About LetoAms

Cypherpunk, DNSSEC. Fedora, openswan, crypto, pinball, speculaas, opensource software,

Why the Sun 2 has the message “Love your country, but never trust its government”

(UPDATE: Sevan Janiyan pointed out the Sun 2 was not a SPARC system)

Yesterday, Alec Muffet posted a few tweets regarding the Sun SPARC 2 bootloader’s DES routines.


Alec figured that message was never supposed to be seen and suggested it was a kind of silent protest of someone in Sun against the US Government. I replied, saying I was pretty sure such a message anywhere in the Sun bootprom code must have originated by John Gilmore. So I asked John, and he did not disappoint. This is what I wrote me back:

> That must have been you? :)

Yes. Vinod Khosla, first President of Sun, came to me at one point
and said to put something hidden, triggered in an unexpected way, into
the ROM Monitor, so that if somebody cloned the Sun Workstation
(violating our software’s copyright), we could do that unexpected thing
to the competitor’s demo workstation at a trade show and thereby prove
that they had cloned it.

The ROM Monitor command that printed “Love your country, but never
trust its government.” was “k2” followed by control-B and RETURN. (To
get to the ROM Monitor from the running Unix display, you had to hold
down the L1 key and hit “A” then release both; the monitor had special
code looking for that key sequence.) I had found that saying years
before on a hand-painted sign tacked up on a pole or tree in central
Pennsylvania, wrote it into one of my notebooks at the time, and
plucked it out as the hidden thing after Vinod asked. In the source
code it was obscured as a set of hex numbers, but in the binary it was
visible. (I didn’t bother to XOR it with something to make it more

The circumstance Vinod was concerned about never did come about;
nobody stole the Sun boot roms. Early 680x0s didn’t come with a
standard MMU, so everybody who wanted virtual memory had to invent
their own, and our roms were very tied to our custom static-ram based
zero-wait-state MMU, which was also patented. A few companies
licensed the board design from us, like Imagen for a laser printer
controller, but they had a license to use the ROMs.

The DES chip slot was intended to speed up DES for high security
networking applications, and we did get it working. I think Bill Joy
was the one who made it a part of the architecture, all the way
through the Sun-3, thinking that we’d use it as part of securing the
network file system. But the chip was expensive and
export-controlled, and our software never really used it (NFS ran in
plaintext and used the sender’s IP address for authentication!). So
it didn’t get stuffed on the production boards, and eventually they
stopped stuffing the support chips too. Tom Lyon wrote a nice device
driver for it, and a “des” user command that would use the chip for
file encryption if it existed (or fall back to software).
In one 68010 based model we also put in a slot for an Intel floating
point chip — I think the one that came with the 80286. We got it
working (I did the initial debugging) and fed it commands and
arguments via manual peek/poke in software. Doing that was somewhat
faster than the software floating point that we were otherwise using,
despite the overhead. But as I recall, Intel wouldn’t sell us the
chips in volume, because they didn’t want to make the 68000 more
competitive against their own x86 chips. So we had to wait til the
68020/68881 came out (Sun-3 era) before we had fast floating point.

It’s nice that Matt Fredette made a Sun-2 emulator. I may have
images of old SunOS release tapes that might work in it. And I have a
Sun “FE Handbook” for the field engineers, that has a lot of the
details about what chips go where, what the jumpers on each board
mean, what the memory map of each board looks like, and etc.

You can forward this info on to whoever…


66% of VPN’s are not in fact broken

So the weakdh.org article got some renewed press again, and it keeps coming up with this rather misleading paragraph:

We further estimate that an academic team can break a 768-bit prime and that a nation-state can break a 1024-bit prime. Breaking the single, most common 1024-bit prime used by web servers would allow passive eavesdropping on connections to 18% of the Top 1 Million HTTPS domains. A second prime would allow passive decryption of connections to 66% of VPN servers and 26% of SSH servers. A close reading of published NSA leaks shows that the agency’s attacks on VPNs are consistent with having achieved such a break.

That’s a rather bold statement. How do the authors know the MODP groups of 66% of VPN servers? This is what they write in their research paper:


We measured how IPsec VPNs use Diffie-Hellman in practice by scanning a 1% random sample of the public IPv4 address space for IKEv1 and IKEv2 (the protocols used to initiate an IPsec VPN connection) in May 2015. We used the ZMap UDP probe module to measure support for Oakley Groups 1 and 2 (two popular 768- and 1024-bit, built-in groups) and which group servers prefer. To test support for individual groups, we offered only the single group in question. To detect default behavior, we offered servers a variety of DH groups, with the lowest priority groups being Oakley Groups 1 and 2.

The first problem is that the first packet exchange performs the DiffieHellman and no IDs or credentials are sent. This means that the server in question does not yet know which configuration this connection maps to, unless it is limited by IP. If it is limited by IP, then their probe likely did not even solicit an answer because they came from the wrong IP – Cisco VPN servers tend to not give error messages and will just drop the packet. If the configuration is not limited by IP, because the connection supports roaming users, then the VPN server cannot yet reject the connection based on a weak MODP group. Imagine the following configuration (in SWAN ipsec.conf syntax):

conn regularusers

conn oldcisco
  rightid="CN=knownweakdevice.com, O=OldVendor, C=CA"

When the weakdh people’s scanner showed up, it allowed the initial request/response packet to be 1024. But once it got further into the negotiation, if would authenticate and identify the remote peer, and refuse to establish the IKE connection with the modp1024 group for everyone except the one known buggy device. But since the scanner had no credentials, it never got that far. So this is a false positive. This is how freeswan and openswan worked, and how libreswan works. When it receives more information, it tries to switch connection to the best matching connection. Note that with only the above two configurations enabled, libreswan would immediately reject an incoming IKE exchange that tries to use modp768 or modp2048, since no loaded connection specifies that modp group. Note the default when not specifying an IKE line is modp2048 and modp1536 for IKEv2, and modp1536 and modp1024 for IKEv1. With a preference for the higher group.

This method of connection switching requires the IKE daemon to know or go through all the connections to look at the modp group. If I understood the strongswan implementation correctly, it does not even bother to reject the incoming request based on modp group. It will just answer whatever the modp group is on the first packet. Later on in the exchange, once it has committed to a specific configuration, it will check for the modp group and reject the connection. So that is false positive number two for the scanner.

The article continues:

Of the 80K hosts that responded with a valid IKE packet, 44.2% were willing to accept an offered proposal from at least one scan. The majority of the remaining hosts responded with a NO-PROPOSAL-CHOSEN message regardless of our proposal. Many of these may be site-to-site VPNs that reject our source address. We consider these hosts “unprofiled” and omit them from the results here.

Well, that means the authors just excluded all freeswan/openswan/libreswan servers that are configured to not allow modp1024 because that is exactly what they would return. Libreswan per default does not allow modp1024 in IKEv2, so this is a rather big false negative!

