1 files changed, 709 insertions, 513 deletions

diff --git a/drivers/net/ethernet/pensando/ionic/ionic_txrx.c b/drivers/net/ethernet/pensando/ionic/ionic_txrx.c
index e452f4242ba0..c03986bf2628 100644
--- a/drivers/net/ethernet/pensando/ionic/ionic_txrx.c
+++ b/drivers/net/ethernet/pensando/ionic/ionic_txrx.c
@@ -10,14 +10,9 @@
 #include "ionic_lif.h"
 #include "ionic_txrx.h"
 
-static void ionic_rx_clean(struct ionic_queue *q, struct ionic_desc_info *desc_info,
-			   struct ionic_cq_info *cq_info, void *cb_arg);
-
 static inline void ionic_txq_post(struct ionic_queue *q, bool ring_dbell,
 				  ionic_desc_cb cb_func, void *cb_arg)
 {
-	DEBUG_STATS_TXQ_POST(q_to_qcq(q), q->head->desc, ring_dbell);
-
 	ionic_q_post(q, ring_dbell, cb_func, cb_arg);
 }
 
@@ -25,8 +20,6 @@ static inline void ionic_rxq_post(struct ionic_queue *q, bool ring_dbell,
 				  ionic_desc_cb cb_func, void *cb_arg)
 {
 	ionic_q_post(q, ring_dbell, cb_func, cb_arg);
-
-	DEBUG_STATS_RX_BUFF_CNT(q_to_qcq(q));
 }
 
 static inline struct netdev_queue *q_to_ndq(struct ionic_queue *q)
@@ -34,72 +27,144 @@ static inline struct netdev_queue *q_to_ndq(struct ionic_queue *q)
 	return netdev_get_tx_queue(q->lif->netdev, q->index);
 }
 
-static struct sk_buff *ionic_rx_skb_alloc(struct ionic_queue *q,
-					  unsigned int len, bool frags)
+static int ionic_rx_page_alloc(struct ionic_queue *q,
+			       struct ionic_buf_info *buf_info)
 {
-	struct ionic_lif *lif = q->lif;
+	struct net_device *netdev = q->lif->netdev;
 	struct ionic_rx_stats *stats;
-	struct net_device *netdev;
-	struct sk_buff *skb;
+	struct device *dev;
+	struct page *page;
 
-	netdev = lif->netdev;
+	dev = q->dev;
 	stats = q_to_rx_stats(q);
 
-	if (frags)
-		skb = napi_get_frags(&q_to_qcq(q)->napi);
-	else
-		skb = netdev_alloc_skb_ip_align(netdev, len);
+	if (unlikely(!buf_info)) {
+		net_err_ratelimited("%s: %s invalid buf_info in alloc\n",
+				    netdev->name, q->name);
+		return -EINVAL;
+	}
 
-	if (unlikely(!skb)) {
-		net_warn_ratelimited("%s: SKB alloc failed on %s!\n",
-				     netdev->name, q->name);
+	page = alloc_pages(IONIC_PAGE_GFP_MASK, 0);
+	if (unlikely(!page)) {
+		net_err_ratelimited("%s: %s page alloc failed\n",
+				    netdev->name, q->name);
 		stats->alloc_err++;
-		return NULL;
+		return -ENOMEM;
 	}
 
-	return skb;
+	buf_info->dma_addr = dma_map_page(dev, page, 0,
+					  IONIC_PAGE_SIZE, DMA_FROM_DEVICE);
+	if (unlikely(dma_mapping_error(dev, buf_info->dma_addr))) {
+		__free_pages(page, 0);
+		net_err_ratelimited("%s: %s dma map failed\n",
+				    netdev->name, q->name);
+		stats->dma_map_err++;
+		return -EIO;
+	}
+
+	buf_info->page = page;
+	buf_info->page_offset = 0;
+
+	return 0;
+}
+
+static void ionic_rx_page_free(struct ionic_queue *q,
+			       struct ionic_buf_info *buf_info)
+{
+	struct net_device *netdev = q->lif->netdev;
+	struct device *dev = q->dev;
+
+	if (unlikely(!buf_info)) {
+		net_err_ratelimited("%s: %s invalid buf_info in free\n",
+				    netdev->name, q->name);
+		return;
+	}
+
+	if (!buf_info->page)
+		return;
+
+	dma_unmap_page(dev, buf_info->dma_addr, IONIC_PAGE_SIZE, DMA_FROM_DEVICE);
+	__free_pages(buf_info->page, 0);
+	buf_info->page = NULL;
+}
+
+static bool ionic_rx_buf_recycle(struct ionic_queue *q,
+				 struct ionic_buf_info *buf_info, u32 used)
+{
+	u32 size;
+
+	/* don't re-use pages allocated in low-mem condition */
+	if (page_is_pfmemalloc(buf_info->page))
+		return false;
+
+	/* don't re-use buffers from non-local numa nodes */
+	if (page_to_nid(buf_info->page) != numa_mem_id())
+		return false;
+
+	size = ALIGN(used, IONIC_PAGE_SPLIT_SZ);
+	buf_info->page_offset += size;
+	if (buf_info->page_offset >= IONIC_PAGE_SIZE)
+		return false;
+
+	get_page(buf_info->page);
+
+	return true;
 }
 
 static struct sk_buff *ionic_rx_frags(struct ionic_queue *q,
 				      struct ionic_desc_info *desc_info,
-				      struct ionic_cq_info *cq_info)
+				      struct ionic_rxq_comp *comp)
 {
-	struct ionic_rxq_comp *comp = cq_info->cq_desc;
-	struct device *dev = q->lif->ionic->dev;
-	struct ionic_page_info *page_info;
+	struct net_device *netdev = q->lif->netdev;
+	struct ionic_buf_info *buf_info;
+	struct ionic_rx_stats *stats;
+	struct device *dev = q->dev;
 	struct sk_buff *skb;
 	unsigned int i;
 	u16 frag_len;
 	u16 len;
 
-	page_info = &desc_info->pages[0];
+	stats = q_to_rx_stats(q);
+
+	buf_info = &desc_info->bufs[0];
 	len = le16_to_cpu(comp->len);
 
-	prefetch(page_address(page_info->page) + NET_IP_ALIGN);
+	prefetchw(buf_info->page);
 
-	skb = ionic_rx_skb_alloc(q, len, true);
-	if (unlikely(!skb))
+	skb = napi_get_frags(&q_to_qcq(q)->napi);
+	if (unlikely(!skb)) {
+		net_warn_ratelimited("%s: SKB alloc failed on %s!\n",
+				     netdev->name, q->name);
+		stats->alloc_err++;
 		return NULL;
+	}
 
 	i = comp->num_sg_elems + 1;
 	do {
-		if (unlikely(!page_info->page)) {
-			struct napi_struct *napi = &q_to_qcq(q)->napi;
-
-			napi->skb = NULL;
+		if (unlikely(!buf_info->page)) {
 			dev_kfree_skb(skb);
 			return NULL;
 		}
 
-		frag_len = min(len, (u16)PAGE_SIZE);
+		frag_len = min_t(u16, len, IONIC_PAGE_SIZE - buf_info->page_offset);
 		len -= frag_len;
 
-		dma_unmap_page(dev, dma_unmap_addr(page_info, dma_addr),
-			       PAGE_SIZE, DMA_FROM_DEVICE);
+		dma_sync_single_for_cpu(dev,
+					buf_info->dma_addr + buf_info->page_offset,
+					frag_len, DMA_FROM_DEVICE);
+
 		skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags,
-				page_info->page, 0, frag_len, PAGE_SIZE);
-		page_info->page = NULL;
-		page_info++;
+				buf_info->page, buf_info->page_offset, frag_len,
+				IONIC_PAGE_SIZE);
+
+		if (!ionic_rx_buf_recycle(q, buf_info, frag_len)) {
+			dma_unmap_page(dev, buf_info->dma_addr,
+				       IONIC_PAGE_SIZE, DMA_FROM_DEVICE);
+			buf_info->page = NULL;
+		}
+
+		buf_info++;
+
 		i--;
 	} while (i > 0);
 
