Date: Fri, 7 Dec 2007 01:35:27 +0100 From: "Ivan Voras" <ivoras@freebsd.org> To: current@freebsd.org Subject: unionfs over cd9660 - solved (?) Message-ID: <9bbcef730712061635k382f8838t33c77756212798ad@mail.gmail.com>
next in thread | raw e-mail | index | archive | help
hi, I've reported (at least twice) a problem with mounting unionfs over cd9660 as the underlying device, that seemed to appear in 7.0-BETA2 or 7.0-BETA3. I've received no independent confirmation of the problems or ideas why it would occur, so I kept trying to solve it when I got time. At last I I've narrowed it down to a strange conclusion: it appears the problem was connected somehow to my build machine, i.e. it's probably not in the code itself. After unsuccessfully trying various dates for cvsup times and similar things, I've created a clean machine as a last resort, and the problem was gone! After comparing the two setups, the most likely cause is a compiler bug when certain optimizations are enabled. The old system had "-Os -mtune=generic", and based on previous experience and reports on the mailing lists I believe the "-Os" part is the most likely culprit. I'll try to verify this theory soon-ish. This solves my problem but the mystery remains why the error condition suddenly appeared at that specific time. Of course, with a buggy compiler almost any change in the code could have provoked it, but still, the rest of the system (and ports) compiled with the flags was working perfectly fine.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9bbcef730712061635k382f8838t33c77756212798ad>