We found that 31.8% of IKEv1 and 19.7% of IKEv2 servers support Oakley Group 1 (768-bit)

In our sample of IKEv1 servers, 2.6% of profiled servers preferred the 768-bit Oakley Group 1

Now regardless of the strong connections the authors missed, they did find a large number of weak ones. There is no excuse whatsoever for an IKEv2 server to allow modp768 or modp1024. It’s even questionable to allow modp1536 because every IKEv2 implementation supports modp2048. The 31.8% of modp768 is also inexcusable. The freeswan/openswan implementations never supported modp768 unless you specifically recompiled it (and the distro versions never had it compiled in). libreswan removed the compile time option altogether. The modp1536 group made it into freeswan – and preferred over modp1024 – before the modp1536 group was actually standardized in an RFC! So unless you specifically configured any of the *swan IKE implementations to use these weak groups, and your remote clients didn’t start out with a weak group, you would be using modp1536. Even back in 1997. This has not been an IKE problem in opensource IKE software ever. But, there is a catch.

The insecure Cisco administrator

86.1% and 91.0% respectively supported Oakley Group 2 (1024-bit)

This statistic matches my 15 year experience with IKE very well. It is very unfortunate that administrators fear reconfiguring a Cisco’s IPsec configuration. As such, when you need to interop with a Cisco, the other party usually has a SOP document that tells them how to configure the Cisco for use with 3DES-SHA1 modp1024. You are lucky if you are not forced to use md5. This is not a technical problem – Cisco VPN Concentrators can do modp2048. It is the humans that are afraid to touch something they are not familiar with that works.

MODP groups in the near future

The ipsecme working group at the IETF is actually working right now to update the set of mandatory to implement algorithms. This also includes promoting new algorithms to SHOULD or MUST and demoting others to MAY or MUST NOT. But it will only do so for the IKEv2 protocol. Most vendors have locked their IKEv1 stacks and will not make any changes to it anymore. So IKEv1 will always stay using modp1024 and modp1536. So if you can, use IKEv2.


The 66% of VPN number is surely very wrong. Measuring that number is per definition impossible, due to the common implementations of IKE daemons that hide the acceptance and rejection of the modp groups from un-authenticated IKE clients. However, I do agree that there are too many old IKE servers deployed that need to have their configurations upgraded to allow stronger modp groups and stronger encryption protocols.

If you want to go beyond modp1536, you should really upgrade to IKEv2. You will also need to move to IKEv2 if you want to move from AES-SHA1 to AES_GCM (or soon chacha20poly1305). Any IKE implementation that does not support IKEv2 is not actively maintained and should not be used.

See also my earlier article weak DH and LogJam attack impact on IKE

UPDATE: A more detailed explanation of the NO_PROPOSAL_CHOSEN case

The way IKE is specified and the way it works in practice is slightly different. The theory of operation is that a client starts with a list of proposals, that can include multiple DH modp groups. But that initial packet already contains the KE payload for one of the DH modp groups. So the client might send a proposal set containing 1024 + 1536 + 2048, but it only sends one KE payload. That KE payload basically denotes the client’s default. In this example, that would be either 1024 or 1536 or 2048. If the server receives this KE + proposal, it needs to make a decision. If the KE modp group is acceptable, use that. If not, then look if the proposal contains any other modp groups it is willing to use. If it does, it should return INVALID_KE with a payload of the acceptable modp group to use. The client receiving this can then select the other modp group it is willing to use and start the IKE exchange from scratch with a new request. (It’s a little more complicated because this reply is unauthenticated, so the client might wait for a bit to see if a “real server” accepts its modp). In practise, this often does not work. For example iphone ios8/ios9 do not process INVALID_KE payloads at all, as they assume a single correct modp group is configured on both sides using their enrollment program. So in practise, the server will accept the client’s DH if it is one of its allowed groups. This can mean that even though client and server both can do a more secure DH group, they end up using the most commonly used shared group. This is usually 1536 or 1024 for IKEv1 and 2048 for IKEv2.

An additional issue with IKEv1 is that the first packet also contains the OAKLEY_AUTHENTICATION_METHOD. If this is mismatched (eg PSK vs RSA) the IKE server will also return NO_PROPOSAL_CHOSEN.

And both both IKEv1 and IKEv2, the initial packet contains encryption/integrity algorithms too. If these mismatch (eg AES vs 3DES or SHA1 vs MD5) then the IKE server will also return NO_PROPOSAL_CHOSEN.

You can test the NO_PROPOSAL_CHOSEN return directed at vpn.nohats.ca. It is configured to accept connections from anyone, but requires modp1536 for either IKEv1 or IKEv2, using X.509/RSA authentication. You can send it PSK requests for modp1024 and get NO_PROPOSAL_CHOSEN.

DNSSEC prevented NetworkSolutions from stealing customer email

A large dutch ISP ran into issues with their Registrar Network Solutions over the weekend. (why anyone would be using NetworkSolutions like it is 1993 is beyond me)

A misunderstanding over a payment caused hundreds of domains to be entered into PENDING-DELETE state. One such example is puiterwijk.org. Some of these domains have been in this state now for days.

What made things worse is that said NetworkSolutions took over running DNS for these domains, including an MX record that points to an actual mail server!

So if you were lucky, your email just bounced. If unlucky, someone else got your emails. The TLS certificate on that mail server doesn’t even match their hostname, so anyone on path can also just MITM it and a traceroute shows 26+ hops all over the place.

However, they did not modify the DS records after taking over the NS records and MX/A records. So those domains that used DNSSEC, including the above mentioned puiterwijk.org, are not resolving at all because the validators are rejecting the NetworkSolutions stolen DNS zones as bogus. So, my emails to this person were not delivered to the rogue MX servers because both he and I deployed DNSSEC. Hoorah, I guess….

Now, taking over MX and causing email failures like this is pretty evil. I would hope this violates some ICANN or PIR agreement but as said Registrar has been a sad registrar since uhm about 1993, I guess nothing is going to change. Let’s hope the Dutch ISP has learned from this and will move to another Registrar soon after this mess gets resolved.

The weak DH and LogJam attack impact on IKE / IPsec (and the *swans)

TL;DR The LogJam downgrade attack does not apply to MODP groups in the IKE protocol, only to TLS, so IKE or IPsec is not impacted.

If you are using libreswan you are not vulnerable to weak MODP groups and using MODP2048 per default unless specifically configured for a lower MODP group.
If you are using openswan with IKEv2 you are using MODP2048, but if you are using IKEv1 you are using MODP1536 which is still much stronger than MODP768 or MODP1024.

Libreswan as a client to a weak server will allow MODP1024 in IKEv1 as the least secure option, and MODP1536 in IKEv2 as the least secure option.
Openswan does not properly implement INVALID_KE, so it cannot connect to another DH group than the one it started out as, so it runs the risk of getting locked out if the server side bumps their minimum MODP group to 2048. openswan defaults to MODP1536 in IKEv1 and MODP2048 in IKEv2

(minor update: Hugh Redelmeier corrected me that freeswan never supported modp768 – the smallest modp was 1024)

A bunch of smart crypto research released a paper describing the LogJam attack. It is a cryptographic attack on weak DiffieHellman groups. It talks about DH in various protocols. It is an interesting read. When you break the DH layer, you obtain the secret session keys needed to decrypt the various protocol (TLS, IPsec, SSH, etc) sessions. The conclusion from the paper is that any DH group smaller than 1024 should not be used as even non-nationstate attackers can obtain the resources needed to break these. They further conclude that they believe it very likely that the standarized 1024 bit DH MODP group has been broken as well by the NSA. It involves large scale computing clusters and sieving and doing a lot of heating up the planet. They can do this because in crypto standards, everyone uses the same MODP prime values for the DH exchange.

