On Tue, 31 Dec 2002 18:14:50 +0100 (CET), Daniela Engert wrote:
>>>Three times NO! You don't need such kind of things. And if you insist
>>>on a GDT selector, why don't you setup a single big one?
>>Because AllocGDTSel is limited to 64k by IBM :-)
>>Three times NO to IBM ;-)
>Nobody stops you from setting up a GDT entry by your own. The lack of
>an appropriate device helper doesn't mean you can't do that. Some years
>ago I inserted two call gates into the GDT to implement high speed
>access to some driver routines which were meant to be called from user
>space with as little latency as possible.
Hmm, nearly the whole area code is done, so I think I will stick to my previous
method. I would need to
redesign most of the internal APIs, but I wouldn't have to allocate those 256 GDT
selectors for the SUB-Area
anymore, still I would need to link multiple e.g. 64k segments together, because the
areas are supposed to be
able to dynamically enlarge and writing "duplicate" routines for use in 16 and 32-bit
mode is needed in any
case as well.
Also because most GCONFIG internals are pure assembly, a flat-selector could make
things worse. If I mistype
something now and a bad offset is used somewhere, it will result in either corrupted
area/data-segment or
exception. By using a flat ring-0 selector, I may accidentaly corrupt some other data,
also I wouldn't be able to
use e.g. carry flag on 64k overflow trick. Instead I would have to use compares and an
additional register (a
shame in search routines).
cu, Kiewitz
-----------
To unsubscribe yourself from this list, send the following message
to [EMAIL PROTECTED]
unsubscribe acpi-os2
end
