[plug] Intermittent hardware problem

[Craig Foster]

> -----Original Message-----
> From: plug-bounces at plug.linux.org.au 
> [mailto:plug-bounces at plug.linux.org.au] On Behalf Of Jim Householder
> Sent: Saturday, 18 December 2004 1:39 PM
> To: plug at plug.linux.org.au
> Subject: [plug] Intermittent hardware problem
> 
> Hi
> 
> One of my machines has a problem.  2.6Ghz Celeron, MSI 
> all-in-one M/B, 256MB ram, new hdd.  hdd was purchased 
> yesterday, the rest last March.  CDRW is 4 years old and DVD 
> reader is 3 months old.  Cables and power supply are 4 years old.
> 
> Over the last month or so it has been spontaneously rebooting 
> from time to time.  Now as likely as not on reboot it hangs 
> after displaying the PCI device listing and before the 
> "Verify DMI pool data..." message.  A reboot at this point 
> ususally works.
> 
> Something is corrupting/replacing files.  No errors are 
> displayed.  Yesterday I installed Windows XP Home and 
> Mandrake 10.1 Official on the new hard disk, and things 
> worked ok.  A reboot to XP later resulted in scandisk running 
> and fragments of files found.  This morning I booted linux 
> and firefox would not start.  Said libssl3.so had a bad ELF 
> header.  On checking, it had totally different data.  Same 
> size etc.  No idea what the original inode was so I could not 
> check that.
> 
> I ran memtest-86 for 1.5 hours, 11 passes of standard test, 
> and no errors were reported.
> 
> I initally thought it was a disk problem as there was a lot 
> of file corruption.  XP Home was used exclusively and VET 
> reported no viruses.  sfc -purgecache fixed the problems in 
> some cases.  I replace the drive but problems continue.  
> Thinking back, if it had been the drive there should have 
> been at least a lot of retries.
> 
> Where do I look next?
> 
> TIA
> Jim

What brand and model of power supply, motherboard, and memory do you have?

*Never* underestimate the corrupting force of a powersupply on the dark side
(no, not MS this time)
You could also set Windows to *not* reboot automaticaally, which will give
some indication of what's going on.
Manually setting the RAM timings & voltages are also suggested with so many
lying/dodgy SPDs around at the moment.

CraigF.