First, a small clarification. When the paper talks about IPsec, they mostly mean the IKE keying protocol, not the ESP packet encryption protocol. IKE is responsible for the secure keying operation that generates symmetric session keys used for the actual packet encryption in the ESP protocol. IKE is the userland daemon, ESP is the kernel level encryption/decryption.

The DH values for IKE are listed in various RFC’s, and can be found in the IANA registry for the older (but more used) IKEv1 GROUP values and for the newer but less used IKEv2 GROUP values. For the published attack, the only type of groups relevant are the MODP groups. The ECP groups do not use exponentiation and thus are not vulnerable to factoring primes.

IKEv1 MODP groups

In IKEv1, the most common modp groups are modp1024 (DH2) and modp1536 (DH5). Some people still want to use modp768 (DH1) but it has strongly decreased over time. Most IKEv1 stacks also support modp2048 (DH14) up to modp8192 (DH18) although notably Cisco administrators in general seem to have stuck with modp1024 and modp1536, just as they have stuck with 3DES and MD5/SHA1. This is mostly because the technical knowledge to configure Cisco’s for VPN is so difficult, that once a Standard Operating Procedure(SOP) is in place, no one wants to, or is allowed to, modify the configuration used. (It’s also the tail end of Cisco asking for extra money to support AES many years ago)

FreeS/WAN never supported modp768 – John Gilmore had good ideas about scaling and NSA computing powers, he was one of the people behind Deep Crack, the EFF 1DES cracker. Openswan which was forked from FreeS/WAN in 2003 had support for modp768 (and 1DES) but did not compile it in per default and neither did any of the Linux distributions that shipped it. But people needed it frequently enough to interoperate that the code was added, but disabled per default. Notably, embedded people using the *swan’s would enable it due to customer demand. It annoyed us so much that sometime around 2005 (I can’t find it in the changelog) we just ripped it out. Of course, it did not prevent people from patching it back in, which some did. Most likely, many unattended embedded devices out there are still doing modp768 bassed VPNs, and as LogJammer shows, that’s within reach of academia and banana republics – so count on these being decrypted routinely by the Five Eyes, Russia, China, and probably most other Western TLA’s.

The default IKEv1 modp group for openswan (and libreswan) was 1536. In July 2014, libreswan changed this to MODP2048. I believe openswan is still using modp1536 (and if you are still using openswan, I recommend that you switch ASAP, see this git commit activity comparison

IKEv2 MODP groups

IKEv2 has seen a very very slow start. I’ve only seen it really being deployed in the last two to three years. Most of these deployments use at least MODP2048, do not support MODP768, but are usually willing to do MODP1536 or even MODP1024. The latter is mostly done for fear of compatibility issues. But also to support an upgrade path for all devices. If your VPN connection supports “either IKEv1 or IKEv2” and your clients are still connecting with IKEv1, then most surely you will need to support MODP1536 and even MODP1024, or you will be losing your older VPN clients.

In April of 2008, as the then release manager of Openswan, we released version 2.6.12 which changed the IKEv2 default to MODP2048. This was a controversial decision – one developer quit the openswan project over this change because he deemed it would cause too many interoperability failures because many IKE implementations did not properly implement INVALID_KE renegotiation or simply did not support MODP2048. No problems emerged after the change, but most likely because simply no one but the certification checkbox people were even testing or running IKEv2 at the time. Since then, libreswan (but not openswan) has implemented proper INVALID_KE renegotiation.

RFC 5114 MODP groups

There is a semi-mysterious RFC 5114 – Additional Diffie-Hellman Groups document. It introduces new MODP groups not with higher sizes, but just with different primes. The justification in the RFC states

The purpose of this document is to provide the parameters and test
data for eight additional groups, in a format consistent with existing RFCs.

It lists one new 1024 bit MODP group and two 2048 bit MODP groups, when we already had RFC 3526, which has MODP groups up to 8192 bit. The odd thing is that when I talked to people in the IPsec community, no one really knew why this document was started. Nothing triggered this document, no one really wanted these, but no one really objected to it either, so the document (originating from Defense contractor BBN) made it to RFC status. Due to this strange history, the RFC 5114 MODP groups are not included in any default proposal set and MUST be manually configured before libreswan will use these groups. We had to support these groups due to demands for RFC compliance and certifications.

IKE and IPsec are not that bad you know

Not that I want to go into a nahnahnahnahnahnah mode, but to those people that keep dissing IKE and IPsec, note that we were again not vulnerable unlike TLS. And when it comes down to cipher mode attacks, IKE has proven to be far more secure than TLS has been in the last three years. So please stop beating up on IKE/IPsec citing some old papers from the late 90’s. Kthanks

Some alt.fan.pratchett quotes from the Usenet days

I’m sad to hear Terry Pratchett has died. I would write “it is not fair” in all caps, but that might get misinterpreted.

I had the pleasure of meeting Terry Pratchett at a book signing in The Netherlands when “The Last Continent” was released. I was getting the book for an Australian friend of mine whom I was about to visit in Australia. She had told me she refused to read Pratchett because all her friends read his books and he was just “too mainstream”. I told Terry this at the book signing when handing him The Last Continent. He thought for a moment and wrote “Carol, Don’t read this one either”.

I greatly enjoyed his books, but also at the time in the late nineties, his postings on Usenet in alt.fan.pratchett. He was the first celebrity that I saw that talked to his fans without a publisher in between. You got to talk to the real Terry. Of course, a lot of talk on alt.fan.pratchett was about whether or not that was really Terry Pratchett. You could tell from the writing though. It was him.

Below are some of the quotes I saved from alt.fan.pratchett, hopefully someone can grab the entire usenet archive from Google before it is lost forever……

on Cuba
I would like to confess that I know my books have been read by a Cuban. At least, I think he is Cuban. Certainly he’s got Cuban relatives. Or knows someone from Cuba. Or knows someone who smokes cigars, at least.

I realise this may put the upcoming US tour in jeopardy, and throw myself on the mercy of the courts.

On the use of IMHO
Oh, as a naturally democratic man I use the simple and common net usuage; it means ‘In my pompous and unwarranted opinion as a person who, despite an apparent intelligence level somewhere between that of a clam and a line-dancer, won’t hesitate to pronounce definitively on any damn subject under the sun.’ Of course, I might have got it wrong. It was just based on long observation:-)

On a hedge
You’ve got to remember that no hedge is ‘natural’; the whole point of cutting and laying the traditional quickthorn is to make a living barrier that even a sex-crazed ram isn’t going to be able to push through.

