Recommendations...

Benjamin Scott dragonhawk at gmail.com
Wed Jun 16 10:25:56 EDT 2010


On Wed, Jun 16, 2010 at 9:20 AM, Michael ODonnell
<michael.odonnell at comcast.net> wrote:
>> Processes can potentially indirectly access more than 4 GiB of RAM
>> by using memory windowing/bank swapping/etc.  This would be similar
>> to "Expanded Memory" from the days of the 8086.  Reserve some
>> range of process-addressable memory.  A special library/system
>> call exchanges that block of memory with another block from the
>> not-directly-addressable RAM.
>
> (cringe!)  I hope you're not speaking from recent experience.

  Says the guy who's stuck with a 32-bit system but still wants to use
large memories.  ;-)

  But seriously: Microsoft Windows Server used this technique back
before Microsoft had their 64-bit stuff working well.  They called it
"Address Windowing Extensions".  I am told MS-SQL could make good use
of it on Win Server 2003, which was 32-bit only, but did fully support
PAE.

  I don't know if Linux even supports this kind of thing; the kernel
people's solution may just be "you should use a 64-bit kernel".

> Well, when you've got an installed base of venerable yet cantankerous
> 32bit apps [...] but now that feature you were using has been
> discontinued....  that's one scenario.  Guess how I know.  >-/

  Ick.  You have my sympathies.  I hate legacy stuff.  (Too bad
there's always a lot of it.)

  And it's amazing how many programmers *still* assume a pointer is
always 32 bits wide.

-- Ben



More information about the gnhlug-discuss mailing list