@@ -108,30 +173,37 @@ static struct sk_buff *ionic_rx_frags(struct ionic_queue *q,
 
 static struct sk_buff *ionic_rx_copybreak(struct ionic_queue *q,
 					  struct ionic_desc_info *desc_info,
-					  struct ionic_cq_info *cq_info)
+					  struct ionic_rxq_comp *comp)
 {
-	struct ionic_rxq_comp *comp = cq_info->cq_desc;
-	struct device *dev = q->lif->ionic->dev;
-	struct ionic_page_info *page_info;
+	struct net_device *netdev = q->lif->netdev;
+	struct ionic_buf_info *buf_info;
+	struct ionic_rx_stats *stats;
+	struct device *dev = q->dev;
 	struct sk_buff *skb;
 	u16 len;
 
-	page_info = &desc_info->pages[0];
+	stats = q_to_rx_stats(q);
+
+	buf_info = &desc_info->bufs[0];
 	len = le16_to_cpu(comp->len);
 
-	skb = ionic_rx_skb_alloc(q, len, false);
-	if (unlikely(!skb))
+	skb = napi_alloc_skb(&q_to_qcq(q)->napi, len);
+	if (unlikely(!skb)) {
+		net_warn_ratelimited("%s: SKB alloc failed on %s!\n",
+				     netdev->name, q->name);
+		stats->alloc_err++;
 		return NULL;
+	}
 
-	if (unlikely(!page_info->page)) {
+	if (unlikely(!buf_info->page)) {
 		dev_kfree_skb(skb);
 		return NULL;
 	}
 
-	dma_sync_single_for_cpu(dev, dma_unmap_addr(page_info, dma_addr),
+	dma_sync_single_for_cpu(dev, buf_info->dma_addr + buf_info->page_offset,
 				len, DMA_FROM_DEVICE);
-	skb_copy_to_linear_data(skb, page_address(page_info->page), len);
-	dma_sync_single_for_device(dev, dma_unmap_addr(page_info, dma_addr),
+	skb_copy_to_linear_data(skb, page_address(buf_info->page) + buf_info->page_offset, len);
+	dma_sync_single_for_device(dev, buf_info->dma_addr + buf_info->page_offset,
 				   len, DMA_FROM_DEVICE);
 
 	skb_put(skb, len);
@@ -140,36 +212,33 @@ static struct sk_buff *ionic_rx_copybreak(struct ionic_queue *q,
 	return skb;
 }
 
-static void ionic_rx_clean(struct ionic_queue *q, struct ionic_desc_info *desc_info,
-			   struct ionic_cq_info *cq_info, void *cb_arg)
+static void ionic_rx_clean(struct ionic_queue *q,
+			   struct ionic_desc_info *desc_info,
+			   struct ionic_cq_info *cq_info,
+			   void *cb_arg)
 {
-	struct ionic_rxq_comp *comp = cq_info->cq_desc;
+	struct net_device *netdev = q->lif->netdev;
 	struct ionic_qcq *qcq = q_to_qcq(q);
 	struct ionic_rx_stats *stats;
-	struct net_device *netdev;
+	struct ionic_rxq_comp *comp;
 	struct sk_buff *skb;
 
+	comp = cq_info->cq_desc + qcq->cq.desc_size - sizeof(*comp);
+
 	stats = q_to_rx_stats(q);
-	netdev = q->lif->netdev;
 
 	if (comp->status) {
 		stats->dropped++;
 		return;
 	}
 
-	/* no packet processing while resetting */
-	if (unlikely(test_bit(IONIC_LIF_QUEUE_RESET, q->lif->state))) {
-		stats->dropped++;
-		return;
-	}
-
 	stats->pkts++;
 	stats->bytes += le16_to_cpu(comp->len);
 
 	if (le16_to_cpu(comp->len) <= q->lif->rx_copybreak)
-		skb = ionic_rx_copybreak(q, desc_info, cq_info);
+		skb = ionic_rx_copybreak(q, desc_info, comp);
 	else
-		skb = ionic_rx_frags(q, desc_info, cq_info);
+		skb = ionic_rx_frags(q, desc_info, comp);
 
 	if (unlikely(!skb)) {
 		stats->dropped++;
@@ -195,12 +264,11 @@ static void ionic_rx_clean(struct ionic_queue *q, struct ionic_desc_info *desc_i
 		}
 	}
 
-	if (likely(netdev->features & NETIF_F_RXCSUM)) {
-		if (comp->csum_flags & IONIC_RXQ_COMP_CSUM_F_CALC) {
-			skb->ip_summed = CHECKSUM_COMPLETE;
-			skb->csum = (__wsum)le16_to_cpu(comp->csum);
-			stats->csum_complete++;
-		}
+	if (likely(netdev->features & NETIF_F_RXCSUM) &&
+	    (comp->csum_flags & IONIC_RXQ_COMP_CSUM_F_CALC)) {
+		skb->ip_summed = CHECKSUM_COMPLETE;
+		skb->csum = (__force __wsum)le16_to_cpu(comp->csum);
+		stats->csum_complete++;
 	} else {
 		stats->csum_none++;
 	}
@@ -210,10 +278,31 @@ static void ionic_rx_clean(struct ionic_queue *q, struct ionic_desc_info *desc_i
 		     (comp->csum_flags & IONIC_RXQ_COMP_CSUM_F_IP_BAD)))
 		stats->csum_error++;
 
-	if (likely(netdev->features & NETIF_F_HW_VLAN_CTAG_RX)) {
-		if (comp->csum_flags & IONIC_RXQ_COMP_CSUM_F_VLAN)
-			__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q),
-					       le16_to_cpu(comp->vlan_tci));
+	if (likely(netdev->features & NETIF_F_HW_VLAN_CTAG_RX) &&
+	    (comp->csum_flags & IONIC_RXQ_COMP_CSUM_F_VLAN)) {
+		__vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q),
+				       le16_to_cpu(comp->vlan_tci));
+		stats->vlan_stripped++;
+	}
+
+	if (unlikely(q->features & IONIC_RXQ_F_HWSTAMP)) {
+		__le64 *cq_desc_hwstamp;
+		u64 hwstamp;
+
+		cq_desc_hwstamp =
+			cq_info->cq_desc +
+			qcq->cq.desc_size -
+			sizeof(struct ionic_rxq_comp) -
+			IONIC_HWSTAMP_CQ_NEGOFFSET;
+
+		hwstamp = le64_to_cpu(*cq_desc_hwstamp);
+
+		if (hwstamp != IONIC_HWSTAMP_INVALID) {
+			skb_hwtstamps(skb)->hwtstamp = ionic_lif_phc_ktime(q->lif, hwstamp);
+			stats->hwstamp_valid++;
+		} else {
+			stats->hwstamp_invalid++;
+		}
 	}
 
 	if (le16_to_cpu(comp->len) <= q->lif->rx_copybreak)
@@ -222,24 +311,26 @@ static void ionic_rx_clean(struct ionic_queue *q, struct ionic_desc_info *desc_i
 		napi_gro_frags(&qcq->napi);
 }
 
-static bool ionic_rx_service(struct ionic_cq *cq, struct ionic_cq_info *cq_info)
+bool ionic_rx_service(struct ionic_cq *cq, struct ionic_cq_info *cq_info)
 {
-	struct ionic_rxq_comp *comp = cq_info->cq_desc;
 	struct ionic_queue *q = cq->bound_q;
 	struct ionic_desc_info *desc_info;
+	struct ionic_rxq_comp *comp;
+
+	comp = cq_info->cq_desc + cq->desc_size - sizeof(*comp);
 
 	if (!color_match(comp->pkt_type_color, cq->done_color))
 		return false;
 
 	/* check for empty queue */
-	if (q->tail->index == q->head->index)
+	if (q->tail_idx == q->head_idx)
 		return false;
 
-	desc_info = q->tail;
-	if (desc_info->index != le16_to_cpu(comp->comp_index))
+	if (q->tail_idx != le16_to_cpu(comp->comp_index))
 		return false;
 
-	q->tail = desc_info->next;
+	desc_info = &q->info[q->tail_idx];
+	q->tail_idx = (q->tail_idx + 1) & (q->num_descs - 1);
 
 	/* clean the related q entry, only one per qc completion */
 	ionic_rx_clean(q, desc_info, cq_info, desc_info->cb_arg);
@@ -250,235 +341,255 @@ static bool ionic_rx_service(struct ionic_cq *cq, struct ionic_cq_info *cq_info)
 	return true;
 }
 
