From owner-freebsd-hackers  Thu Sep 12 13:22:17 2002
Delivered-To: freebsd-hackers@freebsd.org
Received: from mx1.FreeBSD.org (mx1.FreeBSD.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id 911EA37B400
	for <freebsd-hackers@FreeBSD.org>; Thu, 12 Sep 2002 13:22:14 -0700 (PDT)
Received: from nd250009.gab.xdsl.ne.jp (nd250009.gab.xdsl.ne.jp [61.202.250.9])
	by mx1.FreeBSD.org (Postfix) with ESMTP id 6C36043E42
	for <freebsd-hackers@FreeBSD.org>; Thu, 12 Sep 2002 13:22:12 -0700 (PDT)
	(envelope-from nork@FreeBSD.org)
Received: from cl-server.enusure-tech.co.jp (cl-server.ensure-tech.co.jp [211.18.249.19]) (authenticated bits=0) by nd250009.gab.xdsl.ne.jp (8.12.5/8.12.5/NinthNine) with ESMTP id g8CKM004040532 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=NO) for <freebsd-hackers@FreeBSD.org>; Fri, 13 Sep 2002 05:22:02 +0900 (JST) (envelope-from nork@FreeBSD.org)
Message-Id: <200209122022.g8CKM004040532@nd250009.gab.xdsl.ne.jp>
Date: Fri, 13 Sep 2002 05:22:00 +0900
From: Norikatsu Shigemura <nork@FreeBSD.org>
To: freebsd-hackers@FreeBSD.org
Subject: filesytem bsize=64k causes libc/db hash crash
X-Mailer: Sylpheed version 0.8.2 (GTK+ 1.2.10; i386-portbld-freebsd4.6)
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
X-Virus-Scanned: by amavisd-milter (http://amavis.org/)
Sender: owner-freebsd-hackers@FreeBSD.ORG
Precedence: bulk
List-ID: <freebsd-hackers.FreeBSD.ORG>
List-Archive: <http://docs.freebsd.org/mail/> (Web Archive)
List-Help: <mailto:majordomo@FreeBSD.ORG?subject=help> (List Instructions)
List-Subscribe: <mailto:majordomo@FreeBSD.ORG?subject=subscribe%20freebsd-hackers>
List-Unsubscribe: <mailto:majordomo@FreeBSD.ORG?subject=unsubscribe%20freebsd-hackers>
X-Loop: FreeBSD.ORG

	I found a problem on 4-STABLE(maybe 5-CURRENT, too) with libc/db
	(BerkeleyDB 1).  It that when block size on fs is 65536 bytes,
	many programs which use BerkeleyDB(dev_mkdb, cap_mkdb, etc..) are
	crash.

	[REPLICATE THE TEST]
	# dd if=/dev/zero of=/dev/test.image bs=1m count=32
	# vnconfig -s labels -c /dev/vn0c /tmp/test.image
	# newfs -b 65536 -f 8192 /dev/vn0c
	# mount /dev/vn0c /mnt
	# dev_mkdb -f /mnt/dev.db /dev
	zsh: segmentation fault (core dumped)  dev_mkdb -f /mnt/dev.db /dev

	# umount /mnt
	# newfs -b 32768 -f 8192 /dev/vn0c
	# mount /dev/vn0c /mnt
	# dev_mkdb -f /mnt/dev.db /dev
	# ls -alF dev.db
	-rw-r--r--  1 root  wheel  32768 Sep 13 05:06 dev.db


	[ANALYSIS REPORT]
	Because of optimaization for fs, db/hash looks block size on specified
	file in src/lib/libc/db/hash/hash.c.  db/hash's block size limit is
	MAX_BSIZE(=65536) in if (hashp->BSIZE > MAX_BSIZE) ..... But, I can't
	understand others:-).

	I think db1 has boundary condition problems.

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message