lilo- mem=exactmap not valid in RH8/RH9?
 

Greetings.

I have one of those old Compaq Proliant 2500 Dual-PPro servers that I'm trying to squeeze a bit more life out of. It has an oddball 448MB RAM (256+128+64) which worked fine under NT4. Since my company no longer supports NT4, and 2K is way too slow on this old machine, I'm trying to get RH9 to work.

Using the mem statements during install got me past the "Not enough RAM to install" and it got up and running, slowly. It still only recognizes 12MB using the "free" command and dmesg recognizes 16MB via the memory map.

I added this line to the lilo.conf file to no avail:

append " mem=exactmap mem=640k@0 mem=447m@1m"

Still only recognized 12-16MB.

So, I tried booting it via command line:

boot: linux mem=exactmap mem=640k@0 mem=447m@1m

That gives me a message "Invalid Number". It will not accept mem=exactmap at all; it expects a number. But simply using mem=640k@0 mem=447m@1m hangs the system before it loads the kernel. Typing mem=448m doesn't work either; it only sees 16M.

Searches of this board and elsewhere always point to the same solution- use the exactmap statememt to define where the memory is, but that simply isn't working because it won't recognize the "exactmap" statement. And without being able to tell it to "Jump" the chipset-enforced memory "hole" (which is the whole problem) leaves me stuck.

Any pointers, or am I just out of luck?

email: datsunaholic@[nospam]hotmail.com, or reply here. Thank you!

lilo?
 

When I read your question I did not see any mention of the lilo command. After you updated /etc/lilo.conf did you run lilo?

Oops.
 

Yeah, that would be what I missed. It seems to work now.

Now, if I only knew how to do this with GRUB. But no matter- it doesn't matter WHICH boot loader it runs with, just that it DOES.

Thank you.