Return to BSD News archive
Path: sserve!manuel!munnari.oz.au!mips!mips!swrinde!cs.utexas.edu!qt.cs.utexas.edu!yale.edu!ira.uka.de!chx400!bernina!torda From: torda@igc.ethz.ch (Andrew Torda) Newsgroups: comp.unix.bsd Subject: Re: NMI probs and messages. Message-ID: <1992Aug24.221348.14876@bernina.ethz.ch> Date: 24 Aug 92 22:13:48 GMT References: <1992Aug22.151608.19404@ponds.uucp> <cproto.714656525@marsh> Sender: news@bernina.ethz.ch (USENET News System) Organization: Computational Chemistry, ETH, Zuerich Lines: 36 In article <cproto.714656525@marsh> cproto@marsh.cs.curtin.edu.au (Computer Protocol) writes: [... NMI stuff ...] >is right). My diagnostic software claims my RAM is o.k. Also DOS Windows >3.1 runs just fine. Any ideas anybody. I think it could be some corrupt >page table entry ??? [...] I definitely have corrupt page table entries or page directory entries, but I have no idea which. Unfortunately, I don't know if this is a cause in itself or a symptom of some other problem. The most common way for my machine to crash is with a type 12 panic. Peeking at the source, this is coming from a page fault. Sticking in some printf's, this has always been while looking at a user map, rather than kernel map. I used to think it was always happening during a memory read, but recently I saw my first crash while looking at a page which wasn't vm_read. Sadly, this may all be symptomatic of something else. - I get bizarre failures of commands before a crash. It may just be that the vm system is the place which provokes the actual crash. - The error messages described above are the ones I can look at. I have no idea what is broken when the machine just goes splat-reboot without leaving an error message - These problems only happen at 33MHz. Machine runs under an insane load at 8MHz. Does anyone know a little routine which I could stick into appropriate places in the kernel to check the integrity of the page table entries? This might be one way of trying to find what is breaking before it takes the machine down. -Andrew -- Andrew Torda, Computational Chemistry, ETH, Zurich, torda@igc.ethz.ch