Full RAM not recognized on Compaq 2500
I have a Compaq ProLiant 2500 with the following specs:
2x Pentium Pro 200 MHz Processors 640MB Memory Compaq Smart Array Controller - 2DH I first tried installing Redhat 9 and then many other RPM based distros to no avail. None would recognize the full amount of memory and therefore would not install based on a 'not enough mem' related error. I finally went with slackware 10.0 (2.4.26) since it installed fine; however, I later realized that slack does not see the full amount of mem either. Only about 12 megs of RAM is recognized, the OS is mostly using swap =(. I remember trying some kernel memory map parameters on the redhat bootdisc, but I could never get it to see the full RAM. Does anyone have any idea how to get slack to recognize full RAM? Is it a kernel issue, or simply device/chip driver issue with the ProLiant 2500? I plan on recompiling the kernel for this box soon to get SMP support, perhaps this is an issue which could be fixed when making the new kernel? Any help is appreciated, thanks. |
Could you post the output of free -m?
Not to say that I disbelieve you, but it's possible you're just misinterpreting the memory. That said, if it doesn't have shared memory, you can try booting the kernel image with "mem=640" appended. |
Old Compaq Proliants have a known problem with memory detection because of the bios having a memory hole at 15MB.
This requires a lilo.conf switch See http://www.cpqlinux.com/memory.html |
for killerbob; free -m output is as follows:
---------------------------------------------------------------------------------------------- total used free shared buffers cached Mem: 12 12 0 0 1 3 -/+ buffers/cache: 7 4 Swap: 980 22 957 ----------------------------------------------------------------------------------------------- davidsrsb, thanks for the link. I'll have to test out the slack kernel parameter as well, I recall trying a number of the mem=exactmap, mem=(#MB)MB, etc for redhat + mandrake. At least link you sent has all the various Methods in one place. I will post results for refernece. |
Quote:
Regards, dkpw |
Also see the same problem with a C.P.2500 running IPCop firewall. IPCop uses Grub but the isse is the same.
http://www.ipcops.com/index.php?name...5ef6424cdc795f |
heh, I actually use IPCop, but on a generic K6 box, not ProLiant.
alright, now for some good results: I was weary of this working, since I remember trying all the mem paramteres listed in the article on Fedora/Redhat/Mandrake as I said. Using mem=640M@0 didn't work for me on my current slack 10.0 setup, system froze after echoing "....booting the kernel....." However, mem=exactmap mem=640K@0 mem=639M@1M worked like a charm. I appended the parameters to my lilo.conf file and the system boots fine and recognizes all mem. "free" output is now as follows: --------------------------------------------------------------------------------------- total used free shared buffers cached Mem: 644820 63052 581768 0 3980 32020 -/+ buffers/cache: 27052 617768 Swap: 1003672 0 1003672 --------------------------------------------------------------------------------------- As a side note, I did do back and try the same mem parameters on my Fedora 3 bootdisc and they did NOT work, system froze at "...booting the kernel...". Score a big one for Slackware in my book. Thanks for your help and refreshing of the info! |
Had same problem on a Proliant 5000. For what its worth with multiple labels in lilo.conf I incremented a lower but close quess for mem by 1 until it broke. Out of 700M+, gave up after getting 640M, as I remember spaces inside the quotes were important.
image = /boot/bzImage append=" mem=624m@16m " root = /dev/sda1 label = Linux |
Fedora Core 3 on PL2500
For installing Fedora Core 3 onto a Compaq PL2500, refer to:
https://bugzilla.redhat.com/bugzilla....cgi?id=129487 I've just completed an install!!!! |
Quote:
|
All times are GMT -5. The time now is 07:15 AM. |