Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 8 Jul 2001 03:31:59 -0500
From:      John Cappiello <john@apt202.net>
To:        freebsd-stable@freebsd.org
Subject:   Re: Headsup: Tcp ISN generation changes from 4.2 to 4.3
Message-ID:  <20010708033159.A19667@localhost>
In-Reply-To: <20010707002821.A18599-100000@achilles.silby.com>; from silby@silby.com on Sat, Jul 07, 2001 at 10:30:26PM -0500
References:  <20010707002821.A18599-100000@achilles.silby.com>

next in thread | previous in thread | raw e-mail | index | archive | help

--OXfL5xGRrasGEqWY
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Sat, Jul 07, 2001 at 10:30:26PM -0500, Mike Silbersack wrote:
>=20
> If you are running 4.3-release or later, please read the entirety of this
> message; you are probably affected.
>=20
> Shortly before 4.3-release was tagged, the tcp initial sequence number
> generation scheme was changed to fix a weakness which could allow an
> attacker to reset connections.  At that time, it was known that the fix
> would also break TIME_WAIT handling.  The impact of this breakage was
> expected to be small.  Unfortunately, recent reports on -net seem to
> indicate that the breakage is widespread.
>=20
> Consequently, I have just committed patches to -current and -stable which
> allow you to select the tcp initial sequence number generation scheme used
> by your system.  If you don't wish to cvsup just for this patch, you may
> instead obtain it from http://www.silby.com/patches/multiple_isn_schemes.=
patch
> and manually apply the patch yourself.
>=20
> Once a patched / updated kernel is installed, you can change generation
> schemes with the sysctl net.inet.tcp.tcp_seq_genscheme.  0 =3D the older
> random positive increments, 1 =3D the newer randomized scheme.
>=20
> The newer scheme causes problems in cases where a FreeBSD is making many
> outgoing connections per second to the same host.  For example, you may
> have a box which connects to a backend SQL server.  If you have such a
> setup, you are probably seeing many rejected connections each second /
> other oddities in connection setup.  If this is the case, you should
> update and toggle the system back to random positive increments.
>=20
> The newer scheme causes no problems in accepting incoming connections.  As
> a result, you will probably see no problems if your servers mainly handle
> incoming requests and do not make many outgoing requests of their own.
>=20
> This sysctl will only be temporary.  Once a secure _and_ compatible
> initial sequence number generation scheme is implemented, it will become
> the default.  This will be at least a few weeks away, however.  If you are
> seeing the problems described above, you should cvsup (or patch) and flip
> the sysctl now, rather than wait.

I'm sorry, but I just want to make sure I have this right.  I checked my lo=
gs but did not see any report of this problem.  Is there anywhere specific =
I should be looking for this error?  I just don't want to assume I'm not af=
fected by this, and be wrong.


>=20
> Mike "Silby" Silbersack
>=20
>=20
>=20
> To Unsubscribe: send mail to majordomo@FreeBSD.org
> with "unsubscribe freebsd-stable" in the body of the message
>=20
>=20
>=20

Thanks,

John


--OXfL5xGRrasGEqWY
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org

iD8DBQE7SBp/7tlm0/lPd78RAiw+AJwMZtAPcZADWYiIJrOU6NcZ61EB1QCbBr2X
o0AkchY0BLMA5O9lp5mZQ/g=
=vqed
-----END PGP SIGNATURE-----

--OXfL5xGRrasGEqWY--

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




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010708033159.A19667>