I'm desperately trying to understand exactly what successful autosuspend depends on. It seems to me that a common model (observed in cdc_ether and cdc_wdm at least) is - set "needs_remote_wakeup = 1" on open and "needs_remote_wakeup = 0" on close. - use pairs of usb_autopm_get_interface/usb_autopm_put_interface around any device access This works well if the device supports remote wakeup, but when it doesn't then this ends up with a device that only rarely autosuspends. It is of course expected that the device cannot autosuspend between the open and close calls, as needs_remote_wakeup is 1 then. But after close I would expect the device to autosuspend again. It does not. Until some external trigger event happens (e.g. changing the autosuspend timeout). I wonder if the problem might be that the usbcore doesn't notice the changed needs_remote_wakeup? I experimented with wrapping this change between usb_autopm_get_interface and usb_autopm_put_interface and that worked! I.e. status = usb_autopm_get_interface(desc->intf); desc->intf->needs_remote_wakeup = 0; if (!status) usb_autopm_put_interface(desc->intf); causes the device to start autosuspending, while merely doing desc->intf->needs_remote_wakeup = 0; does not. Any comments? Am I totally lost, or should drivers which modify needs_remote_wakeup always wrap the change in usb_autopm_get_interface/ usb_autopm_put_interface to ensure that the change is noticed? Bjørn -- 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
