The DMZ

Thomas Pornin

18:00

@AviD: section 6.1 of RFC 4120 prohibits realm names that do not match some specific categories, defined as "looks like a domain name", "looks like a X.500 name", and "is something else (with an explicit header)"

So an empty realm name (also called "null realm") would be allowed only insofar as it "looks like" a domain name, for some notion of "looks like" which is not defined.

Nate Kerkhofs

18:26

Question: Does chrome already enforce "no SHA1 certificates for HTTPS connections"? Because I'm trying to check a government website that has a sha1 cert valid till 9th September 2015, but I'm getting an error message

And I'm not sure whether the error is from the weak signature or from the website

I'm getting the error in IE11 as well.

Thomas Pornin

@NateKerkhofs Which web site is that ?

Nate Kerkhofs

mijnsociaaldossier.acv-online.be

errors occur both on Mac and Windows, with Chrome 42, IE11 and whatever version Safari is currenlty on

it's not really government, but close to it. Belgian union

Thomas Pornin

@NateKerkhofs From here, chromium is rather happy with the site. It says "obsolete crypto" because HMAC/SHA-1 for integrity check of TLS records, but no issue on the certificate itself.

The certificate I see expires on April 24, 2016, not September 9, 2015.

Nate Kerkhofs

really?

I see an error message: "de gevraagde URL heeft een probleem veroorzaakt"

Thomas Pornin

It is also signed with RSA+SHA-256, not RSA+SHA-1

Nate Kerkhofs

18:33

I should normally see a page where I need to authenticate using my electronic ID

Thomas Pornin

What is the thumbprint (SHA-1 fingerprint) of the site certificate, from your point of view ?

From here, it is 7F 39 9D E1 0A DD C2 AE 85 95 6D 5A 88 01 A4 4D E2 B3 E2 D6

Nate Kerkhofs

‎a3 82 75 d6 d5 61 e4 49 41 d5 35 20 b8 3d 04 a9 dd 34 e6 ee

Thomas Pornin

Not the same certificate...

Which is indicative of something fishy.

Nate Kerkhofs

What website is your cert published for? mine is for *.acv-online.be

Thomas Pornin

*.iamfas.belgium.be

I am getting redirected. Don't know why.

Nate Kerkhofs

18:37

What page are you being redirected to? It's possible it might not work from outside Belgium

Thomas Pornin

I end up on: idp.iamfas.belgium.be/fas/SSOPOST/metaAlias/idp

Nate Kerkhofs

could you try visiting acv-online.be/acv-online/Systeem/eservices.html

which should end you on a page with a few colored cubes

and then click on the first link on the right below those cubes, with the word "werkloosheid" in

"consulteer je e-dossier werkloosheid"

that's the page I get the error on

Thomas Pornin

@NateKerkhofs Redirect again...

Nate Kerkhofs

okay, I guess that it's not accessible from outside Belgium then. outside of using a proxy, there's not much we can do about that

probably a security measure to prevent foreign hackers

Rоry McCune

it's not a flat redirect based on IP address AFAIK

it does GET to the initial domain then a POST to that domain

they gets the favicon

CodesInChaos

18:43

You are in a maze of twisty little redirects, all alike.

Rоry McCune

then goes off to auth.acv-online.be and does some saml stuff

the initial GET returns some obfuscated looking JavaScript

Thomas Pornin

I can see the certificate with OpenSSL and I don't see anything blatantly wrong with it.

Rоry McCune

which I'm guessing causes the POST with some response code..

Thomas Pornin

Sure, it uses SHA-1, but apart from that...

What is the exact error message you get from Chrome ?

Nate Kerkhofs

I get "Request Blocked" in the tab name, but apart from that a reply 200 OK and an error from the website, saying "the requested URL has caused a problem"

since I got "request blocked", I thought it might have had something to do with the client i'm using

Thomas Pornin

18:51

By disabling Javascript I can keep Chromium on "https://mijnsociaaldossier.acv-online.be/" without being redirected -- and it is quite happy with the certificate.

I suppose your problem is NOT the holy war that Chrome wages on SHA-1.

