Re: [BISECTED] Suspend regression in v2.6.31 with Lenovo 3000 v200

[Alan Stern]

On Thu, 24 Jun 2010, Ozan [utf-8] Ã?aÄ?layan wrote:

> Ok so it should be something like this:
> 
> --- linux-2.6.31.orig/drivers/usb/core/hub.c
> +++ linux-2.6.31/drivers/usb/core/hub.c
> @@ -1305,11 +1305,11 @@ void usb_set_device_state(struct usb_dev
>  ·  ·   ·   ·   ·   || new_state == USB_STATE_SUSPENDED)
>  ·  ·   ·   ·   ;·  /* No change to wakeup settings */
>  ·  ·   ·   else if (new_state == USB_STATE_CONFIGURED)
> -·  ·   ·   ·   device_init_wakeup(&udev->dev,
> +·  ·   ·   ·   device_set_wakeup_capable(&udev->dev,
>  ·  ·   ·   ·   ·   (udev->actconfig->desc.bmAttributes
>  ·  ·   ·   ·   ·    & USB_CONFIG_ATT_WAKEUP));
>  ·  ·   ·   else
> -·  ·   ·   ·   device_init_wakeup(&udev->dev, 0);
> +·  ·   ·   ·   device_set_wakeup_capable(&udev->dev, 0);
>  ·  ·   }
>  ·  ·   if (udev->state == USB_STATE_SUSPENDED &&
>  ·  ·   ·   new_state != USB_STATE_SUSPENDED)

Right.

> > You can verify the result by checking the contents of
> > /sys/bus/usb/devices/2-5/power/wakeup (replace the 2-5 with the correct
> > path for the camera device).  With the current code it will say
> > "enabled", but after the change it should say "disabled".
> > 
> 
> I'll make them try this kernel but when I compare the output of
> 
> cat /sys/bus/usb/devices/*/power/wakeup
> 
> between the patched and non-patched kernels on my system, all wakeup values 
> are still enabled. Because as far as I understand the first call still can 
> enable remote wakeup as the 2nd parameter isn't a hardcoded 0. Or should this 
> only affect some devices?

The content of the power/wakeup file depends on two values: the 
can_wakeup flag and the should_wakeup flag.  If can_wakeup is false 
then the file will be empty; otherwise it will contain "disabled" or 
"enabled" according to the value of should_wakeup.

Now, device_init_wakeup() sets both flags whereas 
device_set_wakeup_capable() affects only the can_wakeup flag.  After 
this change is made there should be nothing in the kernel that turns on 
the should_wakeup flag, and consequently remote wakeup should remain 
disabled.

You can check this by grepping for occurrences of "should_wakeup" or 
"dev_set_wakeup_enable" (which affects the should_wakeup flag).

Alan Stern

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