On Tomb Raider
> Same here, maybe I’d like her more if she didn’t look like she was made of plastic.

Never mind about Lara. The Tomb Raider games are the only games where I’ve shot someone because they were getting in the way of the scenery. Who will forget the brown trousers moment with the T-Rex in TR1? And in TR2 I spent some time driving the snomobile and the boat just for the sheer heck of it…

On Doom
“Over the centuries, mankind has tried many ways of combating the forces of evil…prayer, fasting, good works and so on. Up until Doom, no one seemed to have thought about the double-barrel shotgun. Eat leaden death, demon…

On Doom at a con in NZ
At a con in NZ they had a number of machines hooked up and four kids were playing Doom in co-operative mode with the superb Aliens-TC patch, which was incredible well done. They were advancing through the nest, and I explained to them that the only way they’d survive was with some sense of military discipline. Hah, they obviously thought, was does *he* know?

Whereupon an alien stepped out of the corridor wall behind them.

Whereupon they all turned around and fired like Hudson on speed…

Whereupon the guy who had been at the *front* of the line shot all three of his colleages who had been in single file behind him…

…whereupon the alien ate him.

On receiving emails from fans
But in reality people often aren’t writing because they want that information. They want to write an email to a favourite author, and they don’t want to sound too fanboy, so asking a real question gives them, they feel, a reason. This is fine. It’s just that there’s so many :-)

About people posting ideas for DiscWorld in alt.fan.pratchett
You tell *me* what to do, folks. My view is that a) I’m not intending to nick anyone’s suggestions b) I’m not intending to quit reading afp c) if there are ever problems, I can afford really expensive lawyers:-)

On vampire ideas
> I am talking about the carefree use of informal language such as “nicked”,
> “stolen”, even “plagiarised”, and of course the ever popular “it is clearly
> obvious that…”, when referring to things which are, in the end, usually
> entirely unconfirmed references or resonances we *think* we may have
> spotted in PTerry’s work.

A sort of moderating comment…(but best to add some spoiler space)

Probably no other fictional monster has been so written up, filmed and played as The Vampire. I read my way through a shelf of stuff while writing CJ and *of course* there are scattered through it references to the general vampire mythos — as I’ve said elsewhere, there’s little in there about vampires that I had to make up. Vampires didn’t begin and end with Dracula. There were even high-class, Lestat-type vampires in 19th Century fiction. There are *lots* of references in CJ to ‘Dracula’, both the original book and the various influential movies that have been made of it, because they’ve largely defined our image of the vampire (which, before ‘Dracula’, was quite different). And it wouldn’t be DW if I couldn’t take the pi– poke some fun at the Anne Rice wannabes who think vampires are so cool…

Of course, everyone in Transylvania — oops, Uberwald — has a servant called Igor. It’s practically compulsory. Does Igor show Count Duckula a picture gallery anywhere in the series? I don’t know, and I don’t care much — because I can recall ‘pictures of ancestors’ as minor background in vampire movies generally (sometimes the eyes move, and sometimes, ‘hey, did you know you look *just* like the girl who died 300 years ago and here’s her picture?’ Generally, of course, the vampire in the pictures always looks remarkably like the one doing the showing…) And I can’t quite bring myself to believe that ancient retainers grumbling that the new people aren’t a patch on the masters he used to have was invented by Cosgrove-Hall.

I can hardly object to annotations, but I did spend my youthful Satuday mornings watching endless bad (and good) vampire movies (it got so I could *recognise* that bit of road by the dark lake where, in Hammer Horror movies, the coach *always* loses a wheel) and before I wrote CJ I read of lot of historical and reference vampire stuff (as have most modern writers of fictional vampires). We all mine from the same seam. That’s why most vampire movies are remarkably similar — the only suspense is working out which *new* way of killing a vampire is going to be demonstrated this time.

Don’t stop using IPsec just yet

Today, Laura Poitras and Jake Appelbaum spoke at the 31C3 conference and in collaboration with Der Spiegel published an interesting article on VPN exploitation by the NSA.

The “TL;DR” summary of what follows below is: If you configure your IPsec based VPN properly, you are not affected. Always use Perfect Forward Secrecy (“pfs=yes” wich is the default in libreswan IPsec) and avoid PreSharedKeys (authby=secret which is not the default in libreswan IPsec). If you really need to use PSK, use a strong shared secret that cannot be brute forced. The NSA has their own version of IKEcrack running on millions of dollars worth of CPU’s. Also, the NSA sneaks into your router to steal your PSK’s so they can decrypt all your traffic.

Update 1: from media-35513.pdf (“TURMOIL/APEX/APEX High Level Description Document”):

CES generally requires the packets from both sides of an IKE exchange and knowledge of the associated pre-shared key (PSK) in order to have a chance of recovering a key for the corresponding cipher (ESP). A major goal of APEX is to access two sides of key exchanges of traffic of interest”

Update 2: To stop the NSA HappyDance, I just commited code to libreswan IPsec to warn about weak PSKs with a message notifying the sysadmin that as of July 1st, 2015, libreswan will refuse those weak PSKs. To determine the strength of the PSK, I used the Shannon Entropy value. If the PSK has a Shannon Entropy of less than 3.5, a warning will sound and in 6 months it will refuse to use that PSK.

Update 3: Just to make it clear – To break IKE PSKs, you first need to break the initial DiffieHellman exchange, which is usually MODP1024 or MODP1536 in the bad cases (and MODP2048+ in the good cases). The exception to this is IKEv1 Aggressive Mode, where the MAC computed with the PSK as secret key is sent
in the clear. As a result one can mount an offline dictionary attack.

IPsec is not IKE

First, let’s clarify things a little bit. Although everyone is talking about breaking IPsec, what they really mean is breaking the Internet Key Exchange (IKE) that is used to negotiate and create symmetric keys that are used with IPsec for encryption and decryption. IPsec itself can use various ciphers and algorithms. It can use a separate encryption (AES, CAMELLIA, SERPENT, TWOFISH, 3DES, etc) + integrity (SHA2, SHA1, AES_XCBC, MD5, etc) mode or it can use an AEAD cipher that combines these two into one (AES GCM, AES CCM, CAMELLIA GCM, etc)

But Jake said to stop using IPsec

None of what was said by Jake or published by Der Spiegel shows any attack on these ciphers or algorithms. Even the MD5 use within IPsec (HMAC-MD5) is not as weak as people usually believe it to be by confusing it with non-hmac use of MD5. The comments Jake made and the material in the published slides show attacks against weak IKE configurations, router compromises, and possibly against some IPsec hardware accelerators. Below are my comments on those slides that are interesting in the context of IKE and IPsec.


The focus of the VPN decryption team lies with IPsec and SSH. I think that fits the deployment. These two protocols are the most widely used encryption protocols to transfer bulk data. While the hacker community might focus on OpenVPN, that’s not where the real deployment is. The mention of PPTP is curious. From what I see, PPTP is basically dead as a VPN technology, although it is still widely used in Russia as an encapsulation technique for ISPs to connect endusers. It’s probably just listed there for intercept because it is so trivial to do, you might as well gather it.


