System hardlocks after 10-20 minutes!

optilude

Hi,

I got back from 6 weeks without internet yesterday, and decided it may be a good idea to do an emerge -u world. 200-some updates, among them glibc 2.3.4-20040419 (I think) and gcc (both latest ~x86 unstable).

When it got to glibc, it told me I had to use kernel 2.6.5 for NPTL support. I had 2.6.1-gentoo-r1 with NPTL on my old glibc, but I emerged 2.6.5-gentoo-r1, copied my .config over, did menuconfig and all looked ok, and compiled and installed the kernel no problem. A reboot later, I installed the newest glibc and all was well. I then rebooted and started the rest of the updates.

However, about half an hour after I did, my system locked hard, the fan started running like crazy (it's a centrino system with acpi enabled) and the system was completely unresponsive. Not even Alt+SysRq commands were recognised, so I had to take the battery out and pull the plug.

I thought it may be an isolated incident, but I got the same problem again after just using the system normally for a while. It seems now that it locks up after 10-30 minutes of usage; the time seems shorter when the system is under heavy load. I once saw a kernel panic and a memory dump fly past on the logging console, but the message wasn't saved in /var/log/messages, so I don't know the exact error.

I tried taking NPTL out of my USE flags and re-emerging glibc. That was quite difficult, because the system locked up half way through, but I managed by using ebuild and doing compile and install separately. However, the problem persists.

This of course comes at a terrible time! I have a deadline on a project and really need my machine back.

Any ideas would be highly appreciated!

Martin
_________________
--
"Life is both a major and a minor key" -- Travis

georwell · Posted: Tue Jul 27, 2004 10:44 am Post subject:

can you monitor the temp of your machine?

optilude · Posted: Tue Jul 27, 2004 11:42 am Post subject:

mattjgalloway · l33t Joined: 16 Mar 2004 Posts: 761 Location: Coventry, UK

Woah!

That's hot! I wouldn't be surprised if things seg faulted at that heat. The heat will dissipate through all metal things, so it's probably heating your RAM too, which is probably dying. Do a memtest (www.memtest86.com / emerge memtest).

It's probable that your system has finally met it's match with all that heat. If it was on for a long time at 80C then it's gonna be bad news...
_________________
AMD64 3200+, 1024MB RAM, Gentoo Linux
MacBook Core Duo, 1024MB RAM, Leopard

optilude · Posted: Tue Jul 27, 2004 2:39 pm Post subject:

Yes, I've been worried about it for a long time.

But.. how do I get the heat down? It's 15C outside, and it gets that hot just compiling stuff, with the fan running every now and then with varying speeds, as necessary (or so I assume).

I just did a memtest86; I think it's telling me it's got failures. :-(

I'm gonna call the manufacturer and complain.

Martin
_________________
--
"Life is both a major and a minor key" -- Travis