Date: Fri, 4 Dec 1998 15:10:06 -0800 (PST) From: Steve Kargl <sgk@troutmask.apl.washington.edu> To: luoqi@watermarkgroup.com (Luoqi Chen) Cc: freebsd-current@FreeBSD.ORG, kuehn@rz.tu-clausthal.de, dg@FreeBSD.ORG, kirk@mckusick.com Subject: Re: panic: ffs_blkfree: bad size Message-ID: <199812042310.PAA35399@troutmask.apl.washington.edu> In-Reply-To: <199812041546.KAA24938@lor.watermarkgroup.com> from Luoqi Chen at "Dec 4, 1998 10:46:50 am"
next in thread | previous in thread | raw e-mail | index | archive | help
[cc'd to dg and kirk]
According to Luoqi Chen:
> > I had such a panic today too during cvsup activity. Two filesystems
> > don't have softupdates. All others do. The activity was on one
> > of those softupdates filesystems.
> >
> > dev=0x20404, bno = 13, bsize = 8192, size = 8192, fs = /var
> > panic: ffs_blkfree: bad size
> >
> bno is a fs block number and for /var it almost surely can't be 13. This
> still points to a missing VOP_BMAP() somewhere.
>
> While I'm writing this message, I took another look at Kirk's patch. It seems
> to me that he forgot to call VOP_BMAP() for the last buf in the reallocation
> list. Try this patch, and see if you still get the panic,
>
[patch deleted]
Luoqi,
Your patch has survived my torture test. I did:
(1) make -j 32 buildworld (CFLAGS = -O -pipe)
(2) cd sys/compile/GENERIC (COPTFLAGS = -O)
foreach i (*.h)
make -j 16
sleep 10
make clean
sleep 10
end
(3) cd ports/devel/ddd ; make
This would crash my machine without your patch within an hour.
I hit a peak load average of 40 with a load average of around 14
throughout the cycle. Amazingly, with 256 MB on a dual PPro machine,
I never touched the swap space.
--
Steve
finger kargl@troutmask.apl.washington.edu
http://troutmask.apl.washington.edu/~clesceri/kargl.html
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199812042310.PAA35399>