I posted here last month about a diskless ELC machine (dmesg should be
in the archives) that was suffering apparently random failures during
kernel compilation. GCC would exit on SIGBUS or SIGSEGV at unpredictable
points in the compilation process, but always roughly 15-30 minutes into
it.

I have now essentially replaced the entire machine, with no change. I
swapped the motherboard with a new one on suspicion that the CPU might
be fried after years in a fanless machine, and all four SIMMs have been
swapped with new sticks one by one. The only thing that remains the
same is the case, including peripheral ports and monitor, and I doubt
it's any of those.

On the suspicion that the level of EM disturbances in this room might
have something to do with it, I have moved it to a different room. No
change.  

Out of desperation, I downgraded from 1.6 to 1.5 and repeated the same
procedure, and was able to complete the kernel compilation twice in a
row. So I'm looking again at the OS after dismissing it as the culprit
before. 

What changed for this machine between 1.5 and 1.6? What's a good place
to start looking?

Thanks,

Dan
--
/^Dan Debertin$/         
airboss@nodewarrior.org  | Did I sleep a little too late,
www.nodewarrior.org      | or am I awake?        --Byrne