diff --git a/drivers/net/ethernet/cadence/macb.h b/drivers/net/ethernet/cadence/macb.h
index 8ef7995db60c6..83838c03d9734 100644
--- a/drivers/net/ethernet/cadence/macb.h
+++ b/drivers/net/ethernet/cadence/macb.h
@@ -996,6 +996,8 @@ struct macb_dma_desc_ptp {
  *       of the frame
  * @mapping: DMA address of the skb's fragment buffer
  * @size: size of the DMA mapped buffer
+ * @skb_len: skb->len as handed over by the stack, before padding and
+ *           software FCS, only set for the last buffer of the frame
  * @mapped_as_page: true when buffer was mapped with skb_frag_dma_map(),
  *                  false when buffer was mapped with dma_map_single()
  */
@@ -1003,6 +1005,7 @@ struct macb_tx_skb {
 	struct sk_buff		*skb;
 	dma_addr_t		mapping;
 	size_t			size;
+	unsigned int		skb_len;
 	bool			mapped_as_page;
 };
 
diff --git a/drivers/net/ethernet/cadence/macb_main.c b/drivers/net/ethernet/cadence/macb_main.c
index b6e824531081e..31e003a937211 100644
--- a/drivers/net/ethernet/cadence/macb_main.c
+++ b/drivers/net/ethernet/cadence/macb_main.c
@@ -1355,8 +1355,8 @@ static void macb_tx_error_task(struct work_struct *work)
 				bp->dev->stats.tx_packets++;
 				queue->stats.tx_packets++;
 				packets++;
-				bp->dev->stats.tx_bytes += skb->len;
-				queue->stats.tx_bytes += skb->len;
+				bp->dev->stats.tx_bytes += tx_skb->skb_len;
+				queue->stats.tx_bytes += tx_skb->skb_len;
 				bytes += skb->len;
 			}
 		} else {
@@ -1483,8 +1483,8 @@ static int macb_tx_complete(struct macb_queue *queue, int budget)
 					    skb->data);
 				bp->dev->stats.tx_packets++;
 				queue->stats.tx_packets++;
-				bp->dev->stats.tx_bytes += skb->len;
-				queue->stats.tx_bytes += skb->len;
+				bp->dev->stats.tx_bytes += tx_skb->skb_len;
+				queue->stats.tx_bytes += tx_skb->skb_len;
 				packets++;
 				bytes += skb->len;
 			}
@@ -2262,7 +2262,8 @@ static void macb_poll_controller(struct net_device *dev)
 static unsigned int macb_tx_map(struct macb *bp,
 				struct macb_queue *queue,
 				struct sk_buff *skb,
-				unsigned int hdrlen)
+				unsigned int hdrlen,
+				unsigned int skb_len)
 {
 	dma_addr_t mapping;
 	unsigned int len, entry, i, tx_head = queue->tx_head;
@@ -2351,6 +2352,7 @@ static unsigned int macb_tx_map(struct macb *bp,
 
 	/* This is the last buffer of the frame: save socket buffer */
 	tx_skb->skb = skb;
+	tx_skb->skb_len = skb_len;
 
 	/* Update TX ring: update buffer descriptors in reverse order
 	 * to avoid race condition
@@ -2474,6 +2476,13 @@ static inline int macb_clear_csum(struct sk_buff *skb)
 	if (skb->ip_summed != CHECKSUM_PARTIAL)
 		return 0;
 
+	/* The hardware lacks the RFC 768 substitution of 0xffff for a
+	 * UDP checksum that computes to zero, complete UDP in software.
+	 */
+	if (skb->csum_offset == offsetof(struct udphdr, check) &&
+	    !ptp_one_step_sync(skb))
+		return skb_checksum_help(skb);
+
 	/* make sure we can modify the header */
 	if (unlikely(skb_cow_head(skb, 0)))
 		return -1;
@@ -2543,7 +2552,7 @@ static netdev_tx_t macb_start_xmit(struct sk_buff *skb, struct net_device *dev)
 	struct macb *bp = netdev_priv(dev);
 	struct macb_queue *queue = &bp->queues[queue_index];
 	unsigned int desc_cnt, nr_frags, frag_size, f;
-	unsigned int hdrlen;
+	unsigned int hdrlen, skb_len;
 	unsigned long flags;
 	bool is_lso;
 	netdev_tx_t ret = NETDEV_TX_OK;
@@ -2553,6 +2562,7 @@ static netdev_tx_t macb_start_xmit(struct sk_buff *skb, struct net_device *dev)
 		return ret;
 	}
 
+	skb_len = skb->len;
 	if (macb_pad_and_fcs(&skb, dev)) {
 		dev_kfree_skb_any(skb);
 		return ret;
@@ -2618,7 +2628,7 @@ static netdev_tx_t macb_start_xmit(struct sk_buff *skb, struct net_device *dev)
 	}
 
 	/* Map socket buffer for DMA transfer */
-	if (!macb_tx_map(bp, queue, skb, hdrlen)) {
+	if (!macb_tx_map(bp, queue, skb, hdrlen, skb_len)) {
 		dev_kfree_skb_any(skb);
 		goto unlock;
 	}