From owner-freebsd-hackers Tue Nov 28 15:01:17 1995 Return-Path: owner-hackers Received: (from root@localhost) by freefall.freebsd.org (8.6.12/8.6.6) id PAA14024 for hackers-outgoing; Tue, 28 Nov 1995 15:01:17 -0800 Received: from nomad.osmre.gov (nomad.osmre.gov [192.243.129.244]) by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id PAA14004 for ; Tue, 28 Nov 1995 15:01:04 -0800 Received: (from gfoster@localhost) by nomad.osmre.gov (8.6.11/8.6.9) id SAA14720; Tue, 28 Nov 1995 18:00:31 -0500 Date: Tue, 28 Nov 1995 18:00:31 -0500 From: Glen Foster Message-Id: <199511282300.SAA14720@nomad.osmre.gov> To: hackers@freebsd.org Subject: posessed by daemons? Sender: owner-hackers@freebsd.org Precedence: bulk I am having very weird problems with a machine that I am trying to bring from 1026-SNAP to 2.1.0-RELEASE through the make world/build kernel/make world process. The src seemed to extract OK but I found a couple of bit flips in various files that killed the compile. When checking this out I found that md5's on the src tgzs were not consistent, and inconsistent in a weird way. If I do 'md5 *' in the src directory, three or four of the 131 files don't match the md5's on the same files generated on another machine. Different files exhibit this from run to run! However, I can do 'repeat 10000 md5 sgnu.aa | sort | uniq' and not get any mismatches (no matter what value of sgnu.aa I select :-). I am seriously mistrusting this machine at this point! It is a P90, SiS, L2 cache disabled, 2940, 4GB Barracuda, as slow as the CMOS will let me on the memory accesses. I don't know whether to blame the memory or disk sub-systems. Of course, no errors are ever reported on either and for user stuff it runs like a champ, although it is very lightly loaded. There may be some incremental bit rot going on but I haven't seen any in the three weeks or so it has been up. WTF is going on? I can, of course, get more info. on the motherboard if it will help. Thanks, Glen Foster