On 2005-08-22 10:57:47 -0400 (Mon, Aug), R'twick Niceorgaw wrote: > > I also had the same gcc segfault problem recently. After some observation, > I found out whenever cpu temperature goes beyond 50 C I was getting > segfaults. Took out the heat sink and cpu, cleaned them put some arctic > silver between cpu and heat sink and put a fan in front of the box and did > a emerge -Uv world without any problem even though temperature reached > 65. I believe, its the arctic silver that I put between cpu and heat sink > fixed my problem.
I am in the middle of troubleshooting similiar (lockups, not segfaults) situation. Maybe it will be of some use to somebody. During some emerges my machine was getting locked up when cpu temperature got above 55 C (AMD Athlon 2GB), but I decided (yeah - decided ;-) that CPU is not the problem - 55 C is pretty low for CPU I think. As a workaround I have limited the RAM amount available (I have 512 MB physical) to 256 MB by kernel parameter mem=256M. No RAM errors were detected by memtest, but limiting available RAM works - no lockups. The 'sensors' utility (part of lm_sensors package) shows that one (out of two) temperature sensor is a bit too hot: it has now (no emerge in progress) 61 C and warning threshold is on 48 C. I assume that it is not CPU that gets too hot but that small, nice north bridge on my mobo. May it be that the solution is to get some bigger heat sink and more powerful fan not only for CPU. HTH -- No virus found in this outgoing message. Checked by "grep -i virus $MESSAGE" -- gentoo-user@gentoo.org mailing list
