return count;
 }
 
+static void hip04_start_tx_timer(struct hip04_priv *priv)
+{
+       unsigned long ns = priv->tx_coalesce_usecs * NSEC_PER_USEC / 2;
+
+       /* allow timer to fire after half the time at the earliest */
+       hrtimer_start_range_ns(&priv->tx_coalesce_timer, ns_to_ktime(ns),
+                              ns, HRTIMER_MODE_REL);
+}
+
 static int hip04_mac_start_xmit(struct sk_buff *skb, struct net_device *ndev)
 {
        struct hip04_priv *priv = netdev_priv(ndev);
                }
        } else if (!hrtimer_is_queued(&priv->tx_coalesce_timer)) {
                /* cleanup not pending yet, start a new timer */
-               hrtimer_start_expires(&priv->tx_coalesce_timer,
-                                     HRTIMER_MODE_REL);
+               hip04_start_tx_timer(priv);
        }
 
        return NETDEV_TX_OK;
        /* clean up tx descriptors and start a new timer if necessary */
        tx_remaining = hip04_tx_reclaim(ndev, false);
        if (rx < budget && tx_remaining)
-               hrtimer_start_expires(&priv->tx_coalesce_timer, HRTIMER_MODE_REL);
+               hip04_start_tx_timer(priv);
 
        return rx;
 }
        struct hip04_priv *priv;
        struct resource *res;
        unsigned int irq;
-       ktime_t txtime;
        int ret;
 
        ndev = alloc_etherdev(sizeof(struct hip04_priv));
         */
        priv->tx_coalesce_frames = TX_DESC_NUM * 3 / 4;
        priv->tx_coalesce_usecs = 200;
-       /* allow timer to fire after half the time at the earliest */
-       txtime = ktime_set(0, priv->tx_coalesce_usecs * NSEC_PER_USEC / 2);
-       hrtimer_set_expires_range(&priv->tx_coalesce_timer, txtime, txtime);
        priv->tx_coalesce_timer.function = tx_done;
 
        priv->map = syscon_node_to_regmap(arg.np);