rework HAL LInux coldplug to work with future kernel changes

[Artem Kachitchkine]

> You can never tell for sure, what's physical and what not. If you run in
> a virtualized environment, which is already pretty common today, the
> kernel will not even tell you what's a real disk and what what isn't.
> For your Firewire case, you could just guess, by matching on subsystem
> == "ieee1394". But the only interesting property is "removable" or
> "fixed" which is totally independent today, from the kind of bus it is
> connected to.

This is all good in theory, but many people are aware of what they are 
connecting. There are multiple connection points, but the device/system 
boundary is what matters for most people. An IDE disk internally 
connects to the IDE/FireWire bridge, put in an enclosure, connects to a 
FireWire port, which internally connects to a FireWire/PCI bridge and so 
on - but from a person's perspective it's a FireWire disk. It has a 
FireWire logo on it and the matching port on the system has the logo. 
And I want my GNOME disk icon to have the same logo.

It is responsibility of the system software to evaluate reliability of 
the information about physical properties. The fact that in some cases 
it _cannot_ reliably determine this, doesn't mean it should hide this 
information when it _can_. VMWare gets out of its way to emulate an IDE 
CDROM - it walks like a duck, fine, it's a duck. I don't see how 
environments _designed_ to lie about physical properties should affect 
our thinking about environments that don't lie.

-Artem.