Total system freeze - hard reset needed

[McLink]

I am experiencing the problems described in this thread on the Ubuntu forums; after some time of usage (usually a few days), my X will completely lock up, though I am still able to move the mouse. I cannot switch to VTs, and I can't even ssh in. Today, I got out of bed and found my computer locked up again. I could move my mouse to unblank the screen (I have Xscreensaver set to turn my monitor off), but I couldn't do anything else. After a hard reset, my system would throw file system errors, and I've only managed to fix them after several hours in SysRescCd, so I decided it's time to post for help.

It may be related to the Xorg memory leak problem, though I'm not sure; I'll investigate it further.

My system information:
emerge --info

[notHerbert]

Maybe it could be hardware failure. Like harddrive maybe. Those things happen.

[McLink]

Doubt it. I've had problems with locking up for a while now (though I used to turn my computer off when going to sleep, which I haven't been doing a lot anymore lately). Actually, it locked up more frequently but less seriously back when I used GNOME. Back then, I could switch VTs and kill X.

I don't seem to have lost any data, and I got no reports of bad blocks when running e2fsck with the -c option. Also, since other people are experiencing the exact same problems as I am, I'm looking more in the direction of a software problem than a hardware one.

Of course, I'm not truly ruling out the possibility of a hardware failure (though it would suck if it were the case), but since I haven't really noticed anything else out of the ordinary, software failure still seems more likely to me.
_________________

[Dont Panic]

If you are locking up so hard that your ssh server isn't even working, you may want to consider expanding your scope beyond looking for a problem with X. Even if X locks up, your ssh server should still be running.

It may not be the problem, but when I hear about hard lock-ups and file-system corruptions, it might be a good idea to do a memtest, just to eliminate that as a possibility. With 8 GB of RAM, it might actually take a while before you'd hit a possible bad spot in your RAM.

[McLink]

[McLink]

I think I might (with stress on the "think" and "might") have traced the cause of the problem. Just now, my partitions suddenly disappeared entirely; a problem I'd had before and thought was caused by overheating at the time, and "fixed" by placing my computer outside of the desk it's in. Anyway, when I hard-rebooted, my BIOS didn't recognise my "sda" (a 2-year-old 320GB Maxtor SATA II drive). (I've recently placed my computer back inside the desk, but vastly improved the airflow by cutting out the back board and placing the entire cabinet-style desk 5cm/2" from the wall. The computer was only slightly warm to the touch at the time the partitions went AWOL, and wasn't any hotter on the inside when I opened it perhaps some 2 minutes later - so I doubt it's overheating.)

Anyway, I think the cause might have been a faulty SATA cable. I replaced it, and when I turned on my computer, the BIOS immediately recognised my hard drive. And if it isn't the cable, at least I can be fairly certain it's either the hard drive itself, or a motherboard problem: the BIOS never failed to recognise my second drive (a 3-month-old 500GB Western Digital Caviar SATA II drive). If I must, then I can get a new hard drive. The data on this one appears to be fine, so I could probably dd the data over. But I doubt it's a problem with the hard drive is fine, because I've never had any IO errors or anything of the sort, and fsck didn't report any bad blocks.

As for memtest86+, it sort of... gets stuck. Well, I'm not sure it's actually stuck, but when I start it, I get the normal screen, but it's quite unresponsive. The + in the title area blinks.

I will update this topic if I see any more of this.
_________________