Date: Sat, 16 Feb 2008 20:10:35 +0100 From: MFL Commissioner <mfl-commissioner@marino.st> To: Juergen Lock <nox@jelal.kn-bremen.de> Cc: freebsd-current@freebsd.org Subject: Re: 7.0 RC2 kernel panic with Kqemu/AMD64 Message-ID: <47B7352B.1040302@marino.st> In-Reply-To: <20080216175811.GA33393@saturn.kn-bremen.de> References: <43555.82.234.78.29.1203154742.squirrel@secure.synsport.net> <20080216175811.GA33393@saturn.kn-bremen.de>
next in thread | previous in thread | raw e-mail | index | archive | help
Juergen, I would be happy to help debug this, although someone would have to give me step by step instructions on what to do exactly. This is a development machine that (roughly) mirrors a production setup, so I have maintain the SMP/AMD64 configuration. What I am still not quite clear on is this: Is this a Qemu problem or FreeBSD kernel problem? Anyway, I'll do what you ask to provide you with my backtraces, but I doubt that I will be able to interpret the results. John Juergen Lock wrote: > On Sat, Feb 16, 2008 at 03:39:02AM -0600, John Marino wrote: > > Looks like you are running amd64 SMP, it is a known problem that kqemu > doesn't really work in that case (see pkg-message of the qemu port); > the exact cause of this is still unknown - the backtraces I got are rather > weird and not really helpful, see this thread: > http://lists.freebsd.org/pipermail/freebsd-emulation/2007-November/004182.html > and there also is: > http://www.freebsd.org/cgi/query-pr.cgi?pr=113430 > If you or anyone else wants to help debug this you are very welcome of > course (it was suggested this might be a cache coherency issue, so maybe > your backtraces will look different since you're on an AMD CPU - mine is > Intel), otherwise your best bet probably is to use an UP kernel, or an > i386 install, interestingly there kqemu works also with SMP. (well or > disable kqemu of course.) > > Oh, another reason for my weird backtraces might be the minidump race, > and if I read the commitlogs right it is not actually fixed (I obtained > my dumps before it was discovered), the patches only made sure not to > write over the end of swap. So maybe I should try again with minidumps > disabled (if thats possible), one of these days... > > Sorry, > Juergen >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?47B7352B.1040302>