DMA intermittently disabled, spontaneous reboots and hangs ...
 

Hi guys,

I've purchased a 2nd hand machine.

It's a MSI 694D Pro based machine (came without
documentation, so I don't know which of the 4
sub-models it is)

Dual PIII - 1GHz
512 MB of RAM
2 20GB Seagate HDD's
Matsushita DVD-RAM
TNT2 32MB

The thing came with W2K pre-installed (license-tag
on case). It took 7 minutes to boot, response-times
to mouse clicks were in the 1 minute region, ...
After rebooting it came up with the primary slave
HDD being displayed as something that vaguely
resembled Kanji (not the Seagate model name).

After having powered down and up the drive came
all right, but W2K had shot itself in the foot. After a
recovery run of the CD it did the same thing.
At this stage I clean-installed W2K (which then
supposedly happily worked - no crash, fast)
but after a shutdown the 2ndry HDD was defunct
again.

I reset BIOS to conservative values (Award's
fail-safe option).

Ran Knoppix against it, booted fine, did some file
I/O (read-only on first drive, destructive on 2nd)
with no problem. Then installed Slack on 2ndry
HDD with no problem. The next day Slack came
up crawling slowly, DMA disabled. After machine
was up it was almost happy, Mozilla running in KDE,
wife browsing happily. I ssh'd into the box to configure
the kernel for SMP (all of my attempts failed, kernel
always hung after Reading BIOS data successfully)
with a blank screen, and CAPS and SCROLL-L LED's
lit up ...
The next day machine booted slow (DMA turned off)
Manually enabling DMA => HANG.
Reboot, leaving DMA disabled, compiling kernel =>
HANG.
Win2K still reasonably happy.
Decided to scratch Slack for a test, booted Knoppix,
tried to format hdb1 with XFS => Kernel Panic.
Decided to see how hda behaves. Booted W2K,
defragged NTFS. Booted Knoppix, ran ntfsresize,
ntfsresize complaining about bad blocks, bailed out.
Trying to read NTFS => Kernel Panic.

I'm not quite sure whom or what to suspect in this
scenario, the intermittent DMA switch off may be
HDD failure, may be DMA chipset problems, ... Kernel
traps (did I mention the spontaneous reboots? - had
two of those from within Knoppix when accessing HDD)

Any experience that anyone can offer greatly appreciated.

Thanks for reading on, it has become a rather long post :}


Cheers,
Tink

Long shot, but possibly a factor: is all the RAM the same speed or do you have assorted types? Dunno about the hard drives, but if the system has been randomly rebooting or otherwise misbehaving, that could indicate a memory problem. Personally, if I've got more than one stick of RAM installed in a box, I make sure they are all identical to one another, otherwise the likelihood of incompatibilities increases. I can't say if that's the case here but it may be worth taking a look at -- J.W.

Ta, mate, I'll keep posting :\

Cheers,
Tink

Update ... seems the machine had some transport
related problems, after I dismantled it, ripped out
the RAM and re-seated it in the sockets it seems
to be working really well - and I even found out
how to get my SMP kernels (both 2.4.25 and 2.6.4)
to work with it - I have to disable APIC from lilo.




Cheers,
Tink

P.S.: Buzzwords for searches.
SMP DMA BIOS data hang MSI 694D


:)