On Tue, 15 Dec 2009, Oliver Neukum wrote: > Am Montag, 14. Dezember 2009 23:12:06 schrieb Alan Stern: > > On Mon, 14 Dec 2009, Oliver Neukum wrote: > > > > > > Also note that other drivers will reset a device to recover from errors > > > > (usbhid for example). > > > > > > I haven't heard of a switched device with an HID interface. > > > > Eventually one will come along. > > Yes, I have some code to fall back to usb_reset_configuration(). > Should we do so also for storage? I don't think that's such a good idea. For instance, I doubt usb_reset_configuration() will be powerful enough to recover from the errors usb-storage encounters. In fact, even a port reset isn't always powerful enough. Also it doesn't notify drivers about the reset. Really it was meant for use only with non-composite devices, where there's never more than one driver involved. > > > > probably because Windows doesn't use it. So what does Windows do when > > > > such a device gets an error in its mass-storage interface? > > > > > > I don't know. I suspect Windows switches modes in kernel space. > > > > Well, if we don't have any choice then it has to be done. Obviously a > > sysfs attribute is the best way. However I can't help thinking that > > this is only a symptom of something deeper, and it ought to be > > generalized. But it's not clear how... > > The obvious alternative is to put mode switching into the kernel. > If you don't want to do that, there'll be other drawbacks, we are seeing > here. Even if mode switching were in the kernel (and in a few cases it is), there still would be problems. Mode switching often involves changing descriptors. The core interprets these changes as meaning that the device has disconnected and another device was plugged into the port. Face it, mode switching is a bad idea. These devices should have been designed to use multiple configurations instead. Since they weren't, their functionality is severely compromised: It's very difficult to reset them for error recovery. 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
