Not in my experience; modern bioses which support them have a pretty
good (at least in the case of 0e801h and 0e820h) reason to not lie
about them, because windows 95 and NT will puke if they're lied to
in this way, and older bioses, at least the 5 or so I've tested,
have the good sense to return `call not supported' when they're
asked to do these calls.
>If not, it would make sense to try all
>of them and use the maximum value we get
I'd be interested in knowing about modern bioses that support 0e820h
but which don't support 0e801h (820 is a readdir() for a bios memory
map; you walk the map with 0e820h calls, getting the memory map in
painful detail. Since Linux doesn't yet support this, it's a lot of
extra work when 0e801h gets the raw memory sizes up to 4gb.) A lot of
the other calls seem to be supported only sporatically (0c7h doesn't
work on some of the Ppro motherboards I have, and 0e8A doesn't work
on anything I've got) so it may be adding a whole lot of redundant
functionality for only a few malfunctioning bioses.
>; that shouldn't force anybody
>to add "mem=XXX" to their setup.
Well, except people who use Compaq Proliants and Prolineas :-(
____
david parsons \bi/ At times I wonder if Compaq is insane. And then they
\/ plunk a proliant down at my feet and I lose those happy
illusions.