22 @@ void idpf_tx_timeout(struct net_device *netdev, unsigned int txqueue)
 static void idpf_tx_buf_rel_all(struct idpf_tx_queue *txq)
 {
 	struct libeth_sq_napi_stats ss = { };
-	struct idpf_buf_lifo *buf_stack;
-	struct idpf_tx_stash *stash;
 	struct libeth_cq_pp cp = {
 		.dev	= txq->dev,
 		.ss	= &ss,
 	};
-	struct hlist_node *tmp;
-	u32 i, tag;
+	u32 i;
 
 	/* Buffers already cleared, nothing to do */
 	if (!txq->tx_buf)
 		return;
 
 	/* Free all the Tx buffer sk_buffs */
-	for (i = 0; i < txq->desc_count; i++)
+	for (i = 0; i < txq->buf_pool_size; i++)
 		libeth_tx_complete(&txq->tx_buf[i], &cp);
 
 	kfree(txq->tx_buf);
 	txq->tx_buf = NULL;
-
-	if (!idpf_queue_has(FLOW_SCH_EN, txq))
-		return;
-
-	buf_stack = &txq->stash->buf_stack;
-	if (!buf_stack->bufs)
-		return;
-
-	/*
-	 * If a Tx timeout occurred, there are potentially still bufs in the
-	 * hash table, free them here.
-	 */
-	hash_for_each_safe(txq->stash->sched_buf_hash, tag, tmp, stash,
-			   hlist) {
-		if (!stash)
-			continue;
-
-		libeth_tx_complete(&stash->buf, &cp);
-		hash_del(&stash->hlist);
-		idpf_buf_lifo_push(buf_stack, stash);
-	}
-
-	for (i = 0; i < buf_stack->size; i++)
-		kfree(buf_stack->bufs[i]);
-
-	kfree(buf_stack->bufs);
-	buf_stack->bufs = NULL;
 }
 
 /**
@@ -139,6 +74,9 @@ static void idpf_tx_desc_rel(struct idpf_tx_queue *txq)
 	if (!txq->desc_ring)
 		return;
 
+	if (txq->refillq)
+		kfree(txq->refillq->ring);
+
 	dmam_free_coherent(txq->dev, txq->size, txq->desc_ring, txq->dma);
 	txq->desc_ring = NULL;
 	txq->next_to_use = 0;
@@ -195,41 +133,18 @@ static void idpf_tx_desc_rel_all(struct idpf_vport *vport)
  */
 static int idpf_tx_buf_alloc_all(struct idpf_tx_queue *tx_q)
 {
-	struct idpf_buf_lifo *buf_stack;
-	int buf_size;
-	int i;
-
 	/* Allocate book keeping buffers only. Buffers to be supplied to HW
 	 * are allocated by kernel network stack and received as part of skb
 	 */
-	buf_size = sizeof(struct idpf_tx_buf) * tx_q->desc_count;
-	tx_q->tx_buf = kzalloc(buf_size, GFP_KERNEL);
+	if (idpf_queue_has(FLOW_SCH_EN, tx_q))
+		tx_q->buf_pool_size = U16_MAX;
+	else
+		tx_q->buf_pool_size = tx_q->desc_count;
+	tx_q->tx_buf = kcalloc(tx_q->buf_pool_size, sizeof(*tx_q->tx_buf),
+			       GFP_KERNEL);
 	if (!tx_q->tx_buf)
 		return -ENOMEM;
 
-	if (!idpf_queue_has(FLOW_SCH_EN, tx_q))
-		return 0;
-
-	buf_stack = &tx_q->stash->buf_stack;
-
-	/* Initialize tx buf stack for out-of-order completions if
-	 * flow scheduling offload is enabled
-	 */
-	buf_stack->bufs = kcalloc(tx_q->desc_count, sizeof(*buf_stack->bufs),
-				  GFP_KERNEL);
-	if (!buf_stack->bufs)
-		return -ENOMEM;
-
-	buf_stack->size = tx_q->desc_count;
-	buf_stack->top = tx_q->desc_count;
-
-	for (i = 0; i < tx_q->desc_count; i++) {
-		buf_stack->bufs[i] = kzalloc(sizeof(*buf_stack->bufs[i]),
-					     GFP_KERNEL);
-		if (!buf_stack->bufs[i])
-			return -ENOMEM;
-	}
-
 	return 0;
 }
 
@@ -244,6 +159,7 @@ static int idpf_tx_desc_alloc(const struct idpf_vport *vport,
 			      struct idpf_tx_queue *tx_q)
 {
 	struct device *dev = tx_q->dev;
+	struct idpf_sw_queue *refillq;
 	int err;
 
 	err = idpf_tx_buf_alloc_all(tx_q);
@@ -267,6 +183,31 @@ static int idpf_tx_desc_alloc(const struct idpf_vport *vport,
 	tx_q->next_to_clean = 0;
 	idpf_queue_set(GEN_CHK, tx_q);
 
