Date: Mon, 11 Oct 2004 14:54:17 -0400 From: Brian Fundakowski Feldman <green@freebsd.org> To: Robert Watson <rwatson@freebsd.org> Cc: current@freebsd.org Subject: Re: "panic: spin lock entropy harvest mutex not in order list" Message-ID: <20041011185417.GG988@green.homeunix.org> In-Reply-To: <Pine.NEB.3.96L.1041011042445.31040D-100000@fledge.watson.org> References: <16745.57237.472660.492897@jerusalem.litteratus.org> <Pine.NEB.3.96L.1041011042445.31040D-100000@fledge.watson.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Oct 11, 2004 at 04:26:23AM -0400, Robert Watson wrote: > On Sun, 10 Oct 2004, Robert Huff wrote: > > > > > About 2000 UTC Sunday I cvsuped system source and rebuilt. > > "make buildworld" went fine. (Log available) > > "make buildkernel" also good. (Log also available) > > I installed the new kernel and rebooted. > > When the system came up, I got: > > > > Real memory: XXXXXXXXX > > Avail memory: YYYYYYYYY > > panic: spin lock entropy harvest mutex not in order list > > KDB: enter: panic > > [thread 0] > > ddb> > > When I committed the rewrite of the /dev/random entropy harvesting > locking, I ommitted to commit the change to subr_witness.c telling witness > about the lock change; witness requires that all spin locks be identified > in the hard-coded lock order (although maybe only if witness skipspin is > off, hence more people not running into it?). I've merged the change to > subr_witness.c as change 1.182 as of a minute or two ago. Please letme > know if that change doesn't fix the problem. We should really not do that -- makes loading modules that try to use spinlocks ~impossible. For example, the nvidia driver. -- Brian Fundakowski Feldman \'[ FreeBSD ]''''''''''\ <> green@FreeBSD.org \ The Power to Serve! \ Opinions expressed are my own. \,,,,,,,,,,,,,,,,,,,,,,\
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20041011185417.GG988>