On Mon, 2014-08-11 at 16:22 +0800, Greg KH wrote: > On Mon, Aug 11, 2014 at 09:10:35AM +0200, Oliver Neukum wrote: > > On Thu, 2014-08-07 at 20:31 -0700, Greg KH wrote: > > > On Fri, Aug 08, 2014 at 10:56:57AM +0800, w wrote: > > > > > > > > If use symlinks it can work well. > > > > > > > > But why program in userspace can prevent driver release the node? > > > > The program just open the node, nothing else. > > > > > > If it holds the node open, of course the kernel can't release it, the > > > resource is still in use. That's the way it is supposed to be, don't > > > you agree? > > > > Actually on second thought, why? While the device is open there's an > > obvious reason to not throw away an fd and its associated baggage. > > But why the device node? > > Because it's easier that way for the kernel code :) I doubt it. It is more obvious, but it isn't easier. Generically removing the possibility of an attempt to open a disconnected device or do any operation on it should make device drivers simpler. > > That would be necessary only if you were to use the number as an > > index. > > It shouldn't be, so we can change this in the driver core if it's really > an issue. So far, no one has objected to it. I seem to recall quite some complaints. 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
