[tuhdo]

In the Volume 3, it is said that:

"In protected mode, the IA-32 architecture provides a normal physical address space of 4 GBytes (2 32 bytes). This is the address space that the processor can address on its address bus. This address space is flat (unsegmented), with addresses ranging continuously from 0 to FFFFFFFFH. This physical address space can be mapped to read- write memory, read-only memory, and memory mapped I/O. The memory mapping facilities described in this chapter can be used to divide this physical memory up into segments and/or pages."

It correlates to my experience of developing in protected mode in QEMU. Once entered protected mode, I can access to any address above 0x10000 without being wrapped around. When I was writing my first kernel (https://github.com/tuhdo/os-study) in real-mode, indeed A20 must be enabled.

[Vendan]

The A20 gate would affect both real and protected mode, so to be truely compliant, you should be opening it even for a protected mode OS. It was originally a purely hardware hack forcing an address line(A20, or the 21st address line) to stay low, thus "wrapping" addresses from 1mb-2mb to 0-1mb, so it would have affected real or protected mode identically. On modern hardware, there is a chance that A20 is (against the accepted spec) default open, or possible not implemented.

[wolfgke]

> On modern hardware, there is a chance that A20 is (against the accepted spec) default open, or possible not implemented.

To nitpick on this a little bit (consider it as a polite supplement):

In Intel® 64 and IA-32 Architectures Software Developer’s Manual Volume 3 (3A, 3B, 3C & 3D): System Programming Guide

in section

8.7.13.4 External Signal Compatibility

one can read (emphasis by me):

"A20M# pin — On an IA-32 processor, the A20M# pin is typically provided for compatibility with the Intel 286 processor. Asserting this pin causes bit 20 of the physical address to be masked (forced to zero) for all external bus memory accesses. Processors supporting Intel Hyper-Threading Technology provide one A20M# pin, which affects the operation of both logical processors within the physical processor.

The functionality of A20M# is used primarily by older operating systems and not used by modern operating systems. On newer Intel 64 processors, A20M# may be absent.".

TLDR: The accepted spec is that A20M# might not exist.

[Vendan]

fair point, I hadn't bothered to look more, but it's still true that A20(when implemented) gates both real and protected modes, and therefore "ignoring it to focus on protected mode" is invalid