Date: Sun, 30 Dec 2007 20:10:43 +0100 From: Kris Kennaway <kris@FreeBSD.org> To: frank@exit.com Cc: stable@freebsd.org Subject: Re: Repeated crashes in 6.3-PRERELEASE Message-ID: <4777ED33.8000306@FreeBSD.org> In-Reply-To: <1199040933.1579.10.camel@jill.exit.com> References: <1198956498.2124.6.camel@jill.exit.com> <1199040933.1579.10.camel@jill.exit.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Frank Mayhar wrote: > On Sat, 2007-12-29 at 11:28 -0800, Frank Mayhar wrote: >> This is on a very recent -stable (like, as of just a couple of days >> ago). Doing nothing in particular, ZERO_COPY_SOCKETS _not_ specified in >> the config. This has happened twice this morning; if it happens again >> I'll drop back to my older kernel, built back in May. > > Sigh. In the last three days I've had no fewer than four crashes: This > one, another just like it that I didn't get a dump for, a "Fatal trap > 12: page fault while in kernel mode" in buf_splay(): > > #27 0xffffffff802e2cb4 in buf_splay (lblkno=0, xflags=2 '\002', root=0x9000) > at /usr/src/sys/kern/vfs_subr.c:1339 > #28 0xffffffff802e2e08 in buf_vlist_remove (bp=0xffffffffa1919340) > at /usr/src/sys/kern/vfs_subr.c:1371 > #29 0xffffffff802e4dab in brelvp (bp=0xffffffffa1919340) > at /usr/src/sys/kern/vfs_subr.c:1502 > #30 0xffffffff802d28f1 in getnewbuf (slpflag=0, slptimeo=0, size=2, > maxsize=16384) at /usr/src/sys/kern/vfs_bio.c:1801 > > (where bp->b_left is 0x9000) > > and a "panic: backgroundwritedone: lost buffer." I have a dump for each > of the three unique scenarios. At the moment I've dropped back to the > older kernel and am keeping my fingers crossed that it will run okay. > I'm beginning to suspect hardware, although as two of the three are in > the buf code there might be a problem there... If it continues with the > older kernel then it is most likely hardware but if not then there's a > problem. > > Has anyone else encountered anything like this in 6.3-PRERELEASE? Sounds pretty unusual (and e.g. I've not seen other reports of this), but if you are *still* able to run an older kernel without incident then maybe there was an intervening change that interacts badly with your particular hardware. A binary search might be necessary to determine the cause, in that case. Kris
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4777ED33.8000306>