Subject: Re: System freeze
To: Teemu Rinta-aho <>
From: Quentin Garnier <>
List: current-users
Date: 01/13/2005 15:19:40
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Thu, Jan 13, 2005 at 04:02:29PM +0200, Teemu Rinta-aho wrote:
> Christos Zoulas wrote:
> >| Jan 13 00:27:09 dawn /netbsd: vr0: unable to allocate Tx cluster
> >|=20
> >| Well, vr seems to be kind of unstable.  I've seen multiple panics in
> >| xhalf() at bootup (during dhclient) since I have this card (more than
> >| a year), but it hadn't caused any more issues up until now :(  Dunno
> >| if this has something to do with my recent upgrade from 2.99.10 to
> >| 2.99.12.
> >
> >>From the messages it appears that something is leaking memory. You might
> >want to check the code to make sure that it does not leak.
> Somewhat related to this discussion:
> yesterday I ran nessus against my firewall (twin-vr, netbsd-current
> and pf), and nessus reported that it had detected the "Etherleak"
> vulnerability, where ethernet frames are padded with data from some
> other data packets. I checked the vr driver, but it seemed like the
> padding for the ethernet frame is set to zero, and there should
> be no problem. I didn't yet have time to investigate whether the
> problem is really in the vr driver or in nessus. Just wanted
> to mention this as it might be related to the other problems.

It's possible that the vr driver leaves the checksum at the end of
the Ethernet frames.  Supposedly it should remove it, but...  A few
drivers have that bug.

FWIW, on my vr device, tcpdump shows checksums on incoming packets,
but outgoing packets seem correctly padded.

Quentin Garnier - -
"Commala-come-five! / Even when the shadows rise!
To see the world and walk the world / Makes ya glad to be alive."
Susannah's Song, The Dark Tower VI, Stephen King, 2004.

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

Version: GnuPG v1.2.6 (NetBSD)