This slide is harder to make sense of. I think that here the NSA is talking about IKE and IPsec combined, although they call it IPsec. Perhaps once they decrypt traffic, it goes into XKEYSCORE? They don’t seem to use XKEYSCORE as an input to crack the IKE PSKs? I’m amused by the name VULCAN-DEATHGRIP which might indicate that the HiFN (now Exar) Vulcan IPsec accelerator card might have a flaw in it that allows them to decrypt those IPsec VPNs. Also, I hope they properly mask their perl scripts for malicious decrypted traffic :P


This is probably the most interesting slide. It basically states that IPsec VPNs are compromised by router compromises that steal the IKE PSKs. This might be a good time to upgrade your router firmware to the latest version, and change to new strong PSKs (or switch it over to RSA 2048+ instead)
The AppID refers to the XKEYSCORE signature name. I’m glad to see that IKEv2 is finally catching on, even within the NSA :)


This slide is mentioning IPsec scripts, which might refer back to the previously mentioned genericIPSec_wrapper.pl script. This clearly indicates an IKE PSK brute force decrypting engine. If you are using weak PSKs, the NSA is decrypting your IKE traffic which will give them the keys to decrypt your IPsec traffic. Enabling PFS will help you a bit, but ultimtely you should really just switch to using RSA (or ECP) authentication and skip PSK’s alltogether. Note that the FIPS standard already disallows using PSK’s for IKE/IPsec. The NSA knows it is just too dangerous and weak.

Note to the NSA, please see RFC 4301 Section 1.1 which states:

The spelling “IPsec” is preferred and used throughout this and all related IPsec standards. All other capitalizations of IPsec (e.g., IPSEC, IPSec, ipsec) are deprecated.

On slide 31 it states: IPSec: 6640 (protocols) and 6648 (ports)

I don’t know what that refers to. Possible backhaul transport ports or part of an API port ?

On slide 33 it states: Need all the pieces (IKE and ESP for IPSec)

This really shows that the NSA is breaking IPsec by breaking IKE, which mostly seems to come down to weak PSKs and no PFS. And again, in case you still did not get it on slide 39:


“TAO got the configuration files which provided us the PSKs to enable passive exploitation”

So your IPsec VPN is getting owned because your VPN gateway got owned! Run your VPN on a dedicated machine, preferably opensource, and lock remote access via ssh down using strong keys and IP filtering.

And the last interesting nugget is on page 40:


“NSP had an implant which allows passive exploitation with just ESP”.

I read this to mean that the hardware or software of the system running IPsec was compromised, causing it to send valid protocol ESP packets, but creating those in such a way that these could be decrypted without knowing the ESP session keys (from IKE). Possibly by subverting the hardware number generator, or functions related to IV / ICV’s / nonces that would appear to be random but were not. Or less likely if the hardware/software was using padding bytes to reveal decryption keys.


If anything, I would say that IPsec and IKE have withstood the attacks of the NSA. The fact that the NSA needs implants and needs to use brutce force attacks against IKE PSKs shows that despite its kitchen sink nature, IKE and IPsec are doing quite well if you configure it properly and run it on a secure host. But misconfiguration of IKE is something we did a poor job of preventing. We can and should have more automatic IKE/IPsec deployed using strong default configurations.

And you should hear back from The Libreswan Team in the next two months regarding just that. Opportunistic Encryption: Round Two

deFUDing “DNSSEC has failed”

It’s easy to forward or like an article of which the title says something you agree with, without actually reading the content. It seems “DNSSEC has failed” is one of those that showed up in my twitter feed. It’s so defunct of facts, it’s pretty clear most people have just read the title and retweeted it. Hopefully, they will at least read this post before not retweeting it.

It starts comparing the DNSSEC protocol with the SSH protocol. It claims ssh just works. It fails to take into consideration that the two are completely different trust models. With ssh, you tend to own both end points. So there is no issue of PKI or key distribution or trust. And even so, I am willing to bet the author never actually checks his SSH fingerprints when connecting to new servers, and just depends on the ssh leap of faith. Well, security is easy when you can assume you are not under attack. Most ironically, the easiest and only cross-trust-domain SSH public key distribution system out there is the DNSSEC protected SSHFP record. The only other one I know of is GSSAPI and other hooks into centrally-managed systems – eg all systems under one administrative control. Trust is pretty easy when you centrally distribute trust anchors.
Let’s also not talk about the debian ssh disaster, which was a pretty spectacular failure compromising thousands of ssh servers. Let me conclude the ssh comparison with inviting the author to login to bofh.nohats.ca using ssh in a fully trusted matter. He will end up 1) calling me to verify the ssh host key, 2) trusting the SSHFP key in DNSSEC or 3) leap-of-faithing it and hoping he was not under attack.

Next, the author brings up the tremendously easy to use and success of TLS. Never mind that we still need to run SSL 2.x for WinXP compatibility, most servers run TLS 1.0 which is severely broken, we just had the TLS 1.2 resumption/renegotiation disaster, about 4 CA compromises in 2013, and browsers running to patchwork solutions like hardcoded certificate pinning and Certificate Transparency registries that uses “n of m” style “it is probably the right certificate” solutions. There are about 600 (sub)CA’s that can sign any TLS certificate for any domain on the planet, ranging from Russia, to Iran to the U.S – all parties that are known to coerce parties and with a heavy industrial espionage component.

If TLS was successful, port 80 would be dead. Port 80 is the proof that the TLS trust model has failed. It involves money and so these certifications are only available to corporations. While DV certificates are now basically free, they are also considered worthless (protection against passive attack only) EV certificates are now dropping in price – which will make them finally available for more people but unfortunately because of that also lose their value. And all of this security hinges on the (mostly hidden) CABforum and browser vendor decisions on which CAs to include, and which CAs not to include.

While on most proprietary OSes, these preloaded CAs are at least managed by the vendor, the situation in Linux is a disaster. OpenSSL doesn’t support blacklists, NSS does. GnuTLS I don’t even know. Where are these CAs and blacklists? Do you know applications using openssl for their TLS are using the same CA bundle as applications using NSS or GnuTLS? Only very recently has Fedora and RHEL moved to a system where this is guaranteed. Even worse, most python wrappers using TLS don’t even check anything of your TLS connection. Presenting the wrong certificate will just cause that software to continue. And then look at the latest Apple TLS bug where anyone could bypass all TLS checks completely. How the author can even point to TLS as a security success is a complete mystery to me. It’s the worst computer security disaster we ever created! Look at the amount of complete failure TLS deployments at SSL Pulse – 30% allows weak ciphers, 6% has a broken certificate chain, less then 30% supports TLS 1.2, 6% vulnerable to the renegotiation/resumption attack, 70% vulnerable to the BEAST attack, 13% vulnerable to the compression attack, 56% allow RC4 and 50% does not support PFS. And that according to the author is the “success” that DNSSEC should take a lesson from?

