On Mon, 30 Dec 2002 18:20:04 +0100 (CET), Daniela Engert wrote:
>On Mon, 30 Dec 2002 11:59:04 +0100 (CET), Jimi wrote:
>
>>The limit on devices is 8192 and this would make 8k SUB-Block space per device.
>>The problem with that limit is that I would need to allocate 2048 (!) GDT-selectors,
>because I need to map
the
>>space for Ring-0 *and* Ring-3 and that's the whole problem.
>I strongly recommend to get rid of this GDT mapping idea at all!
>We are no longer in the 80286 age and can take advantage of the 4G flat
>kernel address space. Memory objects in this so called linear address
>space are addressable from any context by means of the DOS32FLATDS
Argh. Never knew that one. Of course, this will make some things definitely more easy
to do.
Especially area handling from 16-bit and 32-bit code. I would have used that "trick"
before, I I would have
know that it's possible under OS/2. "PhysToGDTSel - eCX - Size of segment mapped, must
be equal or less
than 64k". :-((
>So, instead of virtual (GDT based) pointer switch to linear (32-bit)
>ones and *expose* them to clients. If a client is so stupid to require
>a virtual mapping then it can do it itself.
No problem at all. The areas are used by gconfig internally and may get used by other
drivers through gconfig
api. Anyway, now I need to rewrite the BASE-Area stuff & API again. For the 3rd time.
*cough*. At least it's
already hotplugable, so not much to do.
cu, Kiewitz
-----------
To unsubscribe yourself from this list, send the following message
to [EMAIL PROTECTED]
unsubscribe acpi-os2
end
