Re: Memory between 640K-1M

Jakub Jelinek (
Mon, 18 May 1998 13:27:27 +0200 (MET DST)

> On Mon, 18 May 1998, David S. Miller wrote:
> ->It's been done already and support for this is working just fine on
> ->UltraSparc for example. See the linux-patches archive incoming
> ->category for my diffs which implement this and are scheduled to go in
> ->when Linus gets a chance to look at it.
> Umm, am I missing something obvious? what archive are you talking about?
> It does not seems to be ftp at vger.rutgers/
(this is contains the incoming queue of patches for Linus)

(this is CVS repository, where it is for quite some time already implemented
and tested). Actually, at the moment it is done on sparc32 and sparc64, as
when I was browsing code for all architectures in cvs, I could not find
something about m68k not using non-contiguous memory. It is very simple:
if a full page in mem_map is filled with reserved pages pointing to memory
hole, then that page is freed into the free pages pool. The last page before
that is marked with PG_skip and one of the struct page fields is overloaded
with pointer to next mem_map entry. With that, those few places in the
kernel (I think it was 1 or 2 places in generic code only) that go one
mem_map entry by one can skip it.

Jakub Jelinek | |
Administrator of SunSITE Czech Republic, MFF, Charles University
Ultralinux - first 64bit OS to take full power of the UltraSparc
Linux version 2.1.102 on a sparc64 machine (498.80 BogoMips).

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to