>>>     struct wsdisplay_softc {
>>>         struct device sc_dv;  /* <- HERE! */
>> There are some who consider that a bug, too...

(Like me - I'd love to see that botch die.)

> Why is is a bug?

Well, strictly, it's not truly a bug.  But I think it's an interface
botch - a design bug, if you will.  If nothing else, it means breaking
LKM binary compatability every time struct device grows, even if the
LKMs don't care about the new elements.

> In FreeBSD there is no such structure nesting and as a result you
> have to do ugly things to do simple things like get a device's name
> etc.  What would be better?

What would be better?  Try this:  struct device has a void * pointer to
the softc; the softc begins with not a struct device but with a struct
device *.  Instead of the pointer cast you do now (under NetBSD), you
do a member dereference and pointer follow.

It means that some operations involve one more memory reference.  In my
opinion, the cleaner code is worth it, even if it does mean a flag day
for all drivers - and if you really can't afford that memory reference,
just do it once and cache the pointer.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B