[PATCH] r8169: IRQ races during change of mtu

IRQ races during change of mtu
- NAPI poll must be enabled prior to IRQ activation or the IRQ handler
  will not know what to do with an incoming packet;
- rtl8169_down() needs to try twice to sync with the IRQ handler when
  it is not issued under !netif_running() protection.

Both changes make it safe to request a change of mtu on a live device.
Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: Jeff Garzik <jgarzik@pobox.com>

drivers/net/r8169.c

@@ -1648,10 +1648,10 @@ static int rtl8169_change_mtu(struct net_device *dev, int new_mtu)
 	if (ret < 0)
 		goto out;
 
-	rtl8169_hw_start(dev);
-
 	netif_poll_enable(dev);
 
+	rtl8169_hw_start(dev);
+
 	rtl8169_request_timer(dev);
 
 out:
@@ -2352,6 +2352,7 @@ static void rtl8169_down(struct net_device *dev)
 {
 	struct rtl8169_private *tp = netdev_priv(dev);
 	void __iomem *ioaddr = tp->mmio_addr;
+	unsigned int poll_locked = 0;
 
 	rtl8169_delete_timer(dev);
 
@@ -2359,6 +2360,7 @@ static void rtl8169_down(struct net_device *dev)
 
 	flush_scheduled_work();
 
+core_down:
 	spin_lock_irq(&tp->lock);
 
 	/* Stop the chip's Tx and Rx DMA processes. */
@@ -2375,11 +2377,28 @@ static void rtl8169_down(struct net_device *dev)
 
 	synchronize_irq(dev->irq);
 
-	netif_poll_disable(dev);
+	if (!poll_locked) {
+		netif_poll_disable(dev);
+		poll_locked++;
+	}
 
 	/* Give a racing hard_start_xmit a few cycles to complete. */
 	synchronize_kernel();
 
+	/*
+	 * And now for the 50k$ question: are IRQ disabled or not ?
+	 *
+	 * Two paths lead here:
+	 * 1) dev->close
+	 *    -> netif_running() is available to sync the current code and the
+	 *       IRQ handler. See rtl8169_interrupt for details.
+	 * 2) dev->change_mtu
+	 *    -> rtl8169_poll can not be issued again and re-enable the
+	 *       interruptions. Let's simply issue the IRQ down sequence again.
+	 */
+	if (RTL_R16(IntrMask))
+		goto core_down;
+
 	rtl8169_tx_clear(tp);
 
 	rtl8169_rx_clear(tp);