> > When "struct usb_device" gets suspend/resume methods, > > wouldn't such a device just turn its suspend() into a disconnect > > so it gets fully re-enumerated on a resume()? > > Firstly, with current device names you must not do so or you could > find your devices rearranged on resume.
While there are USB configurations that don't have to guard against that, I think the ones that DO have to guard against it are so typical that it's not worth writing apps to expect otherwise. For example, I seem to recall usb-storage guards against it by remembering drives after they go off-line. > Secondly, even if you did so, it wouldn't matter, as the device still > exists as far as the block layer is concerned. If the block layer doesn't understand the notion of devices going away, that'd seem like something worth fixing in 2.5 then. With hotpluggable external drives, it's common that they'll go away. - Dave _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