Having listed the “success” stories of ssh and tls, he than looks at DNSSEC. His first factual statement regarding DNSSEC is “Most resolving software supports DNSSEC, but none have enabled it by default“. He is wrong. I have been the package (co)maintainer of bind, unbound and nsd for Fedora and EPEL for many years. On Fedora and EPEL/RHEL the default install enables DNSSEC for all three, and has done so for many years – even before the root was signed with DNSSEC!

Next he claims “To enable DNSSEC we have a ‘tutorial’ by Olaf Kolkman which spans a whopping sixty-nine pages“. As I said, the author is wrong. Simply “yum install bind nsd unbound” and the software comes with dnssec enabled, using the root key as trust anchor.

He then points to a DNSSEC training that takes “multiple days” and claiming that shows the complexity. The same vendor lists a two day DHCP course, a two and a five day “plain DNS & bind” course, and a two and five day IPv6 course. Does that mean those technologies are all failures too because it takes so many days to teach a course about it? According to the author, DHCP would be a failure and no one can deploy it because it “requires” a two day training? It’s just simplistic cherry-picked FUD.

His next argument is “Furthermore, there are almost no simple instructions for enabling DNSSEC, especially none that use KSKs and ZSKs, and include instructions on how to do key rollovers.“.

First, let me ask you how you do an ssh host key rollover? How do you make that seamless so it won’t trigger anyone who ever connected before from seeing a changed key? Where do they go to confirm the changed key is installed by the administrator and not the result of a hacked server or connection? Does this mean ssh has failed too? Again, amusingly enough the only way to rollover ssh keys is piggybacking on SSHFP in DNSSEC or GSSAPI trust models. But I don’t see the author claiming ssh has failed because it cannot do rollovers on its own. More cherry-picking.

The reality is, ssh and DNSSEC can do fine without rollovers. Only TLS requires rollovers because the trust model depends on the fact that you regularly pay the trustees managing that whole disaster. But if you do want or need to do a rollover, with ssh you’re fucked, with TLS you have to pay and with DNSSEC you have a seamless transition method. Let me repeat that – only DNSSEC actually implements a free and seamless key rollover mechanism! So of course, the author just states it is too hard to use. But again does not compare it against TLS or ssh at all. Cherry-picking all the way down. And by the way, there are also good reasons why you do not need to worry about DNSSEC rollovers at all.

As I said before, DNSSEC is already “enabled”, so let’s interpret the author’s quest for “simple instructions to enable DNSSEC” to mean “how do I sign my zone?”. Let us assume he is running bind right now with his zone in /var/named/1sand0s.nl. The author now has to issue the following complicated sequence of commands:

$ dnssec-keygen -f KSK -a RSASHA256 -b 2048 -n ZONE 1sand0s.nl
Generating key pair........+++ ..+++
$ dnssec-keygen -a RSASHA256 -b 1024 -n ZONE 1sand0s.nl
Generating key pair........................................++++++ ..................++++++
$ cat K1sand0s.nl.*.key >> 1sand0s.nl
$ dnssec-signzone -o 1sand0s.nl -k K1sand0s.nl.+008+01699 1sand0s.nl K1sand0s.nl.+008+24363
Verifying the zone using the following algorithms: RSASHA256.
Zone fully signed:
Algorithm: RSASHA256: KSKs: 1 active, 0 stand-by, 0 revoked
ZSKs: 1 active, 0 stand-by, 0 revoked

Edit named.conf and load the file 1sand0s.nl.signed instead of 1sand0s.nl and you’re done. And whenever you change your zone, re-run that dnssec-signer command.
(NOTE: and as pointed out to me, if you don’t have zone edits every week, do run that signer command in a cron job weekly to refresh the DNSSEC signature records)

You can put this into a git hook and maintain the whole thing without ever realising there is a signing step in between. Hardly rocket science. I think we can argue whether or not this is harder or easier than creating a CSR for Apache, get it signed, and than installed (and too many people will forget to actually get the intermediate CAs installed in apache, and their deployment is actually even broken!)

The author’s issue with opendnssec is that it is “yet another daemon” using a “database backend”. I guess he is also not running postfix because that thing has more daemons than I can remember the names of. Opendnssec’s default “database backend” is sqlite. If you hate those, there is a lot of software you’ll going to have to purge from your servers. Sqlite is a perfect backend for small databases that are just too complex for simple text files. For those of us who don’t mind two extra daemons and some sqlite files, to sign your domain from scratch with opendnssec:

# yum install opendnssec
# systemctl enable opendnssec.service
# systemctl start opendnssec.service
# ods-ksmutil zone add –zone 1sand0s.nl –input /var/named/1sand0s.nl –output /var/named/1sand0s.nl.signed
# ods-signer sign 1sand0s.nl

Change named to load the .signed zone instead of the unsigned zone and done. Shockingly complicated! Unsuitable for “common engineers” according to the author.

To make these instructions complete, the last step is to send your DS record to your Registrar which involves using their custom webgui to send them the following data:

yum install ldns
dig dnskey 1sand0s.nl > 1sand0s.nl.dnskey
ldns-key2ds -n 1sand0s.nl.dnskey

And if you want to do all of this within the same bind daemon and without sqlite, you can use bind’s builtin inline-signing management features instead.


Saying the tools are there for ssh (which doesn’t have ANY rollover support) and tls (which over half the world has misconfigured regardless of rollovers) but not for dnssec, is really short-sighted and wrong. What the author has done is taken his prejudice, and written some cherry-picked text justifying it. The only credit I can give him, is that he did not have to rely on the FUD statements by djb.

Development of libreswan vs openswan


UPDATE: Statistics updated to include 2017

Yesterday, openswan released version 2.6.40 to address CVE-2013-6466. You might be confused by its changelog (not the non-updated CHANGES) crediting me for the vast majority of code changes. . Basically all changes are pulled from the libreswan repository and are backports to openswan. The exception is their version of a patch for CVE-2013-6466. Libreswan’s fix is not a band-aid but an updated state machine. The backported libreswan fix is what is going into the updated openswan packages for RHEL5 and RHEL6 and are available at libreswan.org/security/openswan/CVE-2013-6466. RHEL7 will contain libreswan.

Basically, yesterday’s openswan 2.6.40 release brings it up to the initial libreswan-3.0 release of two years ago, plus the two CVE issues. Except that it crashes KLIPS and backported a libreswan commit that broke all non-XAUTH IPsec connections.

History and implementation status of Opportunistic Encryption for IPsec

(as sent to the cryptography mailing list)


In light of the NSA achievements, a few people asked about the FreeS/WAN IPsec OE efforts and whatever happened to it.

The short answer is, we failed and got distracted. The long answer follows below. At the end I will talk about the current plans that have lingered in the last two years to revive this initiative. Below I will use the word “we” a lot. Its meaning changes based on the context as various communities touched, merged, intersected and drifted apart.

NOTE: On September 28, 2013 there is be a memorial service in Ann Arbour for Hugh Daniel, manager of the old IPsec FreeS/WAN Project. Various crypto people will attend, including a bunch of us from freeswan. Hugh would have loved nothing better than his memorial service being used as a focal point to talk about “new OE”, so that’s what we will do on Saturday and Sunday. If you are interested in attending, feel free to contact me.