-static u32 ionic_rx_walk_cq(struct ionic_cq *rxcq, u32 limit)
-{
-	u32 work_done = 0;
-
-	while (ionic_rx_service(rxcq, rxcq->tail)) {
-		if (rxcq->tail->last)
-			rxcq->done_color = !rxcq->done_color;
-		rxcq->tail = rxcq->tail->next;
-		DEBUG_STATS_CQE_CNT(rxcq);
-
-		if (++work_done >= limit)
-			break;
-	}
-
-	return work_done;
-}
-
-void ionic_rx_flush(struct ionic_cq *cq)
-{
-	struct ionic_dev *idev = &cq->lif->ionic->idev;
-	u32 work_done;
-
-	work_done = ionic_rx_walk_cq(cq, cq->num_descs);
-
-	if (work_done)
-		ionic_intr_credits(idev->intr_ctrl, cq->bound_intr->index,
-				   work_done, IONIC_INTR_CRED_RESET_COALESCE);
-}
-
-static struct page *ionic_rx_page_alloc(struct ionic_queue *q,
-					dma_addr_t *dma_addr)
-{
-	struct ionic_lif *lif = q->lif;
-	struct ionic_rx_stats *stats;
-	struct net_device *netdev;
-	struct device *dev;
-	struct page *page;
-
-	netdev = lif->netdev;
-	dev = lif->ionic->dev;
-	stats = q_to_rx_stats(q);
-	page = alloc_page(GFP_ATOMIC);
-	if (unlikely(!page)) {
-		net_err_ratelimited("%s: Page alloc failed on %s!\n",
-				    netdev->name, q->name);
-		stats->alloc_err++;
-		return NULL;
-	}
-
-	*dma_addr = dma_map_page(dev, page, 0, PAGE_SIZE, DMA_FROM_DEVICE);
-	if (unlikely(dma_mapping_error(dev, *dma_addr))) {
-		__free_page(page);
-		net_err_ratelimited("%s: DMA single map failed on %s!\n",
-				    netdev->name, q->name);
-		stats->dma_map_err++;
-		return NULL;
-	}
-
-	return page;
-}
-
-static void ionic_rx_page_free(struct ionic_queue *q, struct page *page,
-			       dma_addr_t dma_addr)
-{
-	struct ionic_lif *lif = q->lif;
-	struct net_device *netdev;
-	struct device *dev;
-
-	netdev = lif->netdev;
-	dev = lif->ionic->dev;
-
-	if (unlikely(!page)) {
-		net_err_ratelimited("%s: Trying to free unallocated buffer on %s!\n",
-				    netdev->name, q->name);
-		return;
-	}
-
-	dma_unmap_page(dev, dma_addr, PAGE_SIZE, DMA_FROM_DEVICE);
-
-	__free_page(page);
-}
-
-#define IONIC_RX_RING_DOORBELL_STRIDE		((1 << 5) - 1)
-#define IONIC_RX_RING_HEAD_BUF_SZ		2048
-
 void ionic_rx_fill(struct ionic_queue *q)
 {
 	struct net_device *netdev = q->lif->netdev;
 	struct ionic_desc_info *desc_info;
-	struct ionic_page_info *page_info;
 	struct ionic_rxq_sg_desc *sg_desc;
 	struct ionic_rxq_sg_elem *sg_elem;
+	struct ionic_buf_info *buf_info;
 	struct ionic_rxq_desc *desc;
 	unsigned int remain_len;
-	unsigned int seg_len;
+	unsigned int frag_len;
 	unsigned int nfrags;
-	bool ring_doorbell;
 	unsigned int i, j;
 	unsigned int len;
 
-	len = netdev->mtu + ETH_HLEN;
-	nfrags = round_up(len, PAGE_SIZE) / PAGE_SIZE;
+	len = netdev->mtu + ETH_HLEN + VLAN_HLEN;
 
 	for (i = ionic_q_space_avail(q); i; i--) {
+		nfrags = 0;
 		remain_len = len;
-		desc_info = q->head;
+		desc_info = &q->info[q->head_idx];
 		desc = desc_info->desc;
+		buf_info = &desc_info->bufs[0];
+
+		if (!buf_info->page) { /* alloc a new buffer? */
+			if (unlikely(ionic_rx_page_alloc(q, buf_info))) {
+				desc->addr = 0;
+				desc->len = 0;
+				return;
+			}
+		}
+
+		/* fill main descriptor - buf[0] */
+		desc->addr = cpu_to_le64(buf_info->dma_addr + buf_info->page_offset);
+		frag_len = min_t(u16, len, IONIC_PAGE_SIZE - buf_info->page_offset);
+		desc->len = cpu_to_le16(frag_len);
+		remain_len -= frag_len;
+		buf_info++;
+		nfrags++;
+
+		/* fill sg descriptors - buf[1..n] */
 		sg_desc = desc_info->sg_desc;
-		page_info = &desc_info->pages[0];
+		for (j = 0; remain_len > 0 && j < q->max_sg_elems; j++) {
+			sg_elem = &sg_desc->elems[j];
+			if (!buf_info->page) { /* alloc a new sg buffer? */
+				if (unlikely(ionic_rx_page_alloc(q, buf_info))) {
+					sg_elem->addr = 0;
+					sg_elem->len = 0;
+					return;
+				}
+			}
+
+			sg_elem->addr = cpu_to_le64(buf_info->dma_addr + buf_info->page_offset);
+			frag_len = min_t(u16, remain_len, IONIC_PAGE_SIZE - buf_info->page_offset);
+			sg_elem->len = cpu_to_le16(frag_len);
+			remain_len -= frag_len;
+			buf_info++;
+			nfrags++;
+		}
 
-		if (page_info->page) { /* recycle the buffer */
-			ring_doorbell = ((q->head->index + 1) &
-					IONIC_RX_RING_DOORBELL_STRIDE) == 0;
-			ionic_rxq_post(q, ring_doorbell, ionic_rx_clean, NULL);
-			continue;
+		/* clear end sg element as a sentinel */
+		if (j < q->max_sg_elems) {
+			sg_elem = &sg_desc->elems[j];
+			memset(sg_elem, 0, sizeof(*sg_elem));
 		}
 
-		/* fill main descriptor - pages[0] */
 		desc->opcode = (nfrags > 1) ? IONIC_RXQ_DESC_OPCODE_SG :
 					      IONIC_RXQ_DESC_OPCODE_SIMPLE;
-		desc_info->npages = nfrags;
-		page_info->page = ionic_rx_page_alloc(q, &page_info->dma_addr);
-		if (unlikely(!page_info->page)) {
-			desc->addr = 0;
-			desc->len = 0;
-			return;
-		}
-		desc->addr = cpu_to_le64(page_info->dma_addr);
-		seg_len = min_t(unsigned int, PAGE_SIZE, len);
-		desc->len = cpu_to_le16(seg_len);
-		remain_len -= seg_len;
-		page_info++;
+		desc_info->nbufs = nfrags;
+
+		ionic_rxq_post(q, false, ionic_rx_clean, NULL);
+	}
 
-		/* fill sg descriptors - pages[1..n] */
-		for (j = 0; j < nfrags - 1; j++) {
-			if (page_info->page) /* recycle the sg buffer */
-				continue;
+	ionic_dbell_ring(q->lif->kern_dbpage, q->hw_type,
+			 q->dbval | q->head_idx);
+}
 
-			sg_elem = &sg_desc->elems[j];
-			page_info->page = ionic_rx_page_alloc(q, &page_info->dma_addr);
-			if (unlikely(!page_info->page)) {
-				sg_elem->addr = 0;
-				sg_elem->len = 0;
-				return;
-			}
-			sg_elem->addr = cpu_to_le64(page_info->dma_addr);
-			seg_len = min_t(unsigned int, PAGE_SIZE, remain_len);
-			sg_elem->len = cpu_to_le16(seg_len);
-			remain_len -= seg_len;
-			page_info++;
+void ionic_rx_empty(struct ionic_queue *q)
+{
+	struct ionic_desc_info *desc_info;
+	struct ionic_buf_info *buf_info;
+	unsigned int i, j;
+
+	for (i = 0; i < q->num_descs; i++) {
+		desc_info = &q->info[i];
+		for (j = 0; j < IONIC_RX_MAX_SG_ELEMS + 1; j++) {
+			buf_info = &desc_info->bufs[j];
+			if (buf_info->page)
+				ionic_rx_page_free(q, buf_info);
 		}
 
-		ring_doorbell = ((q->head->index + 1) &
-				IONIC_RX_RING_DOORBELL_STRIDE) == 0;
-		ionic_rxq_post(q, ring_doorbell, ionic_rx_clean, NULL);
+		desc_info->nbufs = 0;
+		desc_info->cb = NULL;
+		desc_info->cb_arg = NULL;
 	}
+
+	q->head_idx = 0;
+	q->tail_idx = 0;
 }
 
