Am Donnerstag, 26. November 2009 18:22:02 schrieb Alan Stern:
> On Wed, 25 Nov 2009, Oliver Neukum wrote:
> > Yes, all done. As remote wakeup is optional I always want to support
> > devices that don't do it as far as possible.
>
> Such changes don't take effect until the next
> usb_autopm_put_interface(). This might be an issue during close. Is
> there always a usb_autopm_put_interface() call after you set
> intf->needs_remote_wakeup to 0?
Yes, as I currently do in such cases exactly for this reason:
usb_autopm_get_interface(intf);
intf->needs_remote_wakeup = 0;
usb_autopm_put_interface(intf);
> If not, you can be most efficient by doing something like this:
>
> if (!disconnected) {
> intf->needs_remote_wakeup = 0;
> if (currently autosuspended)
How do I do that without a race?
> usb_autopm_get_interface_no_resume(intf);
> usb_autopm_put_interface(intf);
> }
>
> This way if the device is already suspended, it won't be woken up again
> unnecessarily.
This very much argues for having an API call for requesting
remote wakeup.
> By the way, after the upcoming reimplementation of USB runtime PM,
> drivers will have to make sure their autopm_get and autopm_put calls
> balance -- even if this means doing an autopm_put after the disconnect
> routine has returned. You'll want to audit a bunch drivers and fix
> them up.
Yes, I am sure you'll warn me.
Regards
Oliver
--
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