OE in a nutshell

For those not familiar with IPsec OE as per FreeS/WAN implementation. When activated, a host would install a blocking policy for Every packet to an IP address would trigger the kernel to hold the packet and signal the IKE daemon to go find an IPsec policy for that destination. If found, the tunnel would be build, and an IPsec tunnel to the remote IP would be established, and packets would flow. If no policy was found, a “pass” hole was poked so packets would go out unencrypted. Public keys for IP addresses were looked up in the reverse DNS by the IKE daemon based on the destination address. To help with roaming clients (roadwarriors), initiators could store their public key in their FQDN, and convey their FQDN as ID when performing IKE so the remote peer could look up their public key in the forward DNS. This came at the price of two dynamic clients not being able to do OE to each other. (turns out they couldn’t anyway, because of NAT)

What were the reasons for failing to encrypt the internet with OE IPsec (in no particular order):

1) Fragmentation of IPsec kernel stacks

In part due to the early history of FreeS/WAN combined with the export restrictions at the time. Instead of spending more time on IKE and key management for large scale enduser IPsec, we ended up wasting a lot of time fixing the FreeS/WAN KLIPS IPsec stack module for each Linux release. Another IPsec stack, which we dubbed XFRM/NETKEY appeared around 2.6.9 and was backported to 2.4.x. It was terribly incomplete and severely broken. With KLIPS not being within the kernel tree, it was never taken into account. XFRM/NETKEY remained totally unsuitable for OE for a decade. XFRM/NETKEY now has almost all functionality needed – I found out today it shoudl finally have first+last packet caching for dynamic tunnels, which are essential for OE. Since the application’s first packet triggered the IKE mechanism, the application would start retransmitting before IKE was completed. Even when the tunnel finally came up, the application was usually still waiting on that TCP retransmit. David McCullough and I still spend a lot of time fixing up KLIPS to work with the current Linux kernel. Look at ipsec_kversion.h just to see what a nightmare it has been to support Linux 2.0 to 2.6 (libreswan removed support for anything lower then recent 2.4.x kernels)

Linux IPsec Crypto hardware acceleration in practise is only possible with KLIPS + OCF, as the mainstraim async crypto is lacking in hardware driver support. If you want to build OE into people’s router/modem/setup box, this is important, though admittingly less so as time has moved on and even embedded hardware and phones are multicore or have special crypto CPU instructions.

An effort to make the kernel the sole provider of crypto algorithms that everyone could use also failed, and the idea was abandoned when CPU crypto instructions appeared directly accessable from userland.

2) US citizens could not contribute code or patches to FreeS/WAN

This was John Gilmore’s policy to ensure the software remained free for US citizens. If no US citizen touched the code, it would be immune to any presidential National Security Letter. I believe this was actually the main reason for KLIPS not going in mainstream kernel, although personal egos of kernel people seemed to have played a role here as well. Freeswan people really tried had in 2000/2001 to hook KLIPS into the kernel just the way the kernel people wanted. (Ironically, the XFRM/NETKEY hook so bad, it even confuses tcpdump and with it every sysadmin trying to see whether or not their traffic is encrypted) I still don’t fully understand why it was never merged, as the code was GPL, and it should have just been merged in, even against John’s wishes. Someone would have stepped in as maintainer – after all the initial brunt of the work had been done and we had a functional IPsec stack.

In the summer of 2003, I talked to John and together we agreed it was time to fork. Openswan was born to clearly indicate US coders could contribute. However, at that point the (then crappy) FRM/NETKEY IPsec stack was there to prevent OE from working due to the missing first+last packet caching. The FreeS/WAN Project ended and Openswan continued. At first in good pace, but that later slowed down and OE was no longer its focal point. (Due to legal reasons, I cannot go into details regarding the openswan history)

3) Not using DNS without DNSSEC

There were various issues that caused DNSSEC to get massively delayed. We needed DNSSEC to secure our DNS based distributed public key platform. Although it would have worked fine to use DNS against passive attackers (NSA trawling), we believed it was principly wrong to trust cryptographic material that was untrusted and vulnerable against active attacks. So while the developers encouraged people to put keys in DNS even without security, no one else picked it up. It sucks to need to say ‘we told you so’. But we should have really not waited on DNSSEC.

4) Dealing with the DNS working groups at IETF

The DNS community is one of the most pedantic group of people I know. They are very smart, often right, and had been known to be extremely defense of their DNS turf. (Note that things have improved considerably and if you think this is still an issue, I’m happy to try and help)

IETF was divided about the convergence of the “security of the DNS” and the “DNS as PKI” despite that this had always been a goal of DNSSEC for a large group of people within the IETF. The FreeS/WAN people were driving DNSSEC not so much for DNS as for the key distribution. After all, you can detect DNS forging if you know your public keys.

When we had the KEY/SIG records ready to go, it was decreed that it could only be used for the DNS itself. Applications could not use this KEY record. To make that distinction more clear, on the next change in the draft protocol, KEY was obsoleted and DNSKEY introduced. So IPsec keys were relegated back to TXT, since at the time we had no Generic Record format (RFC 3597) support, so waiting for any new RRtype to get any deployment to become usuable would take years. Almost everyone was on bind4 and never upgraded left us with no other choice but the TXT. Even though we wrote the OE and IPSECKEY RFCs, OE’s only deployments were done using TXT records.

5) DNSSEC was delayed by a decade

DNSSEC deployment was slowly gaining traction, but I think we really needed the Kaminsky bug to get that extra push for DNSSEC outside the geeks of the IETF. The US government mandate for DNSSEC in .GOV helped as well. But by this time, OE was mostly forgotten.

djb repeatedly tried to peddle his own warez. While not at all realistic, it always gained a lot of hype and media attention and probably did cause delays of DNSSEC deployment.

Kaminsky himself was shooting down DNSSEC too. I personally heckled him at various Black Hat’s and ICANN conferences until we finally sat down for a couple of hours to talk about DNSSEC’s history and design goals. I’ll claim my 15 minutes of fame for having converted him. It helped having Kaminsky say that although he didn’t like the complexity, he couldn’t see anything better. DNSSEC was needed for everyone.

DNSSEC was gaining traction. Then we ran into a bunch of DNSSEC deployment issues. We had the delays due to NSEC vs NSEC3 with OPTIN, and then on top of that in 2008 when the first big ISP in Sweden turned on DNSSEC in their resolvers all that traction was blown away.

Most consumer routers ran DNS proxies that implemented DNS as “known bitstreams” instead of implemeting the actual DNS protocol. The DNSSEC OK bit caused thousands of routers to drop DNSSEC packets as “invalid DNS”. The only realistic solution: Turn it off and wait two years for those routers to get obsoleted by faster wifi standards and talk to those vendors so they would not repeat their mistake with their next generation of routers.