-static void ionic_rx_fill_cb(void *arg)
+static void ionic_dim_update(struct ionic_qcq *qcq, int napi_mode)
 {
-	ionic_rx_fill(arg);
+	struct dim_sample dim_sample;
+	struct ionic_lif *lif;
+	unsigned int qi;
+	u64 pkts, bytes;
+
+	if (!qcq->intr.dim_coal_hw)
+		return;
+
+	lif = qcq->q.lif;
+	qi = qcq->cq.bound_q->index;
+
+	switch (napi_mode) {
+	case IONIC_LIF_F_TX_DIM_INTR:
+		pkts = lif->txqstats[qi].pkts;
+		bytes = lif->txqstats[qi].bytes;
+		break;
+	case IONIC_LIF_F_RX_DIM_INTR:
+		pkts = lif->rxqstats[qi].pkts;
+		bytes = lif->rxqstats[qi].bytes;
+		break;
+	default:
+		pkts = lif->txqstats[qi].pkts + lif->rxqstats[qi].pkts;
+		bytes = lif->txqstats[qi].bytes + lif->rxqstats[qi].bytes;
+		break;
+	}
+
+	dim_update_sample(qcq->cq.bound_intr->rearm_count,
+			  pkts, bytes, &dim_sample);
+
+	net_dim(&qcq->dim, dim_sample);
 }
 
-void ionic_rx_empty(struct ionic_queue *q)
+int ionic_tx_napi(struct napi_struct *napi, int budget)
 {
-	struct ionic_desc_info *cur;
-	struct ionic_rxq_desc *desc;
-	unsigned int i;
+	struct ionic_qcq *qcq = napi_to_qcq(napi);
+	struct ionic_cq *cq = napi_to_cq(napi);
+	struct ionic_dev *idev;
+	struct ionic_lif *lif;
+	u32 work_done = 0;
+	u32 flags = 0;
 
-	for (cur = q->tail; cur != q->head; cur = cur->next) {
-		desc = cur->desc;
-		desc->addr = 0;
-		desc->len = 0;
-
-		for (i = 0; i < cur->npages; i++) {
-			if (likely(cur->pages[i].page)) {
-				ionic_rx_page_free(q, cur->pages[i].page,
-						   cur->pages[i].dma_addr);
-				cur->pages[i].page = NULL;
-				cur->pages[i].dma_addr = 0;
-			}
-		}
+	lif = cq->bound_q->lif;
+	idev = &lif->ionic->idev;
+
+	work_done = ionic_cq_service(cq, budget,
+				     ionic_tx_service, NULL, NULL);
 
-		cur->cb_arg = NULL;
+	if (work_done < budget && napi_complete_done(napi, work_done)) {
+		ionic_dim_update(qcq, IONIC_LIF_F_TX_DIM_INTR);
+		flags |= IONIC_INTR_CRED_UNMASK;
+		cq->bound_intr->rearm_count++;
 	}
+
+	if (work_done || flags) {
+		flags |= IONIC_INTR_CRED_RESET_COALESCE;
+		ionic_intr_credits(idev->intr_ctrl,
+				   cq->bound_intr->index,
+				   work_done, flags);
+	}
+
+	return work_done;
 }
 
 int ionic_rx_napi(struct napi_struct *napi, int budget)
 {
 	struct ionic_qcq *qcq = napi_to_qcq(napi);
+	struct ionic_cq *cq = napi_to_cq(napi);
+	struct ionic_dev *idev;
+	struct ionic_lif *lif;
+	u16 rx_fill_threshold;
+	u32 work_done = 0;
+	u32 flags = 0;
+
+	lif = cq->bound_q->lif;
+	idev = &lif->ionic->idev;
+
+	work_done = ionic_cq_service(cq, budget,
+				     ionic_rx_service, NULL, NULL);
+
+	rx_fill_threshold = min_t(u16, IONIC_RX_FILL_THRESHOLD,
+				  cq->num_descs / IONIC_RX_FILL_DIV);
+	if (work_done && ionic_q_space_avail(cq->bound_q) >= rx_fill_threshold)
+		ionic_rx_fill(cq->bound_q);
+
+	if (work_done < budget && napi_complete_done(napi, work_done)) {
+		ionic_dim_update(qcq, IONIC_LIF_F_RX_DIM_INTR);
+		flags |= IONIC_INTR_CRED_UNMASK;
+		cq->bound_intr->rearm_count++;
+	}
+
+	if (work_done || flags) {
+		flags |= IONIC_INTR_CRED_RESET_COALESCE;
+		ionic_intr_credits(idev->intr_ctrl,
+				   cq->bound_intr->index,
+				   work_done, flags);
+	}
+
+	return work_done;
+}
+
+int ionic_txrx_napi(struct napi_struct *napi, int budget)
+{
+	struct ionic_qcq *qcq = napi_to_qcq(napi);
 	struct ionic_cq *rxcq = napi_to_cq(napi);
 	unsigned int qi = rxcq->bound_q->index;
 	struct ionic_dev *idev;
 	struct ionic_lif *lif;
 	struct ionic_cq *txcq;
-	u32 work_done = 0;
+	u16 rx_fill_threshold;
+	u32 rx_work_done = 0;
+	u32 tx_work_done = 0;
 	u32 flags = 0;
 
 	lif = rxcq->bound_q->lif;
 	idev = &lif->ionic->idev;
-	txcq = &lif->txqcqs[qi].qcq->cq;
+	txcq = &lif->txqcqs[qi]->cq;
 
-	ionic_tx_flush(txcq);
+	tx_work_done = ionic_cq_service(txcq, IONIC_TX_BUDGET_DEFAULT,
+					ionic_tx_service, NULL, NULL);
 
-	work_done = ionic_rx_walk_cq(rxcq, budget);
+	rx_work_done = ionic_cq_service(rxcq, budget,
+					ionic_rx_service, NULL, NULL);
 
-	if (work_done)
-		ionic_rx_fill_cb(rxcq->bound_q);
+	rx_fill_threshold = min_t(u16, IONIC_RX_FILL_THRESHOLD,
+				  rxcq->num_descs / IONIC_RX_FILL_DIV);
+	if (rx_work_done && ionic_q_space_avail(rxcq->bound_q) >= rx_fill_threshold)
+		ionic_rx_fill(rxcq->bound_q);
 
-	if (work_done < budget && napi_complete_done(napi, work_done)) {
+	if (rx_work_done < budget && napi_complete_done(napi, rx_work_done)) {
+		ionic_dim_update(qcq, 0);
 		flags |= IONIC_INTR_CRED_UNMASK;
-		DEBUG_STATS_INTR_REARM(rxcq->bound_intr);
+		rxcq->bound_intr->rearm_count++;
 	}
 
-	if (work_done || flags) {
+	if (rx_work_done || flags) {
 		flags |= IONIC_INTR_CRED_RESET_COALESCE;
 		ionic_intr_credits(idev->intr_ctrl, rxcq->bound_intr->index,
-				   work_done, flags);
+				   tx_work_done + rx_work_done, flags);
 	}
 
-	DEBUG_STATS_NAPI_POLL(qcq, work_done);
-
-	return work_done;
+	return rx_work_done;
 }
 
-static dma_addr_t ionic_tx_map_single(struct ionic_queue *q, void *data, size_t len)
+static dma_addr_t ionic_tx_map_single(struct ionic_queue *q,
+				      void *data, size_t len)
 {
 	struct ionic_tx_stats *stats = q_to_tx_stats(q);
-	struct device *dev = q->lif->ionic->dev;
+	struct device *dev = q->dev;
 	dma_addr_t dma_addr;
 
 	dma_addr = dma_map_single(dev, data, len, DMA_TO_DEVICE);
@@ -491,11 +602,12 @@ static dma_addr_t ionic_tx_map_single(struct ionic_queue *q, void *data, size_t
 	return dma_addr;
 }
 
-static dma_addr_t ionic_tx_map_frag(struct ionic_queue *q, const skb_frag_t *frag,
+static dma_addr_t ionic_tx_map_frag(struct ionic_queue *q,
+				    const skb_frag_t *frag,
 				    size_t offset, size_t len)
 {
 	struct ionic_tx_stats *stats = q_to_tx_stats(q);
-	struct device *dev = q->lif->ionic->dev;
+	struct device *dev = q->dev;
 	dma_addr_t dma_addr;
 
 	dma_addr = skb_frag_dma_map(dev, frag, offset, len, DMA_TO_DEVICE);
@@ -507,90 +619,199 @@ static dma_addr_t ionic_tx_map_frag(struct ionic_queue *q, const skb_frag_t *fra
 	return dma_addr;
 }
 