Also:

googleonlinesecurity.blogspot.ca/2014/09/…

A certificate that expires in 2015 should not trigger any warning about SHA-1 in Chrome.

@NateKerkhofs If the server tells you that there is a problem, then the issue is on the server side, not the browser.

Nate Kerkhofs

I thought as much, but I wasn't sure, since they only mentioned certs after 2016

Iszi

Anyone got a name for the Apple UEFI 0-day yet? Looks like the media is expecting one. itnews.com.au/News/…

Thomas Pornin

The server has no notion of how much allergic to SHA-1 your browser may be, since the browser does not tell.

Iszi

> "The exact cause for the zero-day vulnerability, which is yet to be named, has not been fully identified."

Nate Kerkhofs

I propose SMARMI: Sleep Mode Apple Rootkit Memory Injector

It's short, explains the problem and evokes memories of smarmy shoarma bars

Iszi

18:57

@NateKerkhofs Don't you mean shawarma?

Nate Kerkhofs

It's called Shoarma in Dutch, I don't know what they call it in other places

Also: meet someone who will never be employed: twitter.com/Gligar13/status/605407871761809408

or has that MLP resume been linked today already?

Iszi

@NateKerkhofs Scariest part about that resume: "He even had his very own nuclear reactor to play with!"

And think, one day we might actually be calling that guy "doctor".

Adi

19:18

Anybody seen these? creativecrafthouse.com/…

CC @Thomas

Nate Kerkhofs

@Iszi I'd rather watch Totalbiscuit make yet another cringeworthy WTF Is of a puzzle platformer

Adi

@Thomas I'm trying to think of efficient ways to crack a text "encrypted" with one of those

The number of possible keys is a bit more than 600k

The whole text (roughly 100 letters) is encrypted with the same "key"

Thomas Pornin

@Adi 600k keys is nothing for a computer

Adi

@ThomasPornin My problem is that I'm unable to escape the one-time pad problem

CodesInChaos

@Adi hm?

Adi

19:21

I can't think of a way to not produce every single text in the universe that is 100 letters

I don't have anything that would hint me at the actual plaintext

Thomas Pornin

The idea is that if there are 600k possible keys, then there are 600k possible decrypted plaintexts.

Adi

so any key that can produce human-readable text is plausible

Thomas Pornin

Then this is just a matter of finding the "right one"

CodesInChaos

I don't get your point. Assuming you know the algorithm, just implement it. Then run it 600k times and check if the output is English

Thomas Pornin

With 100 letters, the number of possible texts is way larger than the number of keys

and probability of a wrong key yielding something "readable" by a human is very small

Adi

19:24

@CodesInChaos To be honest, I'm not sure I get my point either.

Thomas Pornin

Just keep the texts that contain at least one English word of, say, 5 or more letters, and finish the job by hand.

Adi

I just can't wrap my head around the idea that it's possible to get to the right key

Nate Kerkhofs

I think what @adi is saying is that he is trying to find a more efficient way to beat SHA1 than brute-forcing

but he can't because that's the whole point of encryption

Adi

I'm sorry, but I don't think that's what I'm saying

@Thomas @Codes But hey, thanks for the input. I guess the best I can do is to actually try it out and see. I'll ask for more help later

Nate Kerkhofs

Let's break down the problem

Thomas Pornin

19:26

@Adi Try to implement a function that, given a sequence of letters, returns the length of the largest sub-sequence of letters in that sequence that is also a valid word in a given dictionary (say, a dictionary of English words).

CodesInChaos

A key with 600k possibilities is enough to produce every 4 letter combination.

Thomas Pornin

Call f() that function.

Nate Kerkhofs

You got a 100 letter text that's encrypted with SHA1

correct?

and you don't know the key, but it's one of 600K possible keys. correct?

Thomas Pornin

Run f() on the results of all decryptions (by each of the 600k keys) and look at the decrypted results, starting with the sequences with the highest f().

CodesInChaos

Since natural language is highly redundant, producing anything that looks like proper plaintext significantly longer than 4 letters is very unlikely.