We now have the IPSECKEY record format (though RFC 4025 is not useful, see below) and RFC 3597 for the generic DNS record deployed on all DNS servers. And we’re on our way to have DNSSEC on every end node (see also draft-wouters-edns-tcp-chain-query-00 I just submitted to the IETF)

We have a mostly clean working UDP/TCP port 53 transport for DNSSEC on most networks (in part thanks to Google DNS). Although our hotspot handling is still a little rough, with dnssec-trigger the only tool to hack configurable DNSSEC support into the OS for our coffee shop visits when we need to rely on forged DNS.

6) When you’re NAT on the net, you’re NOT on the net.

Opportunistic Encryption relied on a clear peer to peer connection. But we managed to degrade the internet into servers and clients. NAT was the biggest problem, and with CGN around the corner, it’s not something that is going away despite IPv6 offering enough IPs for everyone. In fact, for our “new OE”, this is the biggest hurdle to overcome. When Alice cannot talk to Bob because she cannot reach him due to a (carrier grade) NAT, we are stuck wildly poking holes and hoping packets flow.

7) The reverse DNS tree is dead Jim

OE depended on the reverse tree as a security mechanism that someone who was claiming a public key for a specific IP range was actually the legitimate owner of that IP space. It was the security method for RFC-4025.

But unless you are running in a datacenter, you do not have access to the reverse DNS. It is useless as key distribtion method. On top of that, large IPv6 deployments don’t even care any more to run any authoritative DNS for their reverse.


The IETF tried to revive this OE with the Better Then Nothing Security (“BTNS”) working group. Contrary to the name, they also fell into the “perfect is the enemy of good” trap and most discussion seemed to go into “channel binding” to upgrade anonymous IPsec to some kind of authenticated IPsec – at least by the time I became aware of them. In other words, the most important problem of key distribution was left outside the scope and no one actually seemed to have implemented anything. Though I have to admit, I’m behind on reading the VPN auto-discovery drafts. It is just
very discouraging to still be reading problem statement drafts. More over, I don’t think we should setup IPsec tunnels based on packets hitting the kernel. We have better ways now that we can leverage DNSSEC.

9) We were all complacent

The only interest for IPsec was for corporate VPNs. During the above listed problem periods, OE people gave up. Some walked away from IETF. While everyone gained an always-on portable IP device,
their crypto capabilities were practically non-existent. My current iphone 5 can connect to a corporate VPN, but trying to make it _just_ send out encrypted packets is impossible. Some trickery can be used to cause almost any packet to setup the VPN, but while that’s going on it is still leaking like a sieve. VPN is seen by phone vendors as a method to gain some enterprise users, not as the tool to protect the consumer. The Apple VPN client is a 10+ year old patched version of racoon. The only vendor that took VPNs seriously was RIM and we punished them by not buying their products, because we had other priorities like FourSquare, Facebook and Twitter.

We can only hope that those PRISM players are now put under economic pressure by frightened consumers to fix this. But as long as VPNs and DNSSEC is slow and error-prone, it is better for them not to go there.

The New Opportunistic Encryption

I’ve been brainstorming with various people on how to put IPsec OE back on the table. I’ve discussed this with a bunch of people around me, including the late Hugh Daniel, John Gilmore and Hugh Redelmeier of freeswan.

The packet capturing policy is not a good method because we cannot make any decision on where to find a public key for an IP address. The reverse is unusable, and IP addresses change often. We used it because we had nothing better. But now we do. Since every (secure) platform now runs DNSSEC on the end node, we can use this as our decision making point. Imagine my phone running a DNSSEC resolver (say unbound) and an IKE daemon (say libreswan). The DNS server has access to the set of DNS name and matching IP address. It can lookup the key in the forward DNS zone, and hand over the public key, dns name and IP address to the IKE daemon!

1) User tells browser to go to www.cypherpunks.ca

2) browser does a lookup for the A/AAAA record of www.cypherpunks.ca

3) DNSSEC resolver performs the lookup/validation for the A/AAAA record of www.cypherpunks.ca and additionally looks up the IPSECKEY record of www.cypherpunks.ca.

4a) The resolver will wait with returning the A/AAAA record to the browser until it knows if the IPSECKEY record exists or not. If not, it releases the A/AAA answer to the application. Packets flow in the clear.

4b) The resolver finds an IPSECKEY record. It sends the pubic key, the FQDN and the IP address(es) to the IKE daemon and waits for a response. Meanwhile it does _not_ release the A/AAAA record to the application.

5) The IKE daemon sets up the IPsec tunnel. We haven’t reached agreement yet over how this should be done. There are two choices:

a) The client uses an “@anonymous” ID for itself along with sending its public key inline with IKE. The client is responsible for ensuring there is no MITM attack, as it knows the server’s public key (from DNSSEC). The responding server will just use any key it received inline if it was received for the “@anonymous” ID.

b) The initiator (aka client) uses its own FQDN-based ID. It has preconfigured its DNS so that an IPSECKEY record exists for its FQDN (protected by DNSSEC). The key is not send inline with IKE. Instead, when the responder (aka server) sees the non-anonymous ID, it will perform a DNSSEC secured lookup to obtain the IPSECKEY out of band. Both parties confirm there is no MITM.

The advantage of a) is that it leaks less user information and makes tracking users harder. The client can regularly generate another anonymous keypair. The disadvantage of a) is that it turns peers into clients and servers. And two clients cannot initiate OE to each other.

6) The tunnel is established and the IKE daemon notifies the local DNSSEC server that had instructed it to setup the IPsec tunnel.

7) The resolver releases the IP address to the application.

8) The applications starts sending packets and the IPsec policy encrypts them al.

I’m personally in favour of the @anonymous solution. But there is no reason why support for both could not be implemented.

What are some of the obstacles and work to do:

1) writing the unbound plugin

2) writing the support for @anonymous for the server-side. This includes raw keys for IKEv2 (draft-ietf-ipsecme-oob-pubkey)

3) With NAT, the client suggests an inner-IP. This could be abused or clash, We need to ‘contain’ each connection, possibly using generated ipv6 addresses 4) We cannot use the “gateway” field of RFC-4025, or people could trick a server into giving a client all communication to a certain IP address that does not belong to them

5) anonymous connections should generate throw-away keys to remain anonymous

6) implement draft-wouters-edns-tcp-chain or else latency/RTTs will prevent real-life deployment of DNSSEC validated IPSECKEYs on mobile devices.

7) This allows no upgrading from anonymous to mutually authenticated, but IKE policies can be added to the server/client that would match on different IDs (eg X.509) that work independantly of OE without introducing complicated channel binding promotion code. Other IKEv2 extensions could possible be applied to facilitate promotions.

I’m sure more implementation issues will show up once we get this going, but there are no real fundamental issues why we cannot deploy this in a couple of months of time. My plan is to get libreswan to support this version of OE. Additionally, once we use draft-wouters-edns-tcp-chain, it becomes cheap to do these lookups through the tor network. If the tor exit nodes then also feed each other with DNSSEC cache material, it should make tracing individual clients even harder.

(anyone willing to assist, especially with coding, do contact me)