Page 2 of 2
Re: DOS memory managers?
Posted: Wed Sep 03, 2014 1:17 am
by linguofreak
alexfru wrote:linguofreak wrote:
As to your question "Are you sure, that in v86 mode you can access 00000h - 10FFEFh?", the answer is, "if you are on a 286 or later with A20 enabled, you absolutely can".
You mean 80386, of course, as 80286 had no v86 mode.
I meant 286. Yes, he asked about v86 mode, but the as far as the linear address generated by a given segment:offset, real mode and v86 mode are identical.
Re: DOS memory managers?
Posted: Wed Sep 03, 2014 9:24 pm
by mattrix
Try proofreading Intel manuals .... You'll see a lot of inconsistency, typos and omissions.
Speak of the devil, I'm back to looking at the 386 manual.
The only PM prog I've written was a simple goto PM, display "hello world", and back to RM.
V86 is
way more complicated.
Not sure, but I think all the interupts in V86 generate GPF, and I'm not sure how the interupt handler needs to be set up. I think the interupt handler will need to run in PM, so V86 will have to have a TSS?
All these things seemed OK in in theory, but practice is a whole new kettle of fish.
as far as the linear address generated by a given segment:offset, real mode and v86 mode are identical.
I think switching to V86 is a bit more restrictive than real, no limits > 64k etc, or is it the descriptor table entries?
Re: DOS memory managers?
Posted: Wed Sep 03, 2014 11:28 pm
by alexfru
mattrix wrote:Try proofreading Intel manuals .... You'll see a lot of inconsistency, typos and omissions.
Speak of the devil, I'm back to looking at the 386 manual.
That's probably the best manual to date.
V86 is way more complicated.
Not sure, but I think all the interupts in V86 generate GPF, and I'm not sure how the interupt handler needs to be set up. I think the interupt handler will need to run in PM, so V86 will have to have a TSS?
You've got the manual(s) and
my code. Discover the details yourself.
as far as the linear address generated by a given segment:offset, real mode and v86 mode are identical.
I think switching to V86 is a bit more restrictive than real, no limits > 64k etc, or is it the descriptor table entries?
Officially, the limits are exactly 65535 in real mode.
Simulating "big real mode" in v86 is indeed a lot of fun.