OSDev.org

I 'borrowed' a snippet of code for testing that the A20 line is correctly enabled. It says that A20 is enabled prior to me actually enabling it.

2 questions.

a) Is it possible that it could be?
b) Does anyone know of a fool-proof method of testing it (16-bit code).

32-bit code would be easy enough.. but I am doing this before protected mode, so 16-bit would be preferred.

Errr.. ok I'm confused.

I thought that processors/motherboards were supposed to disable the A20 line at bootup for backward compatibility.

But I've now figured out that my system has A20 enabled by the time the boot loader is exectured, and the same in Bochs.

If I force A20 disabled, then test my 'Is A20 enabled' code, it says that A20 is disabled, so I know its working.

However, when I boot the bootsector of the floppy with either Bochs or from a reboot, A20 is reported as being enabled unless I force it disabled.

Has anyone seen a notice that processors will no longer be backward compatible at boot?

FYI: I have a Pentium 3 on a Microstar MSI-6309 motherboard.

According to Bochs forum, it does start with A20 enabled.. so thats half the problem sorted out.

it is the BIOS which may enable or leave the a20 gate disabled, if memory serves