Do you know any one of those BIOS interrupts wich switches to the Protected Mode?

I guess this is only a problem with some very rare PCI-cards and the most bios routines of all other x86-PCs do not switch to the Protected Mode (after the power on self test of the Bios is done).

So Micosoft uses the BIG-Realmode in one of their "himem.sys" and a lot of DOS-games of the 90" too. But how can it possible, if the BIOS switch to the Protected Mode and back to the Real Mode while running an application that uses the BIG-Realmode?

I reckon this problem was higher conventionalized as it is and the background for this was only for pushing of the comming Windows 95 OS with a comercial lie that more memory can not be access and it is only a bug of the CPU, if the CPU gives the possibility for to access the address above the first MB with th Big-Realmode. So in the intel documentation of their CPUs they give us the garantee that the descriptor cache will be cleared with switching back to the Real Mode. But we known the Intel x86-CPUs do not clear the descriptors cache contrary to their documentation.

Dirk

the only answer possible:

test it

@freecrac: Found a workaround!
In one of the really old versions of FASM, after switching to big real mode FASM modifies the GPF handler which is caused when the application tries to access an offset greater than 0xFFFF in real mode. The GPF handler simply enables Unreal Mode.