Death of a computer? was Re: [BL] More fun with Xvesa and trident card
keesan at sdf.lonestar.org
Sat Oct 30 10:12:56 EDT 2004
On Sat, 30 Oct 2004 3aoo-cvfd at dea.spamcon.org wrote:
> sindi keesan wrote:
>> This computer does not appear to be very good for
>> linux but DOS works fine. Why?
> Linux exercises the hardware more. It is quite common
> to see DOS running happily on a system with a dud RAM
> chip (What me worry?). But on the same system, Linux
> crashes immediately.
This is not crashing immediately, but it is under certain circumstances.
linux unzip in BL2 and startx with an extra task in BL3.
I checked out the Fur Elise for DFI (Diamond Flower Industries(?) and
Damage Free Intelligence) and it only had to do with the cpu or power
Since the problem is not occurring in DOS, it is probably not in the first
64MB if it is a memory problem (tho I rarely use that much RAM in DOS and
ought to test it on rotating a large image file), so I booted both BL2 and
BL3 with the switch mem=16m and both problems are fixed, therefore it is a
memory problem. I have downloaded the DOS version of memtest and the
linux source code for it and will attempt to compile for linux eventually.
If the first SIMM is bad in its second 3/4 the 16m test would still have
let linux work properly so another option is to test the four SIMMs in
pairs until I can recreate these completely reproducible errors.
>> Perhaps this is a bug affecting this model motherboard
> Perhaps. There are fixes in the 2.2.16 kernel sourcecode
> for certain motherboards. To see which motherboards, go
> to a Slackware 7.1 site (or CDrom) and search the following
> file for "motherboard"
> To see which of those fixes has been included in the
> BL3 (or BL2) kernel, read the kernel config file at
> the BasicLinux site. To see which fixes are in the
> bare.i kernel, read the config file that accompanies
> that kernel.
The 'Not a kernel image' problem appears to be motherboard specific but
also intermittent. It occurs in both BL3 and BL2.
ftp.slackware.com/pub/slackware/slackware-7.1/docs 571K, still loading.
Available also in Polish.
I found nothing about DFI motherboards but there is something about buggy
PCI motherboards and linux can try to detect the PCI hardware without
using the BIOS. Also PCI quirks. Maybe I should try bare.i to boot with
and if it works ten times in a row keep it?
But Hal has one computer with the same problem using various standard
linuxes. Also it is intermittent. I will live with it for a while.
>> Does linux strain the hardware more than DOS?
> Yes, Linux uses the available resources to the max.
>> Maybe the SIMMs are causing part of this problem
> That's possible. I recommend a thorough RAM tester,
> like MEMTEST (which can be downloaded free).
Got it even before I read your recommendation.
>> WOuld memory cause this sort of problem with unzipping
> Probably not bad RAM (which usually crashes Linux right
> away). However unzipping a large file does put pressure
> on your CPU and RAM (which might be triggering a hardware
> issue). Is your RAM perhaps being shared with the video
> card? Or is there anything else unusual about your RAM
> (for example, I recall something about certain systems
> having a "memory hole"). Perhaps there is a parameter
> in your CMOS setup for turning off such features?
Four 64MB SIMMs which don't work in all motherboards but did in here.
This was a 2.9MB file, not all that large. Not shared. I had no problems
with two 32 and two 8MB SIMMS. I have 3 of one model and one of another
model 64MB SIMM and maybe they don't like each other?
I will look in CMOS for memory hole if memtest finds nothing.
This is fun. Things that work right are less educational.
(Including commercial linuxes).
> BasLinux mailing list
> BasLinux at lists.ibiblio.org
keesan at sdf.lonestar.org
SDF Public Access UNIX System - http://sdf.lonestar.org
More information about the BasLinux