+	if (!idpf_queue_has(FLOW_SCH_EN, tx_q))
+		return 0;
+
+	refillq = tx_q->refillq;
+	refillq->desc_count = tx_q->buf_pool_size;
+	refillq->ring = kcalloc(refillq->desc_count, sizeof(u32),
+				GFP_KERNEL);
+	if (!refillq->ring) {
+		err = -ENOMEM;
+		goto err_alloc;
+	}
+
+	for (unsigned int i = 0; i < refillq->desc_count; i++)
+		refillq->ring[i] =
+			FIELD_PREP(IDPF_RFL_BI_BUFID_M, i) |
+			FIELD_PREP(IDPF_RFL_BI_GEN_M,
+				   idpf_queue_has(GEN_CHK, refillq));
+
+	/* Go ahead and flip the GEN bit since this counts as filling
+	 * up the ring, i.e. we already ring wrapped.
+	 */
+	idpf_queue_change(GEN_CHK, refillq);
+
+	tx_q->last_re = tx_q->desc_count - IDPF_TX_SPLITQ_RE_MIN_GAP;
+
 	return 0;
 
 err_alloc:
@@ -317,8 +258,6 @@ static int idpf_tx_desc_alloc_all(struct idpf_vport *vport)
 	for (i = 0; i < vport->num_txq_grp; i++) {
 		for (j = 0; j < vport->txq_grps[i].num_txq; j++) {
 			struct idpf_tx_queue *txq = vport->txq_grps[i].txqs[j];
-			u8 gen_bits = 0;
-			u16 bufidx_mask;
 
 			err = idpf_tx_desc_alloc(vport, txq);
 			if (err) {
@@ -327,34 +266,6 @@ static int idpf_tx_desc_alloc_all(struct idpf_vport *vport)
 					i);
 				goto err_out;
 			}
-
-			if (!idpf_is_queue_model_split(vport->txq_model))
-				continue;
-
-			txq->compl_tag_cur_gen = 0;
-
-			/* Determine the number of bits in the bufid
-			 * mask and add one to get the start of the
-			 * generation bits
-			 */
-			bufidx_mask = txq->desc_count - 1;
-			while (bufidx_mask >> 1) {
-				txq->compl_tag_gen_s++;
-				bufidx_mask = bufidx_mask >> 1;
-			}
-			txq->compl_tag_gen_s++;
-
-			gen_bits = IDPF_TX_SPLITQ_COMPL_TAG_WIDTH -
-							txq->compl_tag_gen_s;
-			txq->compl_tag_gen_max = GETMAXVAL(gen_bits);
-
-			/* Set bufid mask based on location of first
-			 * gen bit; it cannot simply be the descriptor
-			 * ring size-1 since we can have size values
-			 * where not all of those bits are set.
-			 */
-			txq->compl_tag_bufid_m =
-				GETMAXVAL(txq->compl_tag_gen_s);
 		}
 
 		if (!idpf_is_queue_model_split(vport->txq_model))
@@ -603,18 +514,18 @@ static int idpf_rx_hdr_buf_alloc_all(struct idpf_buf_queue *bufq)
 }
 
 /**
- * idpf_rx_post_buf_refill - Post buffer id to refill queue
+ * idpf_post_buf_refill - Post buffer id to refill queue
  * @refillq: refill queue to post to
  * @buf_id: buffer id to post
  */
-static void idpf_rx_post_buf_refill(struct idpf_sw_queue *refillq, u16 buf_id)
+static void idpf_post_buf_refill(struct idpf_sw_queue *refillq, u16 buf_id)
 {
 	u32 nta = refillq->next_to_use;
 
 	/* store the buffer ID and the SW maintained GEN bit to the refillq */
 	refillq->ring[nta] =
-		FIELD_PREP(IDPF_RX_BI_BUFID_M, buf_id) |
-		FIELD_PREP(IDPF_RX_BI_GEN_M,
+		FIELD_PREP(IDPF_RFL_BI_BUFID_M, buf_id) |
+		FIELD_PREP(IDPF_RFL_BI_GEN_M,
 			   idpf_queue_has(GEN_CHK, refillq));
 
 	if (unlikely(++nta == refillq->desc_count)) {
@@ -995,6 +906,11 @@ static void idpf_txq_group_rel(struct idpf_vport *vport)
 		struct idpf_txq_group *txq_grp = &vport->txq_grps[i];
 
 		for (j = 0; j < txq_grp->num_txq; j++) {
+			if (flow_sch_en) {
+				kfree(txq_grp->txqs[j]->refillq);
+				txq_grp->txqs[j]->refillq = NULL;
+			}
+
 			kfree(txq_grp->txqs[j]);
 			txq_grp->txqs[j] = NULL;
 		}
@@ -1004,9 +920,6 @@ static void idpf_txq_group_rel(struct idpf_vport *vport)
 
