On 6/23/2008, Philipp Kolmann (kolmann@zid.tuwien.ac.at) wrote:
It fails on random files, since the system is running a few daemons without any problem I'd go for the gcc bug.
Actually, failing at random point when compiling is *much* more likely to be a memory error.
If the 'few daemons' don't happen to ever hit the bad ram, that would explain why they don't fail.
Just try replacing the RAM (RAM is chep nowadays), and see if it fixes the problem.
or test with memtest (if x86):
I have seen bad memory pass memtest for hours on end, so memtest is not a 100% reliable way to test it. The best almost sure-fire test I have found is compiling something big (like X, or even a kernel)...
Again... ram is chep... just replace it and see.
--
Best regards,
Charles