[alt.msdos.programmer] Huge model in Turbo C and Microsoft C

alan@pluto.dss.com (Alan Warwick) (05/07/90)

Does anyone know the specifics of how turbo C and microsoft C deal with huge
data items internally ? The idea of a huge item is that the far pointer to the
data object gets computed on every access to that data item and that the 
pointer is normalized. This is how it can access over 64K of data in one item.
From some initial investigation with TC I noticed that before the references
to the data item calls to PADD@ are made to normalize the pointer to the huge
data item. Does anyone know the complete scoop on how TC and MSC do this ?
My goal is to replace these functions with my own so that I can access really
huge data (say 3MB) while I am running in protected mode. I know that a dos
extender does this kind of thing, but I can't afford one.

						Thanks,

						Alan Wariwck
						alan@doc.dss.com