 		kfree(txq_grp->complq);
 		txq_grp->complq = NULL;
-
-		if (flow_sch_en)
-			kfree(txq_grp->stashes);
 	}
 	kfree(vport->txq_grps);
 	vport->txq_grps = NULL;
@@ -1367,7 +1280,6 @@ static int idpf_txq_group_alloc(struct idpf_vport *vport, u16 num_txq)
 	for (i = 0; i < vport->num_txq_grp; i++) {
 		struct idpf_txq_group *tx_qgrp = &vport->txq_grps[i];
 		struct idpf_adapter *adapter = vport->adapter;
-		struct idpf_txq_stash *stashes;
 		int j;
 
 		tx_qgrp->vport = vport;
@@ -1380,15 +1292,6 @@ static int idpf_txq_group_alloc(struct idpf_vport *vport, u16 num_txq)
 				goto err_alloc;
 		}
 
-		if (split && flow_sch_en) {
-			stashes = kcalloc(num_txq, sizeof(*stashes),
-					  GFP_KERNEL);
-			if (!stashes)
-				goto err_alloc;
-
-			tx_qgrp->stashes = stashes;
-		}
-
 		for (j = 0; j < tx_qgrp->num_txq; j++) {
 			struct idpf_tx_queue *q = tx_qgrp->txqs[j];
 
@@ -1408,12 +1311,14 @@ static int idpf_txq_group_alloc(struct idpf_vport *vport, u16 num_txq)
 			if (!flow_sch_en)
 				continue;
 
-			if (split) {
-				q->stash = &stashes[j];
-				hash_init(q->stash->sched_buf_hash);
-			}
-
 			idpf_queue_set(FLOW_SCH_EN, q);
+
+			q->refillq = kzalloc(sizeof(*q->refillq), GFP_KERNEL);
+			if (!q->refillq)
+				goto err_alloc;
+
+			idpf_queue_set(GEN_CHK, q->refillq);
+			idpf_queue_set(RFL_GEN_CHK, q->refillq);
 		}
 
 		if (!split)
@@ -1697,87 +1602,6 @@ static void idpf_tx_read_tstamp(struct idpf_tx_queue *txq, struct sk_buff *skb)
 	spin_unlock_bh(&tx_tstamp_caps->status_lock);
 }
 