Nate Kerkhofs

19:27

Aside from the fact that sha1 AFAIK is a hashing algorithm and not an encryption algorithm

CodesInChaos

@NateKerkhofs Adnan is talking about the enigma inspired crypto machine he linked. Not SHA1.

Nate Kerkhofs

okay... He linked @thomas, not the post where @thomas mentioned Sha

Gilles

@ThomasPornin It's easy to fool this. Just sayin'. But sure, ok, the risk is low.

Nate Kerkhofs

it highlighted the SHA1 post because that was @ThomasPornin his last post

/temporarilyconfused

Rоry McCune

@NateKerkhofs yeah the linking thing in this room can be a bit odd sometimes

I often get confused by it

CodesInChaos

19:31

If you just use @NateKerkhofs it'll highlight the last post of that user when you mouse-over, but it won't show the link arrow.

Adi

@Thomas That seems like what I should do

@Codes @Thomas I just happened to stumble upon something quite interesting

Aho–Corasick string matching algorithm

In computer science, the Aho–Corasick string matching algorithm is a string searching algorithm invented by Alfred V. Aho and Margaret J. Corasick. It is a kind of dictionary-matching algorithm that locates elements of a finite set of strings (the "dictionary") within an input text. It matches all patterns simultaneously. The complexity of the algorithm is linear in the length of the patterns plus the length of the searched text plus the number of output matches. Note that because all matches are found, there can be a quadratic number of matches if every substring matches (e.g. dictionary = a,...

I found a Python implementation. I'm giving it a try

Thomas Pornin

@Adi I am not sure the problem at hand requires such finesse but if it amuses you then go for it.

Nate Kerkhofs

I assume there's also a Python library that simulates the Enigma encryption?

CodesInChaos

@NateKerkhofs That machine can't really be enigma

it had far more keys.

Thomas Pornin

And Enigma was electrically powered, too. Here we just see some gears.

Nate Kerkhofs

19:41

@CodesInChaos "Enigma IV Encryption cypher machine"

@ThomasPornin please google enigma machine. it was a mechanical crypto device that used 3 of 6 rotating dials and some cables to encrypt messages

Thomas Pornin

@NateKerkhofs I think this is a reuse of the term "Enigma" which is completely unrelated to the machine used in the German army and navy back in WWII.

CodesInChaos

It's called "enigma". But it's clearly inferior to the nazi enigma.

Nate Kerkhofs

it used electricity, but that was just to light some lamps

CodesInChaos

It not only had dials, but also a switchboard. The number of keys vastly exceeded 600k.

Nate Kerkhofs

It's clearly a modern reinterpretation of the WWII enigma machine, down to the name. it still uses the same base method

yeah, I googled it now. Enigma had something like 154 trillion settings (a million million million

CodesInChaos

19:47

@Adi Do you have a spec of that machine/cipher? Or how do you plan on implementing it?

nealmcb

20:24

Looks interesting: "Honorable Mention for the 2014 ACM Doctoral Dissertation Award went to John Criswell of the University of Rochester, and John C. Duchi of Stanford University. They will share a $10,000 prize, with financial sponsorship provided by Google Inc.

Criswell’s dissertation, “Secure Virtual Architecture: Security for Commodity Software Systems,” describes a compiler-based infrastructure designed to address the challenges of securing systems that use commodity operating systems like UNIX or Linux. This Secure Virtual Architecture (SVA) can protect both operating system and appl…

sosp2007.org/papers/sosp139-criswell.pdf

Adi

@CodesInChaos I have no algorithmic knowledge of the machine at all. I'm currently analysing the disks and trying to see how the order and initial positions of the keys related to the input/output

From that, I'll probably reach some sort of function that should be quite easy to program

"Key" is a strong word as there's really no logical/mathematical encrypt(msg, key) = ciphertext

The fixed and simplistic disks make it quite easy to crack by a computer, but quite difficult for me to wrap my head around

Let's see

Thomas Pornin

Is there an official source for hashes of Windows installation ISO ?

Iszi

@ThomasPornin Pretty sure MS used to have them alongside the downloads on TechNet.

How they're doing it since TechNet died, I don't know. Haven't gone looking in awhile.

