Subject: Re: ffs panic with 1.5C (19/11/2000)
To: Frank van der Linden <email@example.com>
From: Darren Reed <firstname.lastname@example.org>
Date: 11/04/2000 08:54:35
In some email I received from Frank van der Linden, sie wrote:
> On Thu, Nov 02, 2000 at 12:53:32PM +0100, Anders Magnusson wrote:
> > The vgonel panic happens very often on a production machine we have here.
> > Is there anyone that knows more about this area of the kernel that can
> > give me some hints about what to search for?
> This panic appears to be the same one as in PRs 11193 and 11416. Since
> it's also present on the 1.5 branch. it's a serious problem.
> What seems to happen here is that some vnode is put on the freelist
> in a bad state (at least, with a lock that's messed up). At some
> point, a filesystem comes along needing a new one, getnewvnode()
> calls vgonel() on it to flush it out and prepare it for reuse, and
> bang, it panics.
> The thing to do is to collect as much information as possible
> about the bad vnode, most importantly which filesystem it belonged
> to previously.
Which parameters, from a ddb backtrace should we be examining and in
what way ? Can you give us a bunch of ddb instructions which will
extract what you need when it panics again ?