Over the past four days, I have struggled with this recurring problem. It first occurred when I installed the newest package version of xbase and xserver-svga from the debian mirrors. I had also switched to WindowMaker, so I thought maybe the problem was there. So far: it hasn't happened to me in another WM, but I've only tried a couple and never stay in them for more than a few mins. Anyway, for a period X would crash on me within about five minutes after login. Everytime it was the same. The xdm-errors log said something like "Caught signal 11, server aborted." Same error everytime. I then attempted to recompile X, in X. X itself crashed about six mins into it. So I compiled XFree86 3.3.2 to use only the matrox mga chipset, and after compiling, did a make install, and put a hold on the parallel debian packages. (xbase and xserver) I also recompiled WindowMaker myself, and installed. All was fine for about 24 hrs, when X crashed on me again while I was in it. The crashes occur whether or not I'm doing something at the time... i.e. not only when I am moving something around or causing mouse or keyboard events. Sometimes I'm not even here when it happens. Besides recompiling things, I have taken several steps to remedy this. First, I opened my case and left it open, and screwed the fan onto my CPU even tighter. I don't think it's overheating, though... I have one of the low-power revisions of the Cyrix/IBM 6x86 PR200+ on an Asus T2P4 HX pentium M/B. Second, I stopped running the rc5des client, but then I started it again after I recompiled, because all seemed well. Third, I changed the RAM settings in the BIOS to 70ns ( I am supposed to have 60 ns EDO ). Anyway, last night I left it in WMaker, and woke up this morning to find the xdm login. The xdm-errors file didn't reveal any errors this time, though. It may very well be a problem with the RAM I've got. I just bought an extra 32 megs (two 16 meg simms) about 1.5 mths ago. To make a long story short, one of the simms must be bad, because I only get a total of 48MB now (even though I put them both in, because the Motherboard won't recognize a single simm). I am hoping this is the problem.... a bad simm, or just bad memory in general. Or heat, but there's not much I can do about that. I have taken the new simms out, and am back to my old 32. I brought the RAM back up to 60 ns. And the rc5des client is still going. If it was the RAM, it shouldn't happen anymore. I hope to get my RAM changed soon, because 32 just doesn't cut it anymore. One last thing: the problem is only in X. I don't get any kernel messages about memory allocation (I do sometimes, but I think that's just my 2.0.32 kernel). This thing only crashes X. Whether I use xdm or not, i get signal11. Compiling X, or anything else not in X, never gives me any problems. If there is anyway of tracking down the problem, I would appreciate it. I was hoping I could trap signal 11 and ignore it, but I have a feeling that's impossible. If anyone has any advice or experience with this , PLEASE let me know. This is the most frustrating experience I have had so far with Linux. I can't even do my work for fear of X dying. -- kiyan@iname.com A scientific truth does not triumph by convincing A. Kiyan Azarbar its opponents and making them see the light, but Ottawa, Canada rather because its opponents eventually die and a Linux 2.0.33 new generation grows up that is familiar with it. 1024/0x9A9EC5EA 4F3ADBDA1EE5850209DD8BB205250ED2F696A7BE ^- Max Planck
Attachment:
pgpqkB9Fec4js.pgp
Description: PGP signature