-/**
- * idpf_tx_clean_stashed_bufs - clean bufs that were stored for
- * out of order completions
- * @txq: queue to clean
- * @compl_tag: completion tag of packet to clean (from completion descriptor)
- * @cleaned: pointer to stats struct to track cleaned packets/bytes
- * @budget: Used to determine if we are in netpoll
- */
-static void idpf_tx_clean_stashed_bufs(struct idpf_tx_queue *txq,
-				       u16 compl_tag,
-				       struct libeth_sq_napi_stats *cleaned,
-				       int budget)
-{
-	struct idpf_tx_stash *stash;
-	struct hlist_node *tmp_buf;
-	struct libeth_cq_pp cp = {
-		.dev	= txq->dev,
-		.ss	= cleaned,
-		.napi	= budget,
-	};
-
-	/* Buffer completion */
-	hash_for_each_possible_safe(txq->stash->sched_buf_hash, stash, tmp_buf,
-				    hlist, compl_tag) {
-		if (unlikely(idpf_tx_buf_compl_tag(&stash->buf) != compl_tag))
-			continue;
-
-		hash_del(&stash->hlist);
-
-		if (stash->buf.type == LIBETH_SQE_SKB &&
-		    (skb_shinfo(stash->buf.skb)->tx_flags & SKBTX_IN_PROGRESS))
-			idpf_tx_read_tstamp(txq, stash->buf.skb);
-
-		libeth_tx_complete(&stash->buf, &cp);
-
-		/* Push shadow buf back onto stack */
-		idpf_buf_lifo_push(&txq->stash->buf_stack, stash);
-	}
-}
-
-/**
- * idpf_stash_flow_sch_buffers - store buffer parameters info to be freed at a
- * later time (only relevant for flow scheduling mode)
- * @txq: Tx queue to clean
- * @tx_buf: buffer to store
- */
-static int idpf_stash_flow_sch_buffers(struct idpf_tx_queue *txq,
-				       struct idpf_tx_buf *tx_buf)
-{
-	struct idpf_tx_stash *stash;
-
-	if (unlikely(tx_buf->type <= LIBETH_SQE_CTX))
-		return 0;
-
-	stash = idpf_buf_lifo_pop(&txq->stash->buf_stack);
-	if (unlikely(!stash)) {
-		net_err_ratelimited("%s: No out-of-order TX buffers left!\n",
-				    netdev_name(txq->netdev));
-
-		return -ENOMEM;
-	}
-
-	/* Store buffer params in shadow buffer */
-	stash->buf.skb = tx_buf->skb;
-	stash->buf.bytes = tx_buf->bytes;
-	stash->buf.packets = tx_buf->packets;
-	stash->buf.type = tx_buf->type;
-	stash->buf.nr_frags = tx_buf->nr_frags;
-	dma_unmap_addr_set(&stash->buf, dma, dma_unmap_addr(tx_buf, dma));
-	dma_unmap_len_set(&stash->buf, len, dma_unmap_len(tx_buf, len));
-	idpf_tx_buf_compl_tag(&stash->buf) = idpf_tx_buf_compl_tag(tx_buf);
-
-	/* Add buffer to buf_hash table to be freed later */
-	hash_add(txq->stash->sched_buf_hash, &stash->hlist,
-		 idpf_tx_buf_compl_tag(&stash->buf));
-
-	tx_buf->type = LIBETH_SQE_EMPTY;
-
-	return 0;
-}
-
 #define idpf_tx_splitq_clean_bump_ntc(txq, ntc, desc, buf)	\
 do {								\
 	if (unlikely(++(ntc) == (txq)->desc_count)) {		\
@@ -1805,14 +1629,8 @@ do {								\
  * Separate packet completion events will be reported on the completion queue,
  * and the buffers will be cleaned separately. The stats are not updated from
  * this function when using flow-based scheduling.
- *
- * Furthermore, in flow scheduling mode, check to make sure there are enough
- * reserve buffers to stash the packet. If there are not, return early, which
- * will leave next_to_clean pointing to the packet that failed to be stashed.
- *
- * Return: false in the scenario above, true otherwise.
  */
-static bool idpf_tx_splitq_clean(struct idpf_tx_queue *tx_q, u16 end,
+static void idpf_tx_splitq_clean(struct idpf_tx_queue *tx_q, u16 end,
 				 int napi_budget,
 				 struct libeth_sq_napi_stats *cleaned,
 				 bool descs_only)
@@ -1826,7 +1644,12 @@ static bool idpf_tx_splitq_clean(struct idpf_tx_queue *tx_q, u16 end,
 		.napi	= napi_budget,
 	};
 	struct idpf_tx_buf *tx_buf;
-	bool clean_complete = true;
+
+	if (descs_only) {
+		/* Bump ring index to mark as cleaned. */
+		tx_q->next_to_clean = end;
+		return;
+	}
 
 	tx_desc = &tx_q->flex_tx[ntc];
 	next_pending_desc = &tx_q->flex_tx[end];
@@ -1846,136 +1669,61 @@ static bool idpf_tx_splitq_clean(struct idpf_tx_queue *tx_q, u16 end,
 			break;
 
 		eop_idx = tx_buf->rs_idx;
+		libeth_tx_complete(tx_buf, &cp);
 
-		if (descs_only) {
-			if (IDPF_TX_BUF_RSV_UNUSED(tx_q) < tx_buf->nr_frags) {
-				clean_complete = false;
-				goto tx_splitq_clean_out;
-			}
-
-			idpf_stash_flow_sch_buffers(tx_q, tx_buf);
+		/* unmap remaining buffers */
+		while (ntc != eop_idx) {
+			idpf_tx_splitq_clean_bump_ntc(tx_q, ntc,
+						      tx_desc, tx_buf);
 
-			while (ntc != eop_idx) {
-				idpf_tx_splitq_clean_bump_ntc(tx_q, ntc,
-							      tx_desc, tx_buf);
-				idpf_stash_flow_sch_buffers(tx_q, tx_buf);
-			}
-		} else {
+			/* unmap any remaining paged data */
 			libeth_tx_complete(tx_buf, &cp);
-
-			/* unmap remaining buffers */
-			while (ntc != eop_idx) {
-				idpf_tx_splitq_clean_bump_ntc(tx_q, ntc,
-							      tx_desc, tx_buf);
-
-				/* unmap any remaining paged data */
-				libeth_tx_complete(tx_buf, &cp);
-			}
 		}
 
 fetch_next_txq_desc:
 		idpf_tx_splitq_clean_bump_ntc(tx_q, ntc, tx_desc, tx_buf);
 	}
 
