Universal backdoor installation tool: What's wrong with system updates

In our blog on Habré, we not only talk about the development of our product - billing for Hydra telecom operators , but also publish materials about working with infrastructure and using technologies.

German journalist and hacker Leif Riga (Leif Ryge) wrote for the publication Ars Technica interesting material that the modern approach to the organization of software updates carries serious information security risks. We bring to your attention the main thoughts of this article.
')

Prehistory

Back in 2014, the Washington Post edited that companies like Apple and Google should invent something like a secret key with which they could get access to their products and which they would store and transfer to special services only if they received a legal solutions".

The authors of the material were then criticized for the very fact of putting forward such an idea - the Internet community strongly opposed it, arguing that if you allow only the “right” people to use such a tool for hacking, like the public servants and special services, then there is still no guarantee that will not be in the wrong hands.

The problem is that such a “golden key”, which opens access to any device, already exists, writes Riga - these are system updates.

Some theory

Riga writes about the existence of "weak links" in which the security chain of software can break, allowing an attacker to deceive the victim's computer and make him "believe" that the malicious update offered to him is in fact legitimate. Such a “directed attack of updates” will allow access to the system of a specific person - this means that other users of the same software will not know about it, which will allow a hacker to go unnoticed for a long time.

Riga has two conditions for this type of attack: the attacker must be able to send updates to the victim’s system, and somehow convince the system that the updates came from a trusted source. If he manages to do this, he will get truly unlimited possibilities for committing unlawful acts. If we imagine that the attacker was able to trick to install a specially generated update of the operating system, then he could get access to the data of any applications used by her, seize all access keys and passwords, as well as all unencrypted data.

In the case of a backdoor in the encryption system, the hacker will be able to decrypt random data encrypted by the victims, and not run malicious software. Riga writes that in his opinion, many people who condemned the idea of ​​the Washington Post material had a backdoor of the latter type.

From the point of view of the attacker, each of the two options has its advantages, writes a German researcher. The first option allows you to passively collect data, including encrypted, for further decryption. The second option allows you to conduct an attack only when conditions for this arise, but if such conditions occur, the hacker gets much more scope for action than in the case of access to the received, but encrypted data. In this case, he will be able to intercept data from all applications and even connected devices, such as microphones or cameras.

With all the scale of possible problems, many software developers have only recently thought about somehow confirming the authenticity of their updates. However, even in the case of most of the projects that are concerned, their creators still leave possible weak links in the chain, I am sure Riga.

What is the problem

The researcher writes that the problem exists in the case of most operating systems. Riga's favorite OS is Debian. Using a simple command, the user can find out how many “weak links” exist in which attackers can try to break the security chain when downloading updates:

sudo apt-key list | grep pub | wc -l 

In the case of the Riga system, this number is nine. Here is how he describes the problem:

Every time when I run the apt-get update command, anyone who has access to one of these nine keys, and who is located between my computer and the servers from which he downloads updates, can send me malicious software, and I will launch it as root user.

Why it became possible

Riga writes that the fact that a huge number of the smartest people involved in the design and development of software, still do not pay attention to this problem, is simply amazing. The researcher explains this with a mixture of "naivety and self-confidence." The developers are confident that they will be able to secure their keys in the face of a real attack, but they did not think that the governments of different countries could simply force them to use their own keys to sign the updates they needed, but harmful to users.

Apple vs FBI: what's really going on

Current events surrounding the FBI’s Apple’s requirements for decrypting iPhone data suggest that people like Riga, who were worried about those very “security weaknesses”, are no longer marginal paranoids, and the problem really exists.

In all of this history, the most important, according to the researcher, is the fact that the FBI requires Apple to provide the Office with a signed update, which will disable the iPhone’s function for deleting data after a certain number of PIN attempts.

In systems like Debian, there is no such functionality - if someone can get access to an encrypted hard disk, then he will have all the possibilities for conducting a brute force password attack. However, it is generally more difficult to do this than in the case of a short PIN code - on computers with a keyboard, people choose much more complicated and long passwords. However, if the attacker can “convince” the computer to run random code without decrypting the disk, then you can simply access the key, and the complexity of its password will become just useless.

Therefore, when Apple says that the FBI is trying to "get us to create a backdoor to our products," in fact, they say that the special service is trying to force the company to use a backdoor, which is already present in these products, according to Riga. The researcher says that the fact that the FBI asks Apple to write some additional code also does not matter in this case - they can always hire other developers for this. The only thing Apple can do, and only she, is to provide the FBI with a signature for updates.

Should we consider the very "weak links" as full backdors? Riga writes that many security experts will not agree with this, since this is not a generally accepted definition of what should be considered a backdoor - after all, this system configuration is no secret to anyone. But in the Apple business, the company itself uses the word "backdoor", describing the "weak link" of the cryptographic protection of its systems.

What does this mean

Apple can win in the current dispute with the FBI, writes Riga. However, they may lose, and this means that in the past the company could concede in this matter to someone else. What if some kind of criminal group, just like the FBI now, wanted to get the opportunity to receive data from secured PIN-codey “iPhones”? And what if this organization found people who understand the design of technology, and they explained to the attackers, who should be “made” to make this possibility a reality? People within the company who have access to this “golden key” can be in grave danger if truly motivated people want to seize it too.

Riga hopes that the situation with the dispute between Apple and the FBI will serve as a wake-up call for the developers of popular systems using a distributed update distribution infrastructure. The researcher believes that, as a result, in the near future, the current inefficiencies of this mechanism should be eliminated, and in order to conduct an attack similar to the one described above, attackers will have to compromise several keys that are kept by different people in different jurisdictions. There are a number of projects that can help achieve this, including Dedic Cothority and Docker's Notary .

Getting rid of the “weak links”, an attack on each of which can become critical, should be a basic requirement for any new software distribution mechanisms.