(no subject)

> We shouldn't allow attaching to a device that's still
> initlizaing, but ROS currently does because of some 
> device that tries to do this, in the PnP manager or
> early boot-phase drivers (it has an auto-generated name).
> Please fix this!

Init function of that driver (PnpDriverInitializeEmpty() in pnpmgr.c) is 
not called at all because of hack in IopCreateDriver(), line 1206.

> Init function of that driver (PnpDriverInitializeEmpty() in pnpmgr.c) is 
> not called at all because of hack in IopCreateDriver(), line 1206.

Sorry, I was wrong.