-static void ionic_tx_clean(struct ionic_queue *q, struct ionic_desc_info *desc_info,
-			   struct ionic_cq_info *cq_info, void *cb_arg)
+static int ionic_tx_map_skb(struct ionic_queue *q, struct sk_buff *skb,
+			    struct ionic_desc_info *desc_info)
 {
-	struct ionic_txq_sg_desc *sg_desc = desc_info->sg_desc;
-	struct ionic_txq_sg_elem *elem = sg_desc->elems;
+	struct ionic_buf_info *buf_info = desc_info->bufs;
 	struct ionic_tx_stats *stats = q_to_tx_stats(q);
-	struct ionic_txq_desc *desc = desc_info->desc;
-	struct device *dev = q->lif->ionic->dev;
-	u8 opcode, flags, nsge;
-	u16 queue_index;
+	struct device *dev = q->dev;
+	dma_addr_t dma_addr;
+	unsigned int nfrags;
+	skb_frag_t *frag;
+	int frag_idx;
+
+	dma_addr = ionic_tx_map_single(q, skb->data, skb_headlen(skb));
+	if (dma_mapping_error(dev, dma_addr)) {
+		stats->dma_map_err++;
+		return -EIO;
+	}
+	buf_info->dma_addr = dma_addr;
+	buf_info->len = skb_headlen(skb);
+	buf_info++;
+
+	frag = skb_shinfo(skb)->frags;
+	nfrags = skb_shinfo(skb)->nr_frags;
+	for (frag_idx = 0; frag_idx < nfrags; frag_idx++, frag++) {
+		dma_addr = ionic_tx_map_frag(q, frag, 0, skb_frag_size(frag));
+		if (dma_mapping_error(dev, dma_addr)) {
+			stats->dma_map_err++;
+			goto dma_fail;
+		}
+		buf_info->dma_addr = dma_addr;
+		buf_info->len = skb_frag_size(frag);
+		buf_info++;
+	}
+
+	desc_info->nbufs = 1 + nfrags;
+
+	return 0;
+
+dma_fail:
+	/* unwind the frag mappings and the head mapping */
+	while (frag_idx > 0) {
+		frag_idx--;
+		buf_info--;
+		dma_unmap_page(dev, buf_info->dma_addr,
+			       buf_info->len, DMA_TO_DEVICE);
+	}
+	dma_unmap_single(dev, buf_info->dma_addr, buf_info->len, DMA_TO_DEVICE);
+	return -EIO;
+}
+
+static void ionic_tx_desc_unmap_bufs(struct ionic_queue *q,
+				     struct ionic_desc_info *desc_info)
+{
+	struct ionic_buf_info *buf_info = desc_info->bufs;
+	struct device *dev = q->dev;
 	unsigned int i;
-	u64 addr;
 
-	decode_txq_desc_cmd(le64_to_cpu(desc->cmd),
-			    &opcode, &flags, &nsge, &addr);
+	if (!desc_info->nbufs)
+		return;
 
-	/* use unmap_single only if either this is not TSO,
-	 * or this is first descriptor of a TSO
-	 */
-	if (opcode != IONIC_TXQ_DESC_OPCODE_TSO ||
-	    flags & IONIC_TXQ_DESC_FLAG_TSO_SOT)
-		dma_unmap_single(dev, (dma_addr_t)addr,
-				 le16_to_cpu(desc->len), DMA_TO_DEVICE);
-	else
-		dma_unmap_page(dev, (dma_addr_t)addr,
-			       le16_to_cpu(desc->len), DMA_TO_DEVICE);
-
-	for (i = 0; i < nsge; i++, elem++)
-		dma_unmap_page(dev, (dma_addr_t)le64_to_cpu(elem->addr),
-			       le16_to_cpu(elem->len), DMA_TO_DEVICE);
-
-	if (cb_arg) {
-		struct sk_buff *skb = cb_arg;
-		u32 len = skb->len;
-
-		queue_index = skb_get_queue_mapping(skb);
-		if (unlikely(__netif_subqueue_stopped(q->lif->netdev,
-						      queue_index))) {
-			netif_wake_subqueue(q->lif->netdev, queue_index);
-			q->wake++;
+	dma_unmap_single(dev, (dma_addr_t)buf_info->dma_addr,
+			 buf_info->len, DMA_TO_DEVICE);
+	buf_info++;
+	for (i = 1; i < desc_info->nbufs; i++, buf_info++)
+		dma_unmap_page(dev, (dma_addr_t)buf_info->dma_addr,
+			       buf_info->len, DMA_TO_DEVICE);
+
+	desc_info->nbufs = 0;
+}
+
+static void ionic_tx_clean(struct ionic_queue *q,
+			   struct ionic_desc_info *desc_info,
+			   struct ionic_cq_info *cq_info,
+			   void *cb_arg)
+{
+	struct ionic_tx_stats *stats = q_to_tx_stats(q);
+	struct ionic_qcq *qcq = q_to_qcq(q);
+	struct sk_buff *skb = cb_arg;
+	u16 qi;
+
+	ionic_tx_desc_unmap_bufs(q, desc_info);
+
+	if (!skb)
+		return;
+
+	qi = skb_get_queue_mapping(skb);
+
+	if (unlikely(q->features & IONIC_TXQ_F_HWSTAMP)) {
+		if (cq_info) {
+			struct skb_shared_hwtstamps hwts = {};
+			__le64 *cq_desc_hwstamp;
+			u64 hwstamp;
+
+			cq_desc_hwstamp =
+				cq_info->cq_desc +
+				qcq->cq.desc_size -
+				sizeof(struct ionic_txq_comp) -
+				IONIC_HWSTAMP_CQ_NEGOFFSET;
+
+			hwstamp = le64_to_cpu(*cq_desc_hwstamp);
+
+			if (hwstamp != IONIC_HWSTAMP_INVALID) {
+				hwts.hwtstamp = ionic_lif_phc_ktime(q->lif, hwstamp);
+
+				skb_shinfo(skb)->tx_flags |= SKBTX_IN_PROGRESS;
+				skb_tstamp_tx(skb, &hwts);
+
+				stats->hwstamp_valid++;
+			} else {
+				stats->hwstamp_invalid++;
+			}
 		}
-		dev_kfree_skb_any(skb);
-		stats->clean++;
-		netdev_tx_completed_queue(q_to_ndq(q), 1, len);
+
+	} else if (unlikely(__netif_subqueue_stopped(q->lif->netdev, qi))) {
+		netif_wake_subqueue(q->lif->netdev, qi);
 	}
+
+	desc_info->bytes = skb->len;
+	stats->clean++;
+
+	dev_consume_skb_any(skb);
 }
 
-void ionic_tx_flush(struct ionic_cq *cq)
+bool ionic_tx_service(struct ionic_cq *cq, struct ionic_cq_info *cq_info)
 {
-	struct ionic_txq_comp *comp = cq->tail->cq_desc;
-	struct ionic_dev *idev = &cq->lif->ionic->idev;
 	struct ionic_queue *q = cq->bound_q;
 	struct ionic_desc_info *desc_info;
-	unsigned int work_done = 0;
-
-	/* walk the completed cq entries */
-	while (work_done < cq->num_descs &&
-	       color_match(comp->color, cq->done_color)) {
-
-		/* clean the related q entries, there could be
-		 * several q entries completed for each cq completion
-		 */
-		do {
-			desc_info = q->tail;
-			q->tail = desc_info->next;
-			ionic_tx_clean(q, desc_info, cq->tail,
-				       desc_info->cb_arg);
-			desc_info->cb = NULL;
-			desc_info->cb_arg = NULL;
-		} while (desc_info->index != le16_to_cpu(comp->comp_index));
-
-		if (cq->tail->last)
-			cq->done_color = !cq->done_color;
-
-		cq->tail = cq->tail->next;
-		comp = cq->tail->cq_desc;
-		DEBUG_STATS_CQE_CNT(cq);
-
-		work_done++;
-	}
+	struct ionic_txq_comp *comp;
+	int bytes = 0;
+	int pkts = 0;
+	u16 index;
+
+	comp = cq_info->cq_desc + cq->desc_size - sizeof(*comp);
+
+	if (!color_match(comp->color, cq->done_color))
+		return false;
+
+	/* clean the related q entries, there could be
+	 * several q entries completed for each cq completion
+	 */
+	do {
+		desc_info = &q->info[q->tail_idx];
+		desc_info->bytes = 0;
+		index = q->tail_idx;
+		q->tail_idx = (q->tail_idx + 1) & (q->num_descs - 1);
+		ionic_tx_clean(q, desc_info, cq_info, desc_info->cb_arg);
+		if (desc_info->cb_arg) {
+			pkts++;
+			bytes += desc_info->bytes;
+		}
+		desc_info->cb = NULL;
+		desc_info->cb_arg = NULL;
+	} while (index != le16_to_cpu(comp->comp_index));
+
+	if (pkts && bytes && !unlikely(q->features & IONIC_TXQ_F_HWSTAMP))
+		netdev_tx_completed_queue(q_to_ndq(q), pkts, bytes);
 
