Openwall GNU/*/Linux - a small security-enhanced Linux distro for servers
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 9 Aug 2018 14:51:20 +0200
From: Solar Designer <solar@...nwall.com>
To: oss-security@...ts.openwall.com
Cc: Matthew Garrett <mjg59@...gle.com>
Subject: Re: Linux TCP implementation vulnerable to Denial of Service (CVE 2018-5390)

Hi,

A co-moderator had rejected Stiepan's message since it "does not provide
any additional content to oss-security readers".  I'm also unhappy about
that, as well as about the focus on legal aspects in Stiepan's postings
in here in general.  However, the message raises an on-topic question
(the request for more detail) and brings up an on-topic issue (the
semi-embargo potentially causing harm).  I feel strongly about us not
getting into censorship, and I feel that rejecting this message would be
it.  So I went for the effort of manually restoring the already-rejected
message into the moderation queue, then approved it.

On Thu, Aug 09, 2018 at 07:12:27AM +0000, Stiepan wrote:
> Could you please provide some more details on the issue?

I agree that more detail must have been posted in here, especially given
that such detail was on linux-distros.

The issue is now also public via CERT:

https://www.kb.cert.org/vuls/id/962459

which links to:

https://git.kernel.org/pub/scm/linux/kernel/git/davem/net.git/commit/?id=1a4f14bab1868b443f0dd3c55b689a478f82e72e

which includes the following detail:

---
Juha-Matti Tilli reported that malicious peers could inject tiny
packets in out_of_order_queue, forcing very expensive calls
to tcp_collapse_ofo_queue() and tcp_prune_ofo_queue() for
every incoming packet.

With tcp_rmem[2] default of 6MB, the ooo queue could
contain ~7000 nodes.

This patch series makes sure we cut cpu cycles enough to
render the attack not critical.

We might in the future go further, like disconnecting
or black-holing proven malicious flows.
---

The CERT Vulnerability Note also talks about a related issue in FreeBSD.

Partial timeline for this issue as I became aware of it is as follows:

2018/07/23 - the commit referenced above
2018/07/23 - notification from CERT to some distros
2018/07/23 - grsecurity tweet linking to the commit
2018/07/27 - posting to linux-distros
2018/08/06 - CERT Vulnerability Note published
2018/08/08 - posting to oss-security

Of course, I am unhappy about this semi-embargo, and even more unhappy
about the semi-violation of linux-distros list policy on only having
non-public issues in there.  However, with CERT involved and with
related issues affecting more than just Linux, there was little I could
do, short of playing full BOFH and breaking the semi-embargo for
everyone.  While I think that would have been for the general public's
benefit overall, I didn't feel about it strongly enough to actually do
it this time.  I apologize for letting this happen.  (At the same time,
I did force another semi-public issue to oss-security right away since
that one didn't involve coordination with so many parties.)

It appears that everyone involved, including the CERT people, Matthew,
and others commenting on the linux-distros thread, were unhappy about
the publication delay.  No one I saw said that they wanted the delay.
Yet somehow CERT didn't pull the trigger sooner.  I guess two weeks
feels very soon for CERT as it is, even if it is a very long embargo for
linux-distros.  Also, I guess the discoverer/reporter of the issue had a
say on it behind the scenes, and other related issues and non-Linux were
considered in CERT's decision-making.

I am also unhappy about the two-day delay between publication of the
CERT Vulnerability Note and the mandatory posting to oss-security (it's
mandatory since the issue was on linux-distros).  I've been pinging
off-list to make this happen at all, and would have probably made the
posting myself if it didn't happen for another day.

> About the same period, our secure e-mail provider suffered an unprecedented DDoS with some e-mail messages never reaching us.
> Since this has business impact,

This is almost certainly unrelated.  (And I dropped the CC's to
ProtonMail and ITU on this reply, not to spam them with further
discussion of the unrelated issue.)

> we consider legal action against the opaque Linux-distros vulnerability-disclosure-among-friends-for-fun-and-profit scheme, that we exposed at the ITU earlier this year. This is digital divide in the works, with real impact for non-club-members.

Personally, I strongly oppose legal threats (let alone action) in our
community.  The way I see it, what we have is primarily a matter of
different opinions on how to handle security issues best, and most
people are genuinely acting the way they think works best for everyone
affected.  With many parties involved in coordinating a disclosure, it
usually becomes difficult.  There isn't necessarily a right or wrong
here.  But whoever brings legal action is definitely wrong.

Ironically, Stiepan had also suggested (here on oss-security a while
ago) that we apply for funding for running the (linux-)distros list (and
I explained in a reply why we shouldn't).

Alexander

Powered by blists - more mailing lists

Your e-mail address:

Please check out the Open Source Software Security Wiki, which is counterpart to this mailing list.

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.