NCM tries to batch TX packets using timeouts (500us) and does not send packets
before 3 packets are in the submit queue. Timeouts take milliseconds on
dde_linux which leads to delayed ACKs and poor performance for the RX case.
Therefore, we send small packets (<100 Bytes) immediately without batching (it
might be an ACK or last packet of a larger transfer).

diff --git a/drivers/net/usb/cdc_ncm.c b/drivers/net/usb/cdc_ncm.c
index 0897fdb..4dd6047 100644
--- a/drivers/net/usb/cdc_ncm.c
+++ b/drivers/net/usb/cdc_ncm.c
@@ -1354,6 +1354,14 @@ cdc_ncm_fill_tx_frame(struct usbnet *dev, struct sk_buff *skb, __le32 sign)
 		}
 		skb_put_data(skb_out, skb->data, skb->len);
 		ctx->tx_curr_frame_payload += skb->len;	/* count real tx payload data */
+
+		/*
+		 * send small packets immediately (e.g., TCP ACKS), just batch with large
+		 * payloads
+		 */
+		if (skb->len < 100)
+			ready2send = 1;
+
 		dev_kfree_skb_any(skb);
 		skb = NULL;