*BSD News Article 94031


Return to BSD News archive

Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!munnari.OZ.AU!news.ecn.uoknor.edu!feed1.news.erols.com!cpk-news-hub1.bbnplanet.com!su-news-feed4.bbnplanet.com!news.bbnplanet.com!coop.net!pacifier!deraadt
From: deraadt@theos.com (Theo de Raadt)
Newsgroups: comp.unix.bsd.bsdi.misc,comp.unix.bsd.misc,comp.security.unix
Subject: Re: *BSD* Security WWW/Mailing List?
Date: 21 Apr 1997 01:55:00 GMT
Organization: Pacifier BBS, Vancouver, Wa.  ((360) 693-0325)
Lines: 131
Message-ID: <DERAADT.97Apr20195500@zeus.pacifier.com>
References: <3356E1CC.299E@softway.com.au> <DERAADT.97Apr18181055@zeus.pacifier.com>
	<5jdgaf$34i@cynic.portal.ca> <DERAADT.97Apr20113509@zeus.pacifier.com>
	<5jeb2b$ata@cynic.portal.ca>
NNTP-Posting-Host: zeus.theos.com
In-reply-to: cjs@cynic.portal.ca's message of 20 Apr 1997 17:06:35 -0700
Xref: euryale.cc.adfa.oz.au comp.unix.bsd.bsdi.misc:6687 comp.unix.bsd.misc:3024 comp.security.unix:33773


In article <5jeb2b$ata@cynic.portal.ca> cjs@cynic.portal.ca (Curt Sampson) writes:

   >   As has OpenBSD. Although I notice your FTP server still has no way
   >   of doing anonymous uploads that are secure from abuse by warez-traders.

   I am not pointing this out to get into a spitting match; I'm pointing
   this out to show that you too have holes that are not closed, Theo.

Please describe a hole that can be fixed in _software_.  Putting a
writeable directory in your anonftp directory is a system admin error.
Of course, since you have the ftp bounce attack in your ftp daemon,
it's a hell of a lot worse of a problem for you!

By the way Curt -- what is your solution?  I see no fix for this in
the NetBSD ftpd.

   You regularly insinuate that you have fixed holes that others
   haven't, but you never describe the exact holes you fix: you just
   say `look at our change logs,' which don't describe the attacks at
   all.

Go read bugtraq.  Through postings made by Secure Networks (our friends)
we have made the world aware of a number of problems.  But come on.


   Without a proper analysis of the attacks, which you refuse to
   provide, there's no real evidence that many of the things you have
   `fixed' were ever broken.

HAHAHAHA!

`proper analysis'. Hahahahahaha.

We find a buffer overflow.  We fix it.  We don't _CARE_ if it is
exploitable; we just fix it.  Sometimes we later find it is
exploitable.  Sometimes it's immediately evident.

Sometimes it is just a bug that I can SIGSEGV a program with.  Know
what? I hate core files lying around my machine too.  But I guess I
should be doing some `proper analysis' before fixing those.

Curt, you're way out there!

Let me provide one of my favorite examples.  In August I went over
lpr.c with Todd and we fixed a couple of buffer overflows.  None of us
checked for exploitability (see, sometimes we do, sometimes we don't).
Then we shipped our 2.0 release.  Ho hum, time passed as it tends to.
About 4 months later someone posted an article to Bugtraq. It included
source to an exploit.  Wham, root. Neat.

We had shipped a release months earlier.  We didn't have the hole!  I
saw the exploit code and I had to ASK the author if it still worked in
OpenBSD, because I thought it was a new one!  It wasn't a new hole.
We had it fixed.  (The exploit author was actually a little bit upset
that he hadn't been the first to find it ;-)

There are many other examples.

In essence we are fixing holes before they become KNOWN holes.  That
is what being CAREFUL is all about.  Obviously you know zero about how
an operating system becomes secure.  Through diligence and paranoia,
by fixing BUGS OF ALL KINDS.

I invite you to step onto the podium and suggest that going through
the source tree carefully looking for problem areas and fixing them
isn't going to close a few holes.  Or a lot of holes.  To be honest,
it's a lot of holes.  They fall into classes, and in some classes like
mktemp there are perhaps 500 in the tree, of which perhaps one quarter
are exploitable for some nasty effect.

One was so nasty that a BSDi person got extremely upset that we hadn't
told them beforehands.  Heh, that /etc/security bug sure was funny.  I
presume you have it fixed, right?

But now, if you'd rather chat about security problems before they hit
you in the face, hey, all I can suggest is you start running your
backups more often.  I certainly am not going to waste my time talking
to anyone with such a careless attitude towards fixing problems which
are primarily just simple bugs with bad and `exploitable' side
effects.

Fixing a buffer overflow or a writable directory race isn't rocket
science!

   >And the source routing controls in the kernel appear
   >completely insufficient compared to the threat.

   A typical insinuation. What exactly does the OpenBSD kernel do with
   source routed packets that the NetBSD kernel doesn't, Theo?

It drops them dead, and it bitches too.  Always.  It won't forward
them and it won't accept them.  And if you are silly enough to ask the
kernel to let them through, a bunch of applications have been
carefully modified (AND FIXED if they did it wrong, grr, including
sendmail, hint, hint) to specifically and carefully deal with them.
Is your rshd safe?  Did you not see the posting about Neils Provos'
new source routing attack?  The secnet posting about the new source
routing attack mentioned a few definately vulnerable programs but
didn't mention anything about other possible programs which might be.
Hmm, I wonder if there are any.  Do you feel safe?

   >I look forward to the day we are able to look at cvs logs for the
   >NetBSD source tree!

   So do we all. But I think that's been gone over already.

Let's go over it again.  There are two possible reasons why NetBSD
doesn't make their CVS tree available to the public like all the other
free OS groups do (OpenBSD, FreeBSD, Linux to a limited extent too
now)
	1. Either they have bad things in there.
	2. Or they don't want to let it out of their grubby hands.

Are there any other reasons?  Oh right -- there's a pr1v4t3 reason and
there is a s3cr3t document with USL!  Some free operating system.

   >The fixes are there, and they are shared with the world.

   The fixes are not as important as proper descriptions of the problem,
   Theo. It's very difficult to work back to the original problem from
   your fix, and thus even more difficult to verify that your fix is
   correct.

Balony.  You are making a fool of yourself.  The fixes are clearly
more important.  You are too lazy to read the diffs, and you would
like to blame us for your slackness.

--
This space not left unintentionally unblank.		deraadt@openbsd.org
www.OpenBSD.org -- We're fixing security problems so you can sleep at night.
(If it wasn't so fascinating I might get some sleep myself...)