Re: Having weird seqfaults

[Pascal Hambourg <pascal@plouf.fr.eu.org>]

Goren Buckwalk a écrit :
> 
> Could the segfaulting be caused by overheating? I thought when older
> CPUs overheated they just ran until the circuits melted (this is an AMD
> K6-2 450). More recent ones shutdown before damage can occur. But
> whatever occurs I thought happend to everything, there isn't any way an
> individual process could get terminated because they were causing an
> high usage leading to overheating, right?

The real world is analog, not boolean. There is a gap between the
maximum proper operating temperature and the core meltdown temperature.
In this gap, many more or less tiny dysfunctions can happen. For exemple
a single bit error which transforms a 1 into 0 or vice versa, changing
the meaning of an instruction and causing the segfault of the process
executing that instruction.