On Tue, 27 Nov 2012, Oliver Neukum wrote: > On Tuesday 27 November 2012 10:30:02 Alan Stern wrote: > > > I disagree. The usbfs interface is not as capable as the kernel's > > internal API; that has always been true. One of its limitations is the > > inability to request remote wakeups. We could add that to usbfs, but > > for now it isn't there. > > Yes. > > > If that limitation means the buggy modem will crash whenever it is > > being driven by a user program and the system suspends, so be it. We > > As far as the device is under control of usbfs that is a defensible viewpoint. > > > shouldn't expect the kernel to work around hardware bugs when the > > device in question isn't even under the control of a kernel driver. > > That is not a position that is useful. In particular there's necessarily > (if you use a modular kernel) a window where a device is configured > due to the kernel's action, but not yet bound to a driver. We'd crash the > device if we go to a system suspend then. In fact, it would not crash. Daniele said so; the crash occurs only when the modem is attached to a network. If it hasn't been bound to a driver yet, it can't be attached. > That is no good and one more > reason this must be handled in usbcore, not in cdc-acm. Alan Stern -- 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
