Re: HIGH MEMORY access

From: david parsons (orc@pell.portland.or.us)
Date: Fri Mar 03 2000 - 14:15:44 EST


> > > > >> From: Manfred Spraul [mailto:manfreds@colorfullife.com]

> > > > >> * max_pfn is set to the end if the last usable block.
> > > > >> * then the ACPI tables follow --> they are behind max_pfn -->
> > > > >> behind the
> > > > >> end of mem_map.

> > > > <nathan.zook@amd.com> wrote:
> > > > >This is changing. The new patches will be reclaim-enabled, so
> > > > >those pages will be mapped, and the max_pfn-class variables will
> > > > >reflect their presence.

> > > > From: orc@pell.portland.or.us [mailto:orc@pell.portland.or.us]
> > > > Only if these pages are set as acpi reclaim memory;
> > > > if they're marked as nvs and happen to be above the last
> > > > ram section, max_pfn won't catch them.

> > nathan.zook@amd.com wrote:

> > > Marking them NVS is an error, but I've been leaning towards
> > > mapping the NVS pages anyway.

> > From: david parsons [mailto:orc@pell.portland.or.us]
> > Though if you find a bios that stuffs the nvs pages up near the
> > top of 4gb, won't that make mem_map[] very large and thus
> > potentially take all memory away from low-memory systems?

nathan.zook@amd.com wrote:
>
> Oh the joys of an inadequate spec! Sounds like it might be time for Smart
> Reclaim(tm). Oh well. It will get worse.

    I suspect that mapping roms and acpi memory into the regular memory
    segment might not be the most optimal way of doing things. The PCI
    code, once upon a time (and still?) set up their own memory segment
    to get at pci stuff that was in awkward places; this might be a
    good solution to mismatches between Linux's memory model and the way
    vendors might scatter things around in memory.

> And those Award bioses are the ones that dropped e801.

    Quite possibly. I'm pretty some of the older award bioses still did
    e801, but since I wrote the fancy memory detection code e801 and 88
    could both set the motherboard on fire and I'd never know until the
    fire department arrives.

                  ____
    david parsons \bi/ I'm not expecting any of the new Award bioses are likely
                   \/ to ever spontaneously backport to any of my 366s, either.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Tue Mar 07 2000 - 21:00:15 EST