Comment 7 for bug 183109

Ben, thanks for your investigations. I actually see one more functional difference, the new code added setting the speed variable within the baud rate case statement in kobil_set_termios().

- case 9600:
   default:
+ speed = 9600;
+ case 9600:

It looks like the speed variable was never properly initialized within that function - maybe the latest code is broken because the lines Ben pointed out above now _really use_ that variable?
A patched binary kernel module (generic kernel on IX86) for testing would be perfect as I can only test with the Hardy live CD. I would then unload the driver loaded during boot and then load the patched one.
If it's possible to properly build the Hardy kernel on a Gutsy machine, I can do that on my own. I would pull the Hardy kernel source using the live CD and copy the code together with the Hardy kernel configuration from /boot to my Gutsy system, then compile the kernel there. Finally I would unload/load the kernel module on the Hardy Live CD system.