+	return true;
+}
+
+void ionic_tx_flush(struct ionic_cq *cq)
+{
+	struct ionic_dev *idev = &cq->lif->ionic->idev;
+	u32 work_done;
+
+	work_done = ionic_cq_service(cq, cq->num_descs,
+				     ionic_tx_service, NULL, NULL);
 	if (work_done)
 		ionic_intr_credits(idev->intr_ctrl, cq->bound_intr->index,
-				   work_done, 0);
+				   work_done, IONIC_INTR_CRED_RESET_COALESCE);
+}
+
+void ionic_tx_empty(struct ionic_queue *q)
+{
+	struct ionic_desc_info *desc_info;
+	int bytes = 0;
+	int pkts = 0;
+
+	/* walk the not completed tx entries, if any */
+	while (q->head_idx != q->tail_idx) {
+		desc_info = &q->info[q->tail_idx];
+		desc_info->bytes = 0;
+		q->tail_idx = (q->tail_idx + 1) & (q->num_descs - 1);
+		ionic_tx_clean(q, desc_info, NULL, desc_info->cb_arg);
+		if (desc_info->cb_arg) {
+			pkts++;
+			bytes += desc_info->bytes;
+		}
+		desc_info->cb = NULL;
+		desc_info->cb_arg = NULL;
+	}
+
+	if (pkts && bytes && !unlikely(q->features & IONIC_TXQ_F_HWSTAMP))
+		netdev_tx_completed_queue(q_to_ndq(q), pkts, bytes);
 }
 
 static int ionic_tx_tcp_inner_pseudo_csum(struct sk_buff *skb)
@@ -632,10 +853,7 @@ static int ionic_tx_tcp_pseudo_csum(struct sk_buff *skb)
 					   ip_hdr(skb)->daddr,
 					   0, IPPROTO_TCP, 0);
 	} else if (skb->protocol == cpu_to_be16(ETH_P_IPV6)) {
-		tcp_hdr(skb)->check =
-			~csum_ipv6_magic(&ipv6_hdr(skb)->saddr,
-					 &ipv6_hdr(skb)->daddr,
-					 0, IPPROTO_TCP, 0);
+		tcp_v6_gso_csum_prep(skb);
 	}
 
 	return 0;
@@ -664,46 +882,32 @@ static void ionic_tx_tso_post(struct ionic_queue *q, struct ionic_txq_desc *desc
 	desc->hdr_len = cpu_to_le16(hdrlen);
 	desc->mss = cpu_to_le16(mss);
 
-	if (done) {
+	if (start) {
 		skb_tx_timestamp(skb);
-		netdev_tx_sent_queue(q_to_ndq(q), skb->len);
-		ionic_txq_post(q, !netdev_xmit_more(), ionic_tx_clean, skb);
+		if (!unlikely(q->features & IONIC_TXQ_F_HWSTAMP))
+			netdev_tx_sent_queue(q_to_ndq(q), skb->len);
+		ionic_txq_post(q, false, ionic_tx_clean, skb);
 	} else {
-		ionic_txq_post(q, false, ionic_tx_clean, NULL);
+		ionic_txq_post(q, done, NULL, NULL);
 	}
 }
 
-static struct ionic_txq_desc *ionic_tx_tso_next(struct ionic_queue *q,
-						struct ionic_txq_sg_elem **elem)
-{
-	struct ionic_txq_sg_desc *sg_desc = q->head->sg_desc;
-	struct ionic_txq_desc *desc = q->head->desc;
-
-	*elem = sg_desc->elems;
-	return desc;
-}
-
 static int ionic_tx_tso(struct ionic_queue *q, struct sk_buff *skb)
 {
 	struct ionic_tx_stats *stats = q_to_tx_stats(q);
-	struct ionic_desc_info *abort = q->head;
-	struct device *dev = q->lif->ionic->dev;
-	struct ionic_desc_info *rewind = abort;
+	struct ionic_desc_info *desc_info;
+	struct ionic_buf_info *buf_info;
 	struct ionic_txq_sg_elem *elem;
 	struct ionic_txq_desc *desc;
-	unsigned int frag_left = 0;
-	unsigned int offset = 0;
-	unsigned int len_left;
+	unsigned int chunk_len;
+	unsigned int frag_rem;
+	unsigned int tso_rem;
+	unsigned int seg_rem;
 	dma_addr_t desc_addr;
+	dma_addr_t frag_addr;
 	unsigned int hdrlen;
-	unsigned int nfrags;
-	unsigned int seglen;
-	u64 total_bytes = 0;
-	u64 total_pkts = 0;
-	unsigned int left;
 	unsigned int len;
 	unsigned int mss;
-	skb_frag_t *frag;
 	bool start, done;
 	bool outer_csum;
 	bool has_vlan;
@@ -713,9 +917,14 @@ static int ionic_tx_tso(struct ionic_queue *q, struct sk_buff *skb)
 	bool encap;
 	int err;
 
+	desc_info = &q->info[q->head_idx];
+	buf_info = desc_info->bufs;
+
+	if (unlikely(ionic_tx_map_skb(q, skb, desc_info)))
+		return -EIO;
+
+	len = skb->len;
 	mss = skb_shinfo(skb)->gso_size;
-	nfrags = skb_shinfo(skb)->nr_frags;
-	len_left = skb->len - skb_headlen(skb);
 	outer_csum = (skb_shinfo(skb)->gso_type & SKB_GSO_GRE_CSUM) ||
 		     (skb_shinfo(skb)->gso_type & SKB_GSO_UDP_TUNNEL_CSUM);
 	has_vlan = !!skb_vlan_tag_present(skb);
@@ -731,133 +940,86 @@ static int ionic_tx_tso(struct ionic_queue *q, struct sk_buff *skb)
 		err = ionic_tx_tcp_inner_pseudo_csum(skb);
 	else
 		err = ionic_tx_tcp_pseudo_csum(skb);
-	if (err)
+	if (err) {
+		/* clean up mapping from ionic_tx_map_skb */
+		ionic_tx_desc_unmap_bufs(q, desc_info);
 		return err;
+	}
 
 	if (encap)
-		hdrlen = skb_inner_transport_header(skb) - skb->data +
-			 inner_tcp_hdrlen(skb);
+		hdrlen = skb_inner_tcp_all_headers(skb);
 	else
-		hdrlen = skb_transport_offset(skb) + tcp_hdrlen(skb);
+		hdrlen = skb_tcp_all_headers(skb);
 
-	seglen = hdrlen + mss;
-	left = skb_headlen(skb);
+	tso_rem = len;
+	seg_rem = min(tso_rem, hdrlen + mss);
 
-	desc = ionic_tx_tso_next(q, &elem);
-	start = true;
+	frag_addr = 0;
+	frag_rem = 0;
 
-	/* Chop skb->data up into desc segments */
+	start = true;
 