-tx_splitq_clean_out:
 	tx_q->next_to_clean = ntc;
-
-	return clean_complete;
 }
 
-#define idpf_tx_clean_buf_ring_bump_ntc(txq, ntc, buf)	\
-do {							\
-	(buf)++;					\
-	(ntc)++;					\
-	if (unlikely((ntc) == (txq)->desc_count)) {	\
-		buf = (txq)->tx_buf;			\
-		ntc = 0;				\
-	}						\
-} while (0)
-
 /**
- * idpf_tx_clean_buf_ring - clean flow scheduling TX queue buffers
+ * idpf_tx_clean_bufs - clean flow scheduling TX queue buffers
  * @txq: queue to clean
- * @compl_tag: completion tag of packet to clean (from completion descriptor)
+ * @buf_id: packet's starting buffer ID, from completion descriptor
  * @cleaned: pointer to stats struct to track cleaned packets/bytes
  * @budget: Used to determine if we are in netpoll
  *
- * Cleans all buffers associated with the input completion tag either from the
- * TX buffer ring or from the hash table if the buffers were previously
- * stashed. Returns the byte/segment count for the cleaned packet associated
- * this completion tag.
+ * Clean all buffers associated with the packet starting at buf_id. Returns the
+ * byte/segment count for the cleaned packet.
  */
-static bool idpf_tx_clean_buf_ring(struct idpf_tx_queue *txq, u16 compl_tag,
-				   struct libeth_sq_napi_stats *cleaned,
-				   int budget)
+static void idpf_tx_clean_bufs(struct idpf_tx_queue *txq, u32 buf_id,
+			       struct libeth_sq_napi_stats *cleaned,
+			       int budget)
 {
-	u16 idx = compl_tag & txq->compl_tag_bufid_m;
 	struct idpf_tx_buf *tx_buf = NULL;
 	struct libeth_cq_pp cp = {
 		.dev	= txq->dev,
 		.ss	= cleaned,
 		.napi	= budget,
 	};
-	u16 ntc, orig_idx = idx;
-
-	tx_buf = &txq->tx_buf[idx];
-
-	if (unlikely(tx_buf->type <= LIBETH_SQE_CTX ||
-		     idpf_tx_buf_compl_tag(tx_buf) != compl_tag))
-		return false;
 
+	tx_buf = &txq->tx_buf[buf_id];
 	if (tx_buf->type == LIBETH_SQE_SKB) {
 		if (skb_shinfo(tx_buf->skb)->tx_flags & SKBTX_IN_PROGRESS)
 			idpf_tx_read_tstamp(txq, tx_buf->skb);
 
 		libeth_tx_complete(tx_buf, &cp);
+		idpf_post_buf_refill(txq->refillq, buf_id);
 	}
 
-	idpf_tx_clean_buf_ring_bump_ntc(txq, idx, tx_buf);
+	while (idpf_tx_buf_next(tx_buf) != IDPF_TXBUF_NULL) {
+		buf_id = idpf_tx_buf_next(tx_buf);
 
-	while (idpf_tx_buf_compl_tag(tx_buf) == compl_tag) {
+		tx_buf = &txq->tx_buf[buf_id];
 		libeth_tx_complete(tx_buf, &cp);
-		idpf_tx_clean_buf_ring_bump_ntc(txq, idx, tx_buf);
+		idpf_post_buf_refill(txq->refillq, buf_id);
 	}
-
-	/*
-	 * It's possible the packet we just cleaned was an out of order
-	 * completion, which means we can stash the buffers starting from
-	 * the original next_to_clean and reuse the descriptors. We need
-	 * to compare the descriptor ring next_to_clean packet's "first" buffer
-	 * to the "first" buffer of the packet we just cleaned to determine if
-	 * this is the case. Howevever, next_to_clean can point to either a
-	 * reserved buffer that corresponds to a context descriptor used for the
-	 * next_to_clean packet (TSO packet) or the "first" buffer (single
-	 * packet). The orig_idx from the packet we just cleaned will always
-	 * point to the "first" buffer. If next_to_clean points to a reserved
-	 * buffer, let's bump ntc once and start the comparison from there.
-	 */
-	ntc = txq->next_to_clean;
-	tx_buf = &txq->tx_buf[ntc];
-
-	if (tx_buf->type == LIBETH_SQE_CTX)
-		idpf_tx_clean_buf_ring_bump_ntc(txq, ntc, tx_buf);
-
-	/*
-	 * If ntc still points to a different "first" buffer, clean the
-	 * descriptor ring and stash all of the buffers for later cleaning. If
-	 * we cannot stash all of the buffers, next_to_clean will point to the
-	 * "first" buffer of the packet that could not be stashed and cleaning
-	 * will start there next time.
-	 */
-	if (unlikely(tx_buf != &txq->tx_buf[orig_idx] &&
-		     !idpf_tx_splitq_clean(txq, orig_idx, budget, cleaned,
-					   true)))
-		return true;
-
-	/*
-	 * Otherwise, update next_to_clean to reflect the cleaning that was
-	 * done above.
-	 */
-	txq->next_to_clean = idx;
-
-	return true;
 }
 
