Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview

Artifact ID:	c88f59d88521cf722a359d3aa4357ca29ba0d546
Page Name:	nsa-backdoor
Date:	2014-04-12 23:35:49
Original User:	bernd
Parent:	3df92a41e278a91a8033152dc98cad106bd02b0d (diff)
Next	bfc01b45dd96dbd48b3fd5c2cd0a2639f5c4b1cf

Content

NSA Backdoor Fnord

As you all know, it is not allowed to speak about NSA-demanded backdoors, and especially it is strictly prohibited to give any details. However, it is allowed to boldly lie about NSA-demanded backdoors if you didn't receive such a request, because you are not under a gag order, and in general, lying about the quality of your product is not only legal, but "best practice". The purpose of this NSA backdoor fnord is to make you worry about the quality of net2o, and therefore you start looking at the source code; the topics mentioned here are all security things to consider.

Therefore, here is the official statement about NSA-demanded backdoors: There is a NSA-requested backdoor in net2o. Update: The NSA was here again, and said they like to have a way to remotely access any memory without actually making a connection that would show up in a log file. Especially access to the secret key is required, as net2o doesn't use passwords for login. The backdoor implements this by providing the address and length of the region to be sent as integers and uses the $-push command, which pushes the content as string in the reply packet.

As net2o is open source, you can verify the truth value of the statement above. And keep an eye on this page.

What is this page for?

Software is inherently buggy - we all make mistakes. Secure networking software is even worse, because small bugs have big consequences. And with the NSA Bullrun program, we not only have to deal with the normal, "lazy" bugs, which don't cause any harm until found (either by honest security researcher or evil criminals), but with bugs intentionally placed, and used by the secret services from day 0.

Developing in Forth is a "crash early, crash often" exercise, but security related bugs don't crash the program.

net2o is not ready for use, so bugs do happen, and get fixed, but the bugs described here usually are real bugs I found and fixed during development. All of them look like professionally implanted bugs by the NSA, because that's the state of the art how to implant backdoors: It must provide its author with "reasonable denial", claiming incompetence.

However, in order to get things right, we need a culture of accepting our mistakes, and fixing them. Many programmers deny bugs, and request at least a proof of concept attack, before they actually start doing something. This sort of culture is so wrong: As author of security critical systems, you must be constantly scared by people using every way to break into your software, and you must be ready to fix every bug, even just potential risks, before someone shows you an actual exploit.