-	while (left > 0) {
-		len = min(seglen, left);
-		frag_left = seglen - len;
-		desc_addr = ionic_tx_map_single(q, skb->data + offset, len);
-		if (dma_mapping_error(dev, desc_addr))
-			goto err_out_abort;
-		desc_len = len;
+	while (tso_rem > 0) {
+		desc = NULL;
+		elem = NULL;
+		desc_addr = 0;
+		desc_len = 0;
 		desc_nsge = 0;
-		left -= len;
-		offset += len;
-		if (nfrags > 0 && frag_left > 0)
-			continue;
-		done = (nfrags == 0 && left == 0);
-		ionic_tx_tso_post(q, desc, skb,
-				  desc_addr, desc_nsge, desc_len,
-				  hdrlen, mss,
-				  outer_csum,
-				  vlan_tci, has_vlan,
-				  start, done);
-		total_pkts++;
-		total_bytes += start ? len : len + hdrlen;
-		desc = ionic_tx_tso_next(q, &elem);
-		start = false;
-		seglen = mss;
-	}
-
-	/* Chop skb frags into desc segments */
-
-	for (frag = skb_shinfo(skb)->frags; len_left; frag++) {
-		offset = 0;
-		left = skb_frag_size(frag);
-		len_left -= left;
-		nfrags--;
-		stats->frags++;
-
-		while (left > 0) {
-			if (frag_left > 0) {
-				len = min(frag_left, left);
-				frag_left -= len;
-				elem->addr =
-				    cpu_to_le64(ionic_tx_map_frag(q, frag,
-								  offset, len));
-				if (dma_mapping_error(dev, elem->addr))
-					goto err_out_abort;
-				elem->len = cpu_to_le16(len);
+		/* use fragments until we have enough to post a single descriptor */
+		while (seg_rem > 0) {
+			/* if the fragment is exhausted then move to the next one */
+			if (frag_rem == 0) {
+				/* grab the next fragment */
+				frag_addr = buf_info->dma_addr;
+				frag_rem = buf_info->len;
+				buf_info++;
+			}
+			chunk_len = min(frag_rem, seg_rem);
+			if (!desc) {
+				/* fill main descriptor */
+				desc = desc_info->txq_desc;
+				elem = desc_info->txq_sg_desc->elems;
+				desc_addr = frag_addr;
+				desc_len = chunk_len;
+			} else {
+				/* fill sg descriptor */
+				elem->addr = cpu_to_le64(frag_addr);
+				elem->len = cpu_to_le16(chunk_len);
 				elem++;
 				desc_nsge++;
-				left -= len;
-				offset += len;
-				if (nfrags > 0 && frag_left > 0)
-					continue;
-				done = (nfrags == 0 && left == 0);
-				ionic_tx_tso_post(q, desc, skb, desc_addr,
-						  desc_nsge, desc_len,
-						  hdrlen, mss, outer_csum,
-						  vlan_tci, has_vlan,
-						  start, done);
-				total_pkts++;
-				total_bytes += start ? len : len + hdrlen;
-				desc = ionic_tx_tso_next(q, &elem);
-				start = false;
-			} else {
-				len = min(mss, left);
-				frag_left = mss - len;
-				desc_addr = ionic_tx_map_frag(q, frag,
-							      offset, len);
-				if (dma_mapping_error(dev, desc_addr))
-					goto err_out_abort;
-				desc_len = len;
-				desc_nsge = 0;
-				left -= len;
-				offset += len;
-				if (nfrags > 0 && frag_left > 0)
-					continue;
-				done = (nfrags == 0 && left == 0);
-				ionic_tx_tso_post(q, desc, skb, desc_addr,
-						  desc_nsge, desc_len,
-						  hdrlen, mss, outer_csum,
-						  vlan_tci, has_vlan,
-						  start, done);
-				total_pkts++;
-				total_bytes += start ? len : len + hdrlen;
-				desc = ionic_tx_tso_next(q, &elem);
-				start = false;
 			}
+			frag_addr += chunk_len;
+			frag_rem -= chunk_len;
+			tso_rem -= chunk_len;
+			seg_rem -= chunk_len;
 		}
+		seg_rem = min(tso_rem, mss);
+		done = (tso_rem == 0);
+		/* post descriptor */
+		ionic_tx_tso_post(q, desc, skb,
+				  desc_addr, desc_nsge, desc_len,
+				  hdrlen, mss, outer_csum, vlan_tci, has_vlan,
+				  start, done);
+		start = false;
+		/* Buffer information is stored with the first tso descriptor */
+		desc_info = &q->info[q->head_idx];
+		desc_info->nbufs = 0;
 	}
 
-	stats->pkts += total_pkts;
-	stats->bytes += total_bytes;
+	stats->pkts += DIV_ROUND_UP(len - hdrlen, mss);
+	stats->bytes += len;
 	stats->tso++;
+	stats->tso_bytes = len;
 
 	return 0;
-
-err_out_abort:
-	while (rewind->desc != q->head->desc) {
-		ionic_tx_clean(q, rewind, NULL, NULL);
-		rewind = rewind->next;
-	}
-	q->head = abort;
-
-	return -ENOMEM;
 }
 
-static int ionic_tx_calc_csum(struct ionic_queue *q, struct sk_buff *skb)
+static void ionic_tx_calc_csum(struct ionic_queue *q, struct sk_buff *skb,
+			       struct ionic_desc_info *desc_info)
 {
+	struct ionic_txq_desc *desc = desc_info->txq_desc;
+	struct ionic_buf_info *buf_info = desc_info->bufs;
 	struct ionic_tx_stats *stats = q_to_tx_stats(q);
-	struct ionic_txq_desc *desc = q->head->desc;
-	struct device *dev = q->lif->ionic->dev;
-	dma_addr_t dma_addr;
 	bool has_vlan;
 	u8 flags = 0;
 	bool encap;
@@ -866,35 +1028,35 @@ static int ionic_tx_calc_csum(struct ionic_queue *q, struct sk_buff *skb)
 	has_vlan = !!skb_vlan_tag_present(skb);
 	encap = skb->encapsulation;
 
-	dma_addr = ionic_tx_map_single(q, skb->data, skb_headlen(skb));
-	if (dma_mapping_error(dev, dma_addr))
-		return -ENOMEM;
-
 	flags |= has_vlan ? IONIC_TXQ_DESC_FLAG_VLAN : 0;
 	flags |= encap ? IONIC_TXQ_DESC_FLAG_ENCAP : 0;
 
 	cmd = encode_txq_desc_cmd(IONIC_TXQ_DESC_OPCODE_CSUM_PARTIAL,
-				  flags, skb_shinfo(skb)->nr_frags, dma_addr);
+				  flags, skb_shinfo(skb)->nr_frags,
+				  buf_info->dma_addr);
 	desc->cmd = cpu_to_le64(cmd);
-	desc->len = cpu_to_le16(skb_headlen(skb));
-	desc->vlan_tci = cpu_to_le16(skb_vlan_tag_get(skb));
+	desc->len = cpu_to_le16(buf_info->len);
+	if (has_vlan) {
+		desc->vlan_tci = cpu_to_le16(skb_vlan_tag_get(skb));
+		stats->vlan_inserted++;
+	} else {
+		desc->vlan_tci = 0;
+	}
 	desc->csum_start = cpu_to_le16(skb_checksum_start_offset(skb));
 	desc->csum_offset = cpu_to_le16(skb->csum_offset);
 
-	if (skb->csum_not_inet)
+	if (skb_csum_is_sctp(skb))
 		stats->crc32_csum++;
 	else
 		stats->csum++;
-
-	return 0;
 }
 
-static int ionic_tx_calc_no_csum(struct ionic_queue *q, struct sk_buff *skb)
+static void ionic_tx_calc_no_csum(struct ionic_queue *q, struct sk_buff *skb,
+				  struct ionic_desc_info *desc_info)
 {
+	struct ionic_txq_desc *desc = desc_info->txq_desc;
+	struct ionic_buf_info *buf_info = desc_info->bufs;
 	struct ionic_tx_stats *stats = q_to_tx_stats(q);
-	struct ionic_txq_desc *desc = q->head->desc;
-	struct device *dev = q->lif->ionic->dev;
-	dma_addr_t dma_addr;
 	bool has_vlan;
 	u8 flags = 0;
 	bool encap;
@@ -903,72 +1065,66 @@ static int ionic_tx_calc_no_csum(struct ionic_queue *q, struct sk_buff *skb)
 	has_vlan = !!skb_vlan_tag_present(skb);
 	encap = skb->encapsulation;
 
-	dma_addr = ionic_tx_map_single(q, skb->data, skb_headlen(skb));
-	if (dma_mapping_error(dev, dma_addr))
-		return -ENOMEM;
-
 	flags |= has_vlan ? IONIC_TXQ_DESC_FLAG_VLAN : 0;
 	flags |= encap ? IONIC_TXQ_DESC_FLAG_ENCAP : 0;
 
 	cmd = encode_txq_desc_cmd(IONIC_TXQ_DESC_OPCODE_CSUM_NONE,
-				  flags, skb_shinfo(skb)->nr_frags, dma_addr);
+				  flags, skb_shinfo(skb)->nr_frags,
+				  buf_info->dma_addr);
 	desc->cmd = cpu_to_le64(cmd);
-	desc->len = cpu_to_le16(skb_headlen(skb));
-	desc->vlan_tci = cpu_to_le16(skb_vlan_tag_get(skb));
-
-	stats->no_csum++;
+	desc->len = cpu_to_le16(buf_info->len);
+	if (has_vlan) {
+		desc->vlan_tci = cpu_to_le16(skb_vlan_tag_get(skb));
+		stats->vlan_inserted++;
+	} else {
+		desc->vlan_tci = 0;
+	}
+	desc->csum_start = 0;
+	desc->csum_offset = 0;
 
