Re: x86 physical memory probe?

• To: Alastair Reid <reid@cs.utah.edu>
• Subject: Re: x86 physical memory probe?
• From: "Jonathan S. Shapiro" <shap@cs.jhu.edu>
• Date: Mon, 27 Nov 2000 17:21:54 -0500
• CC: "Jonathan S. Shapiro" <shap@eros-os.org>, oskit-users@fast.cs.utah.edu
• Organization: Johns Hopkins University
• References: <NEBBJFHFADKEOJFBOIEJEEFHCLAA.reid@cs.utah.edu>
• Sender: owner-oskit-users@fast.cs.utah.edu

I was aware that multiboot reported the total amount of physical memory
(less what it steals, of course). I would have expected it to start the
kernel unmapped, as it is difficult for real-mode kernels to get back to
the unmapped state. I confess I haven't looked at the multiboot spec in
a very long time.

When reporting memory regions, does multiboot also report distinct
regions for areas that are differently reachable by DMA? It seems to me
that if multiboot knows enough to discover discontiguous physical
memory, then it probably doesn't involve much greater machine dependency
to report (e.g.) the 1M, 16M, and "rest of memory" regions as distinct
on a 32M PC (there are DMA channels that only reach the first 1M, and
likewise some that only reach the 16M region).


Jonathan

• RE: x86 physical memory probe?
• From: "Alastair Reid" <reid@cs.utah.edu>

• Prev: RE: x86 physical memory probe?
• Next: relocating kernel to start at 0x00000000
• Index(es):
  • Main
  • Thread