Mail Archives: djgpp/2000/10/12/14:31:07
"Graham Reeds" <grahamr AT dtn DOT ntl DOT com> wrote in message
news:JAlF5.31810$uq5 DOT 615829 AT news6-win DOT server DOT ntlworld DOT com...
<snip>
> I thought that it was similar to the _farpokex() functions, in that it
> disables memory protection for a small amount of time. I also thought I
was
> on the right track since it doesn't crash when ran (unlike the other
options
> I have tried.)
It doesn't crash because it doesn't violate the access rights of _dos_ds.
That doesn't mean you did it correctly. The dosmem*() functions are for
accessing the real mode address space only. That's fine for VGA, but if
you're using a VESA LFB it's no good. The _farpeek/poke functions are more
like movedata (in fact I think movedata() may even use those functions).
None of these functions disable memory protection in any way. You seem to
have some confusion about this, so I highly advise you to read chapter 18 of
the djgpp faq, especially 18.7.
<snip>
> _movedatal(_my_ds(), primarysurface->dsimage, videoselector, videomemory,
> remainder);
This looks wrong to me. The fourth parameter is suposed to be the offset
within videoselector, and instead you're passing the linear address that you
got from __dpmi_allocate_dos_memory(). This is definately wrong. You
probably meant to pass 0, since you probably want to copy data starting at
the beginning of the LFB, right? It's not suprising that you get a GPF from
this mistake, since you violate the access rights of videoselector (by
exceeding its limit boundary).
> Casting (-I think that is the right term) the primarysurface->dsimage with
> just an asterix removes the warning but it still gpf's out when ran:
It's called dereferencing the pointer, and that's not what you want to do.
You want to pass the pointer itself, since it's the offset within the ds
selector (which is what movedata needs). If you want to get rid of the
warning, typecast it to unsigned int. The typecast won't actually change the
value passed, but it will tell the compiler that is what you wanted to do
and get rid of the warning. The crashing was caused by the mistake you made
with the fourth parameter.
Hope this helps.
Marp
- Raw text -