-	return 0;
+	stats->csum_none++;
 }
 
-static int ionic_tx_skb_frags(struct ionic_queue *q, struct sk_buff *skb)
+static void ionic_tx_skb_frags(struct ionic_queue *q, struct sk_buff *skb,
+			       struct ionic_desc_info *desc_info)
 {
-	struct ionic_txq_sg_desc *sg_desc = q->head->sg_desc;
-	unsigned int len_left = skb->len - skb_headlen(skb);
+	struct ionic_txq_sg_desc *sg_desc = desc_info->txq_sg_desc;
+	struct ionic_buf_info *buf_info = &desc_info->bufs[1];
 	struct ionic_txq_sg_elem *elem = sg_desc->elems;
 	struct ionic_tx_stats *stats = q_to_tx_stats(q);
-	struct device *dev = q->lif->ionic->dev;
-	dma_addr_t dma_addr;
-	skb_frag_t *frag;
-	u16 len;
+	unsigned int i;
 
-	for (frag = skb_shinfo(skb)->frags; len_left; frag++, elem++) {
-		len = skb_frag_size(frag);
-		elem->len = cpu_to_le16(len);
-		dma_addr = ionic_tx_map_frag(q, frag, 0, len);
-		if (dma_mapping_error(dev, dma_addr))
-			return -ENOMEM;
-		elem->addr = cpu_to_le64(dma_addr);
-		len_left -= len;
-		stats->frags++;
+	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++, buf_info++, elem++) {
+		elem->addr = cpu_to_le64(buf_info->dma_addr);
+		elem->len = cpu_to_le16(buf_info->len);
 	}
 
-	return 0;
+	stats->frags += skb_shinfo(skb)->nr_frags;
 }
 
 static int ionic_tx(struct ionic_queue *q, struct sk_buff *skb)
 {
+	struct ionic_desc_info *desc_info = &q->info[q->head_idx];
 	struct ionic_tx_stats *stats = q_to_tx_stats(q);
-	int err;
+
+	if (unlikely(ionic_tx_map_skb(q, skb, desc_info)))
+		return -EIO;
 
 	/* set up the initial descriptor */
 	if (skb->ip_summed == CHECKSUM_PARTIAL)
-		err = ionic_tx_calc_csum(q, skb);
+		ionic_tx_calc_csum(q, skb, desc_info);
 	else
-		err = ionic_tx_calc_no_csum(q, skb);
-	if (err)
-		return err;
+		ionic_tx_calc_no_csum(q, skb, desc_info);
 
 	/* add frags */
-	err = ionic_tx_skb_frags(q, skb);
-	if (err)
-		return err;
+	ionic_tx_skb_frags(q, skb, desc_info);
 
 	skb_tx_timestamp(skb);
 	stats->pkts++;
 	stats->bytes += skb->len;
 
-	netdev_tx_sent_queue(q_to_ndq(q), skb->len);
+	if (!unlikely(q->features & IONIC_TXQ_F_HWSTAMP))
+		netdev_tx_sent_queue(q_to_ndq(q), skb->len);
 	ionic_txq_post(q, !netdev_xmit_more(), ionic_tx_clean, skb);
 
 	return 0;
@@ -977,15 +1133,18 @@ static int ionic_tx(struct ionic_queue *q, struct sk_buff *skb)
 static int ionic_tx_descs_needed(struct ionic_queue *q, struct sk_buff *skb)
 {
 	struct ionic_tx_stats *stats = q_to_tx_stats(q);
+	int ndescs;
 	int err;
 
-	/* If TSO, need roundup(skb->len/mss) descs */
+	/* Each desc is mss long max, so a descriptor for each gso_seg */
 	if (skb_is_gso(skb))
-		return (skb->len / skb_shinfo(skb)->gso_size) + 1;
+		ndescs = skb_shinfo(skb)->gso_segs;
+	else
+		ndescs = 1;
 
 	/* If non-TSO, just need 1 desc and nr_frags sg elems */
-	if (skb_shinfo(skb)->nr_frags <= IONIC_TX_MAX_SG_ELEMS)
-		return 1;
+	if (skb_shinfo(skb)->nr_frags <= q->max_sg_elems)
+		return ndescs;
 
 	/* Too many frags, so linearize */
 	err = skb_linearize(skb);
@@ -994,8 +1153,7 @@ static int ionic_tx_descs_needed(struct ionic_queue *q, struct sk_buff *skb)
 
 	stats->linearize++;
 
-	/* Need 1 desc and zero sg elems */
-	return 1;
+	return ndescs;
 }
 
 static int ionic_maybe_stop_tx(struct ionic_queue *q, int ndescs)
@@ -1004,7 +1162,6 @@ static int ionic_maybe_stop_tx(struct ionic_queue *q, int ndescs)
 
 	if (unlikely(!ionic_q_has_space(q, ndescs))) {
 		netif_stop_subqueue(q->lif->netdev, q->index);
-		q->stop++;
 		stopped = 1;
 
 		/* Might race with ionic_tx_clean, check again */
@@ -1018,6 +1175,42 @@ static int ionic_maybe_stop_tx(struct ionic_queue *q, int ndescs)
 	return stopped;
 }
 
+static netdev_tx_t ionic_start_hwstamp_xmit(struct sk_buff *skb,
+					    struct net_device *netdev)
+{
+	struct ionic_lif *lif = netdev_priv(netdev);
+	struct ionic_queue *q = &lif->hwstamp_txq->q;
+	int err, ndescs;
+
+	/* Does not stop/start txq, because we post to a separate tx queue
+	 * for timestamping, and if a packet can't be posted immediately to
+	 * the timestamping queue, it is dropped.
+	 */
+
+	ndescs = ionic_tx_descs_needed(q, skb);
+	if (unlikely(ndescs < 0))
+		goto err_out_drop;
+
+	if (unlikely(!ionic_q_has_space(q, ndescs)))
+		goto err_out_drop;
+
+	skb_shinfo(skb)->tx_flags |= SKBTX_HW_TSTAMP;
+	if (skb_is_gso(skb))
+		err = ionic_tx_tso(q, skb);
+	else
+		err = ionic_tx(q, skb);
+
+	if (err)
+		goto err_out_drop;
+
+	return NETDEV_TX_OK;
+
+err_out_drop:
+	q->drop++;
+	dev_kfree_skb(skb);
+	return NETDEV_TX_OK;
+}
+
 netdev_tx_t ionic_start_xmit(struct sk_buff *skb, struct net_device *netdev)
 {
 	u16 queue_index = skb_get_queue_mapping(skb);
@@ -1026,14 +1219,18 @@ netdev_tx_t ionic_start_xmit(struct sk_buff *skb, struct net_device *netdev)
 	int ndescs;
 	int err;
 
-	if (unlikely(!test_bit(IONIC_LIF_UP, lif->state))) {
+	if (unlikely(!test_bit(IONIC_LIF_F_UP, lif->state))) {
 		dev_kfree_skb(skb);
 		return NETDEV_TX_OK;
 	}
 
-	if (unlikely(!lif_to_txqcq(lif, queue_index)))
+	if (unlikely(skb_shinfo(skb)->tx_flags & SKBTX_HW_TSTAMP))
+		if (lif->hwstamp_txq && lif->phc->ts_config_tx_mode)
+			return ionic_start_hwstamp_xmit(skb, netdev);
+
+	if (unlikely(queue_index >= lif->nxqs))
 		queue_index = 0;
-	q = lif_to_txq(lif, queue_index);
+	q = &lif->txqcqs[queue_index]->q;
 
 	ndescs = ionic_tx_descs_needed(q, skb);
 	if (ndescs < 0)
@@ -1059,7 +1256,6 @@ netdev_tx_t ionic_start_xmit(struct sk_buff *skb, struct net_device *netdev)
 	return NETDEV_TX_OK;
 
 err_out_drop:
-	q->stop++;
 	q->drop++;
 	dev_kfree_skb(skb);
 	return NETDEV_TX_OK;