xfs
[Top] [All Lists]

Re: 2.4.11-pre2-xfs

To: Seth Mos <knuffie@xxxxxxxxx>
Subject: Re: 2.4.11-pre2-xfs
From: utz lehmann <xfs@xxxxxxxxxx>
Date: Fri, 5 Oct 2001 22:40:34 +0200
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <Pine.BSI.4.10.10110052121220.303-100000@xxxxxxxxxxxxx>
References: <3BBE04A0.64D2A0BA@xxxxxxxxxxx> <Pine.BSI.4.10.10110052121220.303-100000@xxxxxxxxxxxxx>
Sender: owner-linux-xfs@xxxxxxxxxxx
User-agent: Mutt/1.2.5i
Hi Seth

Seth Mos [knuffie@xxxxxxxxx] wrote:
> The dell PE 2500 is ServerWorks LE based, 2GB ram with both 2.4.10 and
> 2.4.11-pre3.
> 
> The only way to fix it is not using HIGHMEM. As soon as I compile without
> HIGHMEM (4GB) the box is stable and does not deadlock or crash even under
> heavy load. I have about a month before the system must go into production
> so if anyone has some hints or tests I could do they are most welcome.
> 
> I can not get it over my heart to tell that we cannot use half the
> memory available. There goes my reputation :-/

Maybe I have a solution for you (and others).

I found a patch (linux-2.4.2-vm-1-2-3-gbyte.patch) in the redhat kernel src
rpm. It allow you to change the standard vm user/kernel split of 3/1 GB to
2/2 and 1/3 GB. Without a HIGHMEM kernel your max available memory is kernel
spilt size - 128MB. 896MB default, and 1920 or 2944MB with the patch.

At work we have athlon based CAE workstations and numbercruncher with 1GB or
1.5GB RAM. They running 2.4.7 or 2.4.9 linux-xfs kernels (of course .-) with
this patch. I modified the patch to 2.75/1.25 resp. 2.25/1.75GB. So they can
use all their memory without HIGHMEM.

Advantages are no performance loss due HIGHMEM and *NO* HIGHMEM trouble.

Disadvantage is that the possible userspace size per process is reduced.
If you don't have big processes it won't be a problem. I saw >1GB sized
processes with the 2.25/1.75GB split.


hope that helps.

utz


<Prev in Thread] Current Thread [Next in Thread>