[mythtv-users] OT: Backend Crashes with Machine Check Exception

[Todd Pearsall]

Every 24 to 48 hours I get a hard crash on my Myth Backend:

kernel: CPU 0: Machine Check Exception: 0000000000000004
kernel: Bank 4: b200000000070f0f
Kernel panic: CPU context corrupt

My reseach pointed to a likely hardware issue, so one at a time I tried the following:
- swap RAM with a known "good" chip
- swap RAM with a different known "good" chip
- ran memtets86 for a week (no failures)
- moved around the PCI cards
- swapped video cards
- tried different PVR500 cards
- swapped CPUs
- kernal updates to 2.6.12

After each change the system still crashed with the exact same error.  More research pointed to a potential issue with the nVideo chipset and DMA.  I then set the system to boot with no DMA and it is rock solid.  The only problem is that 3 or 4 PVR500 recordings and a stream or two leave the system in a 50% cpu wait state due to the slow disk.

I'm having a hard time believing there is a DMA issue with the nVideo chipset and there is virtually no discussion of it, so I assume it's something hardware related to the disk.

Next I'm planning to pull the CDROM and swap the IDE cables.  Does anyone have other ideas I should try?  I've been holding off on:
- a power supply swap since I don't have a big enough spare and would hate to spend the money for a "test".
- drive swap since it will be a pain
- motherboard swap since it will be an expensive test

BTW, it is an ABIT NF8 motherboard that I am otherwise very happyy with.

Thanks,
Todd
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mythtv.org/pipermail/mythtv-users/attachments/20051121/52e1f59b/attachment.htm