Re: [PATCH V3] input: Fix USB autosuspend on bcm5974

On Wed, Oct 12, 2011 at 09:03:00AM +0200, Oliver Neukum wrote:

Am Dienstag, 11. Oktober 2011, 18:43:55 schrieb Alan Stern:

quoted

On Tue, 11 Oct 2011, Dmitry Torokhov wrote:

quoted

quoted

Hmm, reading Alan's comment I can see why wakeup might be beneficial
when system is in the sleep state, but while it is running and there is
no driver or driver wants to disable wakeups I think we should
accommodate it.

Currently, wakeup is enabled for USB devices during autosuspend if and
only if one or more of the interface drivers has set the
intf->needs_remote_wakeup flag.  Therefore all that usbhid should need 
to do is clear that flag when the device file is closed.

Now I am confused because Matthew reported that his device generated
remote wakeups without setting needs_remote_wakeup at all.
Matthew, could you clarify?

I /think/ Alan's wrong here - do_remote_wakeup is set to
device_may_wakeup(), which will be true if the device can generate 
remote wakeups even if nothing's asked for them. I'm basing this on 
autosuspend_check() and choose_wakeup(). Alan, am I misinterpreting this 
code?

-- 
Matthew Garrett | mjg59@srcf.ucam.org