Re: [RFC] Delay EHCI initialization until khubd is idle

[Alan Stern <stern@xxxxxxxxxxxxxxxxxxx>]

On Fri, 5 Sep 2008, Frans Pop wrote:

> The "dead keyboard" issue discussed in the other thread does IMO 
> demonstrate that there is a real risk of issues because of the current 
> lack of coordination between drivers which makes the outcome sensitive to 
> miniscule timing differences and races.

The "dead keyboard" problem is only tangentially related to EHCI
initialization.  Below is a patch for 2.6.27-rc5 which removes the
transient-disconnect handling that caused the "dead keyboard" problem.  
You might want to test it; even without the "delay EHCI init" patch it
should eliminate any more dead keyboards.

Alan Stern



Index: usb-2.6/drivers/usb/core/hub.c
===================================================================
--- usb-2.6.orig/drivers/usb/core/hub.c
+++ usb-2.6/drivers/usb/core/hub.c
@@ -2683,35 +2683,17 @@ static void hub_port_connect_change(stru
 				USB_PORT_STAT_C_ENABLE);
 #endif
 
-	/* Try to use the debounce delay for protection against
-	 * port-enable changes caused, for example, by EMI.
-	 */
-	if (portchange & (USB_PORT_STAT_C_CONNECTION |
-				USB_PORT_STAT_C_ENABLE)) {
-		status = hub_port_debounce(hub, port1);
-		if (status < 0) {
-			if (printk_ratelimit())
-				dev_err (hub_dev, "connect-debounce failed, "
-						"port %d disabled\n", port1);
-			portstatus &= ~USB_PORT_STAT_CONNECTION;
-		} else {
-			portstatus = status;
-		}
-	}
-
 	/* Try to resuscitate an existing device */
 	udev = hdev->children[port1-1];
 	if ((portstatus & USB_PORT_STAT_CONNECTION) && udev &&
 			udev->state != USB_STATE_NOTATTACHED) {
-
 		usb_lock_device(udev);
 		if (portstatus & USB_PORT_STAT_ENABLE) {
 			status = 0;		/* Nothing to do */
-		} else if (!udev->persist_enabled) {
-			status = -ENODEV;	/* Mustn't resuscitate */
 
 #ifdef CONFIG_USB_SUSPEND
-		} else if (udev->state == USB_STATE_SUSPENDED) {
+		} else if (udev->state == USB_STATE_SUSPENDED &&
+				udev->persist_enabled) {
 			/* For a suspended device, treat this as a
 			 * remote wakeup event.
 			 */
@@ -2726,7 +2708,7 @@ static void hub_port_connect_change(stru
 #endif
 
 		} else {
-			status = usb_reset_device(udev);
+			status = -ENODEV;	/* Don't resuscitate */
 		}
 		usb_unlock_device(udev);
 
@@ -2741,6 +2723,19 @@ static void hub_port_connect_change(stru
 		usb_disconnect(&hdev->children[port1-1]);
 	clear_bit(port1, hub->change_bits);
 
+	if (portchange & (USB_PORT_STAT_C_CONNECTION |
+				USB_PORT_STAT_C_ENABLE)) {
+		status = hub_port_debounce(hub, port1);
+		if (status < 0) {
+			if (printk_ratelimit())
+				dev_err (hub_dev, "connect-debounce failed, "
+						"port %d disabled\n", port1);
+			portstatus &= ~USB_PORT_STAT_CONNECTION;
+		} else {
+			portstatus = status;
+		}
+	}
+
 	/* Return now if debouncing failed or nothing is connected */
 	if (!(portstatus & USB_PORT_STAT_CONNECTION)) {
 
@@ -2748,7 +2743,7 @@ static void hub_port_connect_change(stru
 		if ((wHubCharacteristics & HUB_CHAR_LPSM) < 2
 				&& !(portstatus & (1 << USB_PORT_FEAT_POWER)))
 			set_port_feature(hdev, port1, USB_PORT_FEAT_POWER);
- 
+
 		if (portstatus & USB_PORT_STAT_ENABLE)
   			goto done;
 		return;

--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html