Ben Rosenberg wrote:
If anyone has any suggestions at this point I'll try them. I've attached a plain text file called hw which is the output of hwinfo --all. If anyone sees anything weird about the output then please let me know. Other wise I will have to find another solution to this. This machine ran 7.1, 7.3 and 8.0 without a problem so I know it's not the machine.
I don't know, sure looks like a hardware fault to me. Like duff memory. Does 8.1 still have the memtest86 option on the boot CD / DVD? If so give it a go, or give it a go on the 8.0 version. It is possible a fault has developed *since* or *while* you were installing 8.1. It's always possible you zapped a chip when you had the case open, or a power surge... So, put 8.0 back on it and see what happens, but try memtest86 first. You say you switched acpi off, or I'd have suspected that. Maybe you have DMA related problems? Earlier versions of SuSE (in my experience anyway) defaulted to having DMA turned off, and had to be set up manually (hdparm command in /etc/init.d/boot.local) but suse 8.1 has selected what I would consider the "right" settings for my drive by itself. Maybe on yours it's done the wrong thing and you need to manually set more conservative settings. In the first instance just turn DMA off in yast2 and see if the freezing stops (bear with the slower disk access for the time being). Last June I had very similar random crashes start appearing once I upgraded to kernel 2.4.4 on disk access through the onboard Promise IDE/RAID controller *only* when DMA was enabled. I got to be able to reproduce it consistently by doing a big bonnie test (ie: using the disk subsystem so heavily that it raised the chances of it happening during the test to nearly 100%). In the end I didn't wait for this to be fixed, but switched to using the on-board VIA IDE controller instead, though the original problem is very likely fixed by now. (the whole sorry tale is archived in the lkml) You're probably not experiencing the exact same problem, but DMA may still not be set up correctly. Or it might be a newly-introduced bug in the 2.4.19 kernel - or the *SuSE* 2.4.19 kernel - for your IDE controller. BTW, the hw file you said was attached, wasn't :-) Probably suse's mail server stripping it. -- Rachel