 /**
@@ -1994,22 +1742,17 @@ static void idpf_tx_handle_rs_completion(struct idpf_tx_queue *txq,
 					 struct libeth_sq_napi_stats *cleaned,
 					 int budget)
 {
-	u16 compl_tag;
+	/* RS completion contains queue head for queue based scheduling or
+	 * completion tag for flow based scheduling.
+	 */
+	u16 rs_compl_val = le16_to_cpu(desc->q_head_compl_tag.q_head);
 
 	if (!idpf_queue_has(FLOW_SCH_EN, txq)) {
-		u16 head = le16_to_cpu(desc->q_head_compl_tag.q_head);
-
-		idpf_tx_splitq_clean(txq, head, budget, cleaned, false);
+		idpf_tx_splitq_clean(txq, rs_compl_val, budget, cleaned, false);
 		return;
 	}
 
-	compl_tag = le16_to_cpu(desc->q_head_compl_tag.compl_tag);
-
-	/* If we didn't clean anything on the ring, this packet must be
-	 * in the hash table. Go clean it there.
-	 */
-	if (!idpf_tx_clean_buf_ring(txq, compl_tag, cleaned, budget))
-		idpf_tx_clean_stashed_bufs(txq, compl_tag, cleaned, budget);
+	idpf_tx_clean_bufs(txq, rs_compl_val, cleaned, budget);
 }
 
 /**
@@ -2126,8 +1869,7 @@ fetch_next_desc:
 		/* Update BQL */
 		nq = netdev_get_tx_queue(tx_q->netdev, tx_q->idx);
 
-		dont_wake = !complq_ok || IDPF_TX_BUF_RSV_LOW(tx_q) ||
-			    np->state != __IDPF_VPORT_UP ||
+		dont_wake = !complq_ok || np->state != __IDPF_VPORT_UP ||
 			    !netif_carrier_ok(tx_q->netdev);
 		/* Check if the TXQ needs to and can be restarted */
 		__netif_txq_completed_wake(nq, tx_q->cleaned_pkts, tx_q->cleaned_bytes,
@@ -2184,15 +1926,21 @@ void idpf_tx_splitq_build_flow_desc(union idpf_tx_flex_desc *desc,
 	desc->flow.qw1.compl_tag = cpu_to_le16(params->compl_tag);
 }
 
-/* Global conditions to tell whether the txq (and related resources)
- * has room to allow the use of "size" descriptors.
+/**
+ * idpf_tx_splitq_has_room - check if enough Tx splitq resources are available
+ * @tx_q: the queue to be checked
+ * @descs_needed: number of descriptors required for this packet
+ * @bufs_needed: number of Tx buffers required for this packet
+ *
+ * Return: 0 if no room available, 1 otherwise
  */
-static int idpf_txq_has_room(struct idpf_tx_queue *tx_q, u32 size)
+static int idpf_txq_has_room(struct idpf_tx_queue *tx_q, u32 descs_needed,
+			     u32 bufs_needed)
 {
-	if (IDPF_DESC_UNUSED(tx_q) < size ||
+	if (IDPF_DESC_UNUSED(tx_q) < descs_needed ||
 	    IDPF_TX_COMPLQ_PENDING(tx_q->txq_grp) >
 		IDPF_TX_COMPLQ_OVERFLOW_THRESH(tx_q->txq_grp->complq) ||
-	    IDPF_TX_BUF_RSV_LOW(tx_q))
+	    idpf_tx_splitq_get_free_bufs(tx_q->refillq) < bufs_needed)
 		return 0;
 	return 1;
 }
@@ -2201,14 +1949,21 @@ static int idpf_txq_has_room(struct idpf_tx_queue *tx_q, u32 size)
  * idpf_tx_maybe_stop_splitq - 1st level check for Tx splitq stop conditions
  * @tx_q: the queue to be checked
  * @descs_needed: number of descriptors required for this packet
+ * @bufs_needed: number of buffers needed for this packet
  *
- * Returns 0 if stop is not needed
+ * Return: 0 if stop is not needed
  */
 static int idpf_tx_maybe_stop_splitq(struct idpf_tx_queue *tx_q,
-				     unsigned int descs_needed)
+				     u32 descs_needed,
+				     u32 bufs_needed)
 {
+	/* Since we have multiple resources to check for splitq, our
+	 * start,stop_thrs becomes a boolean check instead of a count
+	 * threshold.
+	 */
 	if (netif_subqueue_maybe_stop(tx_q->netdev, tx_q->idx,
-				      idpf_txq_has_room(tx_q, descs_needed),
+				      idpf_txq_has_room(tx_q, descs_needed,
+							bufs_needed),
 				      1, 1))
 		return 0;
 
@@ -2250,14 +2005,16 @@ void idpf_tx_buf_hw_update(struct idpf_tx_queue *tx_q, u32 val,
 }
 
