Hi.
I have been using usbip for a while and think it is a nice and useful
tool. However, there is one thing that annoys me quite much, namely the
fact that the devices are identified by their USB bus IDs even at the
protocol level.
In practice, this means that when the hardware configuration of my usbip
server changes, I need to reconfigure the clients with the new bus IDs.
Any decent web server provides means for mapping URL paths to real file
system paths, such that I can freely relocate content in my file system
while keeping the existing URLs intact. I think usbip should follow the
same pattern. Specifically, the bind command should accept a new argument
specifying an arbitrary name for the exported device. On the client side,
the attach command would be run with this name as argument rather than the
bus ID.
This idea could be extended by allowing multiple devices to be exported
under the same name. Doing this would create a pool of devices, and a
client trying to import a device using the pool's name would be assigned a
free device from the pool. This would be very useful if you happen to own
expensive USB devices (such as copy protection dongles) which you would
like to utilize efficiently across a geographically distributed
organization.
What do you think of this idea? Would it make sense to develop usbip into
this direction, or should problems of this kind be addressed at higher
level?
--
Kaarle Ritvanen
Data King Ltd.
--
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
