Derek, from your paper:
> This was a fun set of runs. The most exciting thing to happen was
> the complete failure of gcc under two seperate runs under the 2.2.19
> kernel! gcc died on the inital run with "parse error" messages. To
> make sure nothing had gotten corrupted after the first run/failure,
> I re-extracted the sources, re-made the config, rebooted the box and
> tried again. The second time through it succeeded. The third time
> through it died with another "parse error", this time at a different
> point in the code. The conclusion was that 2.2.19 just couldn't
> handle this load on this hardware. (My suspicion is that it was some
> sort of IDE/hard drive failure. I'm not impressed with the Toshiba
> hard drive in this machine and it possibly spit out a bad bit or
> two. Or some kind of very occasional random flakiness with the IDE
> interface and 2.2.19.)
When is the last time the RAM was tested in that machine? The same gcc
parse errors were copping up on a freebsd system during a make
buildworld (heavy memory/compiler activity). The machine had 96MB ram
total and would throw the errors randomly on seemingly good
code. memtest86 was run and it found a faulty module, but not on the
first pass, it took a few passes for the corruption to occur. The
module was replaced (leaving me with 32MB less RAM than before),
and the errors stopped completely.
This archive was generated by hypermail 2.1.3 : Fri Aug 01 2014 - 16:35:50 EDT