 /**
- * idpf_tx_desc_count_required - calculate number of Tx descriptors needed
+ * idpf_tx_res_count_required - get number of Tx resources needed for this pkt
  * @txq: queue to send buffer on
  * @skb: send buffer
+ * @bufs_needed: (output) number of buffers needed for this skb.
  *
- * Returns number of data descriptors needed for this skb.
+ * Return: number of data descriptors and buffers needed for this skb.
  */
-unsigned int idpf_tx_desc_count_required(struct idpf_tx_queue *txq,
-					 struct sk_buff *skb)
+unsigned int idpf_tx_res_count_required(struct idpf_tx_queue *txq,
+					struct sk_buff *skb,
+					u32 *bufs_needed)
 {
 	const struct skb_shared_info *shinfo;
 	unsigned int count = 0, i;
@@ -2268,6 +2025,7 @@ unsigned int idpf_tx_desc_count_required(struct idpf_tx_queue *txq,
 		return count;
 
 	shinfo = skb_shinfo(skb);
+	*bufs_needed += shinfo->nr_frags;
 	for (i = 0; i < shinfo->nr_frags; i++) {
 		unsigned int size;
 
@@ -2297,71 +2055,89 @@ unsigned int idpf_tx_desc_count_required(struct idpf_tx_queue *txq,
 }
 
 /**
- * idpf_tx_dma_map_error - handle TX DMA map errors
- * @txq: queue to send buffer on
- * @skb: send buffer
- * @first: original first buffer info buffer for packet
- * @idx: starting point on ring to unwind
+ * idpf_tx_splitq_bump_ntu - adjust NTU and generation
+ * @txq: the tx ring to wrap
+ * @ntu: ring index to bump
  */
-void idpf_tx_dma_map_error(struct idpf_tx_queue *txq, struct sk_buff *skb,
-			   struct idpf_tx_buf *first, u16 idx)
+static unsigned int idpf_tx_splitq_bump_ntu(struct idpf_tx_queue *txq, u16 ntu)
 {
-	struct libeth_sq_napi_stats ss = { };
-	struct libeth_cq_pp cp = {
-		.dev	= txq->dev,
-		.ss	= &ss,
-	};
+	ntu++;
 
-	u64_stats_update_begin(&txq->stats_sync);
-	u64_stats_inc(&txq->q_stats.dma_map_errs);
-	u64_stats_update_end(&txq->stats_sync);
+	if (ntu == txq->desc_count)
+		ntu = 0;
 
-	/* clear dma mappings for failed tx_buf map */
-	for (;;) {
-		struct idpf_tx_buf *tx_buf;
+	return ntu;
+}
 
-		tx_buf = &txq->tx_buf[idx];
-		libeth_tx_complete(tx_buf, &cp);
-		if (tx_buf == first)
-			break;
-		if (idx == 0)
-			idx = txq->desc_count;
-		idx--;
-	}
+/**
+ * idpf_tx_get_free_buf_id - get a free buffer ID from the refill queue
+ * @refillq: refill queue to get buffer ID from
+ * @buf_id: return buffer ID
+ *
+ * Return: true if a buffer ID was found, false if not
+ */
+static bool idpf_tx_get_free_buf_id(struct idpf_sw_queue *refillq,
+				    u32 *buf_id)
+{
+	u32 ntc = refillq->next_to_clean;
+	u32 refill_desc;
 
-	if (skb_is_gso(skb)) {
-		union idpf_tx_flex_desc *tx_desc;
+	refill_desc = refillq->ring[ntc];
 
-		/* If we failed a DMA mapping for a TSO packet, we will have
-		 * used one additional descriptor for a context
-		 * descriptor. Reset that here.
-		 */
-		tx_desc = &txq->flex_tx[idx];
-		memset(tx_desc, 0, sizeof(*tx_desc));
-		if (idx == 0)
-			idx = txq->desc_count;
-		idx--;
+	if (unlikely(idpf_queue_has(RFL_GEN_CHK, refillq) !=
+		     !!(refill_desc & IDPF_RFL_BI_GEN_M)))
+		return false;
+
+	*buf_id = FIELD_GET(IDPF_RFL_BI_BUFID_M, refill_desc);
+
+	if (unlikely(++ntc == refillq->desc_count)) {
+		idpf_queue_change(RFL_GEN_CHK, refillq);
+		ntc = 0;
 	}
 
-	/* Update tail in case netdev_xmit_more was previously true */
-	idpf_tx_buf_hw_update(txq, idx, false);
+	refillq->next_to_clean = ntc;
+
+	return true;
 }
 
