Re: File Corruption

H. Peter Anvin (
26 Sep 1997 08:11:56 GMT

Followup to: <>
By author: Jeremy Fitzhardinge <>
In newsgroup:
> I saw file corruption for the first time in ages under 2.1.56 last
> night. I'd been running all day (~12 hours) doing edit-compile cycles
> on some Java code, and the Java compiler stopped working. One of the
> class files had been corrupted such that an 'n' became a 'k'
> (1101110->1101011). It looked like a single character change;
> everything around it was OK. I had real work to do, so I just rebooted
> and continued. There was no apparent disk corruption; it was just in
> memory.
> I haven't reproduced it, nor seen anything like it before (except around
> 1.1 when there was a buffer-corrupting race condition or something). I
> was also dialed up to my ISP with ppp, and using Netscape 4.03 to
> browse. There was no swapping.

In 100% of the cases when something like this has happened, especially
randomly, it has been memory-related, usually a bad SIMM or a bad
motherboard, but occationally bad cache. I swear by ECC these days...


    PGP: 2047/2A960705 BA 03 D3 2C 14 A8 A8 BD  1E DF FE 69 EE 35 BD 74
    See for web page and full PGP public key
        I am Bahá'í -- ask me about it or see
   "To love another person is to see the face of God." -- Les Misérables