Date: Tue, 20 Jul 2004 08:58:49 +0200 From: "Willem Jan Withagen" <wjw@withagen.nl> To: "Robert Watson" <rwatson@freebsd.org> Cc: current@freebsd.org Subject: Re: panic: Duplicate free of item 0xffffff005c4a8600 fromzone0xffffff007fed4780(Mbuf) Message-ID: <15fd01c46e27$02e49050$471b3dd4@digiware.nl> References: <Pine.NEB.3.96L.1040719095628.38770D-100000@fledge.watson.org>
next in thread | previous in thread | raw e-mail | index | archive | help
From: "Robert Watson" <rwatson@freebsd.org> > On Mon, 19 Jul 2004, Willem Jan Withagen wrote: > > > > I'll starts some more tests before I'm of to bed. > > > > This mornings result: > > ==== > > System call getdirentries returning with the following locks held: > > exclusive sleep mutex bdone lock r = 0 (0xffffffff805fd080) locked @ /home2/src/ > > sys/kern/vfs_bio.c:3767 > > panic: witness_warn > > cpuid = 0; > > KDB: stack backtrace: > > kdspin lock sched lock held by 0xffffff007b6cc940 for > 5 seconds > > panic: spin lock held too long > > cpuid = 0; > > KDB: enter: panic > > ==== > > > > But no way to get into the debugger. Does not look like it is much > > network related??? > > Doesn't look very network related, although it could be that increased > concurrency and lack of waiting on Giant open up a race of some sort. Can > you confirm "options DDB" and "options KDB" are both in your kernel > config? You may want to consider commenting out "#define PREEMPTION" in > the copy of params.h for the architecture you're running on and see if > that helps. Won't help interrupt processing latency, but probably won't > hurt server throughput, and your box is a server box so it might be worth > trying. The first time that it made it through the night since quite some days... The real stress test, "make -j32 buildworld" is still too much however. --WjW
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?15fd01c46e27$02e49050$471b3dd4>