DS>Perhaps I am just showing my inexperience with nonlinear
DS>addressing, 3 years of C cobwebs and rust, or just plain
DS>being naive... but shouldn't the following two pieces of
DS>code generate the same result ?  The first one (based on
DS>the example you posted) works, the second (very similar to
DS>the MAKEP macro) does not.

DS>  ptr1616 = (void * _Seg16) ( phys.asel[0] << 16 );
DS>  VGARAMptr = (void *) ptr1616;

DS>  VGARAMptr = (void *) ((void * _Seg16) ( phys.asel[0] << 16 ));

I'm suspecting that the first code generates a thunk, and that the
second one the (void*) overrides the (void * _Seg16) before the thunk is
generated.  Just a thought.  It's stuff like this that drives me mad
when I have to do anything unusual in C.  You might try looking at an
assembly version of the two pieces of code to see what the compiler
thinks the difference is.

Rob.
___
 X SLMR 2.1a X MS: Where do you want to go uhhh... some time next year?

--- Maximus/2 2.01wb