Thomas Pornin

@Iszi I have looked for the MD5, SHA-1, SHA-256 and SHA-512 hash values on Google

Only the SHA-1 hits things, and they are all Torrent references. Nothing from Microsoft.

Iszi

The TechNet Subscriber Downloads section still has SHA1 hashes available even though I'm pretty sure I can't actually download the files anymore.

Are there a few in particular you're looking for?

You might actually be able to pull them up yourself, since I can still view them without an active subscription. technet.microsoft.com/en-us/subscriptions/downloads

May need to log in with a Microsoft account.

Thomas Pornin

20:47

@Iszi I have an ISO which has been downloaded and installed, but now I must convince an auditor that the ISO which was installed is a genuine one from Microsoft.

Preferably without having to download it again.

Even more preferably without having to download and install it again.

Iszi

Which version/edition?

(...And shouldn't you be checking this before you install?) ;-)

Thomas Pornin

@Iszi I am not the one who did the download and forgot to call the auditor to witness it.

I am given the mission to, as always, save the day.

Iszi

I know the feeling.

Thomas Pornin

It is some Windows Server 2012 R2, with French support (at least)

Iszi

Need version, edition, service pack level, and distribution type. e.g.: Windows 7 Professional SP1 Retail.

Thomas Pornin

20:52

From what I find on the Web, it would correspond to build 6052780

Iszi

Server 2012 R2 w/ Update, Volume License - sound right?

Thomas Pornin

@Iszi Yes

Well, from what I see here

I am supposed to save things, but I don't get told where the things are or even what they are, only that they are in some un-saved status right now

Iszi

That's the description I got off Google for that build ID, but it doesn't match what I'm seeing on TechNet.

Oh, wait... right...

Probably the "with French support" might be borking me.

You sure it's supposed to end in 80, not 08?

You have the original ISO file name?

Thomas Pornin

I found the hash

Iszi

Ah, spiffy.

Thomas Pornin

20:59

The original file name is distinct: it ends with 6052772

Iszi

Glad I could be of help... if I was. If I wasn't, then meh. It was interesting nonetheless.

Thomas Pornin

@Iszi You pointed me at the right place, so thanks.

Iszi

@ThomasPornin That was the code I kept hitting.

@ThomasPornin No problem. Good luck with it.

Thomas Pornin

@Iszi The day is saved.

I hope it will, say, stay saved.

Iszi

:-D

And now, I must depart. See ya later.

Arperum

21:52

cc @RoryAlsop I seem to have a need to go to the UK next year.

David Freitag

22:03

Alestorm - You Are A Pirate

►

Arperum

@DavidFreitag I'm going to see them Saturday (or friday, think it's saturday anyway)

David Freitag

DO WHAT YOU WANT 'CAUSE A PIRATE IS FREE!!! YOU ARE A PIRATE.

Arperum

Bloodbound is for thursday. Sabaton will have to wait until early august before I see them again (for the third time this year)

ALso: DAMMIT TELENET. I DON'T CARE THAT I USED 90% OF MY DATA LIMIT FOR THE MONTH IF IT RESETS IN 48 HOURS. (I couldn't send my message because they had to forcibly show me the warning, but they couldn't show the warning because that's not how chat works.

Rory Alsop

@Arperum nice!

Arperum

@RoryAlsop Yea, I really want to see that tour, but no clue how I'll get there yet.

Rory Alsop

22:21

Gloryhammer - The unicorn invasion of Dundee Lyrics

►

Arperum

@RoryAlsop They are fun live. Seen them once I think. ANd their second CD has been mastered and send to the record label according to facebook.

Rory Alsop

@Arperum my kids are now asking if they can go see them in Glasgow next year

David Freitag

23

A: Translate your English source code into a foreign language

Python, Koine Greek - Πύθων My favorite programming language, in my favorite foreign language--perfect! And it doesn't hurt that the name is already Greek. The translator program in Python 3 (thank goodness for native Unicode support): with open(__file__, encoding="utf-8") as f: code = f.r...

Arperum

@RoryAlsop I assume I am to blame for that?

Transcript for