xhci: clear root port wake on bits if controller isn't wake-up capable
authorLu Baolu <baolu.lu@linux.intel.com>
Tue, 24 Jun 2014 14:14:43 +0000 (17:14 +0300)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Tue, 24 Jun 2014 16:29:35 +0000 (12:29 -0400)
When xHCI PCI host is suspended, if do_wakeup is false in xhci_pci_suspend,
xhci_bus_suspend needs to clear all root port wake on bits. Otherwise some Intel
platforms may get a spurious wakeup, even if PCI PME# is disabled.

This patch should be back-ported to kernels as old as 2.6.37, that
contains the commit 9777e3ce907d4cb5a513902a87ecd03b52499569
"USB: xHCI: bus power management implementation".

Cc: stable@vger.kernel.org # 2.6.37
Signed-off-by: Lu Baolu <baolu.lu@linux.intel.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/usb/host/xhci-hub.c

index 2b998c6..aa79e87 100644 (file)
@@ -22,6 +22,7 @@
 
 
 #include <linux/slab.h>
+#include <linux/device.h>
 #include <asm/unaligned.h>
 
 #include "xhci.h"
@@ -1139,7 +1140,9 @@ int xhci_bus_suspend(struct usb_hcd *hcd)
                 * including the USB 3.0 roothub, but only if CONFIG_PM_RUNTIME
                 * is enabled, so also enable remote wake here.
                 */
-               if (hcd->self.root_hub->do_remote_wakeup) {
+               if (hcd->self.root_hub->do_remote_wakeup
+                               && device_may_wakeup(hcd->self.controller)) {
+
                        if (t1 & PORT_CONNECT) {
                                t2 |= PORT_WKOC_E | PORT_WKDISC_E;
                                t2 &= ~PORT_WKCONN_E;