 /**
- * idpf_tx_splitq_bump_ntu - adjust NTU and generation
- * @txq: the tx ring to wrap
- * @ntu: ring index to bump
+ * idpf_tx_splitq_pkt_err_unmap - Unmap buffers and bump tail in case of error
+ * @txq: Tx queue to unwind
+ * @params: pointer to splitq params struct
+ * @first: starting buffer for packet to unmap
  */
-static unsigned int idpf_tx_splitq_bump_ntu(struct idpf_tx_queue *txq, u16 ntu)
+static void idpf_tx_splitq_pkt_err_unmap(struct idpf_tx_queue *txq,
+					 struct idpf_tx_splitq_params *params,
+					 struct idpf_tx_buf *first)
 {
-	ntu++;
+	struct idpf_sw_queue *refillq = txq->refillq;
+	struct libeth_sq_napi_stats ss = { };
+	struct idpf_tx_buf *tx_buf = first;
+	struct libeth_cq_pp cp = {
+		.dev    = txq->dev,
+		.ss     = &ss,
+	};
 
-	if (ntu == txq->desc_count) {
-		ntu = 0;
-		txq->compl_tag_cur_gen = IDPF_TX_ADJ_COMPL_TAG_GEN(txq);
+	u64_stats_update_begin(&txq->stats_sync);
+	u64_stats_inc(&txq->q_stats.dma_map_errs);
+	u64_stats_update_end(&txq->stats_sync);
+
+	libeth_tx_complete(tx_buf, &cp);
+	while (idpf_tx_buf_next(tx_buf) != IDPF_TXBUF_NULL) {
+		tx_buf = &txq->tx_buf[idpf_tx_buf_next(tx_buf)];
+		libeth_tx_complete(tx_buf, &cp);
 	}
 
-	return ntu;
+	/* Update tail in case netdev_xmit_more was previously true. */
+	idpf_tx_buf_hw_update(txq, params->prev_ntu, false);
+
+	if (!refillq)
+		return;
+
+	/* Restore refillq state to avoid leaking tags. */
+	if (params->prev_refill_gen != idpf_queue_has(RFL_GEN_CHK, refillq))
+		idpf_queue_change(RFL_GEN_CHK, refillq);
+	refillq->next_to_clean = params->prev_refill_ntc;
 }
 
 /**
@@ -2385,6 +2161,7 @@ static void idpf_tx_splitq_map(struct idpf_tx_queue *tx_q,
 	struct netdev_queue *nq;
 	struct sk_buff *skb;
 	skb_frag_t *frag;
+	u32 next_buf_id;
 	u16 td_cmd = 0;
 	dma_addr_t dma;
 
@@ -2402,17 +2179,16 @@ static void idpf_tx_splitq_map(struct idpf_tx_queue *tx_q,
 	tx_buf = first;
 	first->nr_frags = 0;
 
-	params->compl_tag =
-		(tx_q->compl_tag_cur_gen << tx_q->compl_tag_gen_s) | i;
-
 	for (frag = &skb_shinfo(skb)->frags[0];; frag++) {
 		unsigned int max_data = IDPF_TX_MAX_DESC_DATA_ALIGNED;
 
-		if (dma_mapping_error(tx_q->dev, dma))
-			return idpf_tx_dma_map_error(tx_q, skb, first, i);
+		if (unlikely(dma_mapping_error(tx_q->dev, dma))) {
+			idpf_tx_buf_next(tx_buf) = IDPF_TXBUF_NULL;
+			return idpf_tx_splitq_pkt_err_unmap(tx_q, params,
+							    first);
+		}
 
 		first->nr_frags++;
-		idpf_tx_buf_compl_tag(tx_buf) = params->compl_tag;
 		tx_buf->type = LIBETH_SQE_FRAG;
 
 		/* record length, and DMA address */
@@ -2468,29 +2244,12 @@ static void idpf_tx_splitq_map(struct idpf_tx_queue *tx_q,
 						  max_data);
 
 			if (unlikely(++i == tx_q->desc_count)) {
-				tx_buf = tx_q->tx_buf;
 				tx_desc = &tx_q->flex_tx[0];
 				i = 0;
-				tx_q->compl_tag_cur_gen =
-					IDPF_TX_ADJ_COMPL_TAG_GEN(tx_q);
 			} else {
-				tx_buf++;
 				tx_desc++;
 			}
 
-			/* Since this packet has a buffer that is going to span
-			 * multiple descriptors, it's going to leave holes in
-			 * to the TX buffer ring. To ensure these holes do not
-			 * cause issues in the cleaning routines, we will clear
-			 * them of any stale data and assign them the same
-			 * completion tag as the current packet. Then when the
-			 * packet is being cleaned, the cleaning routines will
-			 * simply pass over these holes and finish cleaning the
-			 * rest of the packet.
-			 */
-			tx_buf->type = LIBETH_SQE_EMPTY;
-			idpf_tx_buf_compl_tag(tx_buf) = params->compl_tag;
-
 			/* Adjust the DMA offset and the remaining size of the