This just makes UHCI stop using the HCD glue
wakeup flags.
- Dave
This makes UHCI stop using the HCD glue wakeup flags to report whether
the controller can wake the system. The existing code was wrong anyway;
having a PCI PM capability doesn't imply it reports PME# is supported.
I skimmed Intel's ICH7 datasheet and that basically says the wakeup
signaling gets routed only through ACPI registers. (On the other hand,
many VIA chips provide the PCI PM capabilities...) I think that doing
this correctly with UHCI is going to require the ACPI folk to associate
the /proc/acpi/wakeup identifiers (and wakeup enable/disable flags)
with the relevant /sys/devices/pci*/... devices.
Index: g26/drivers/usb/host/uhci-hcd.c
===================================================================
--- g26.orig/drivers/usb/host/uhci-hcd.c 2005-11-06 11:30:01.000000000 -0800
+++ g26/drivers/usb/host/uhci-hcd.c 2005-11-06 11:30:33.000000000 -0800
@@ -478,8 +478,6 @@ static int uhci_start(struct usb_hcd *hc
struct dentry *dentry;
hcd->uses_new_polling = 1;
- if (pci_find_capability(to_pci_dev(uhci_dev(uhci)), PCI_CAP_ID_PM))
- hcd->can_wakeup = 1; /* Assume it supports PME# */
dentry = debugfs_create_file(hcd->self.bus_name,
S_IFREG|S_IRUGO|S_IWUSR, uhci_debugfs_root, uhci,
