Microsoft admits to signing rootkit malware in supply-chain fiasco

Celebrate Nite

Come On Baby It's Party Time!
kiwifarms.net
Joined
Jan 2, 2020
According to this, it relates to "Gaming Environments"... which is a broad-as-fuck term that could mean the XBOX to any/all PCs. Unsure if this also affects the newly announced/released Win11.

Article: https://www.bleepingcomputer.com/ne...gning-rootkit-malware-in-supply-chain-fiasco/
Archive: https://archive.md/srWWu
----------------------------------------------------------------------------
Microsoft has now confirmed signing a malicious driver being distributed within gaming environments.

This driver, called "Netfilter," is in fact a rootkit that was observed communicating with Chinese command-and-control (C2) IPs.

G Data malware analyst Karsten Hahn first took notice of this event last week and was joined by the wider infosec. community in tracing and analyzing the malicious drivers bearing the seal of Microsoft.

This incident has once again exposed threats to software supply-chain security, except this time it stemmed from a weakness in Microsoft's code-signing process.

"Netfilter" driver is rootkit signed by Microsoft

Last week, G Data's cybersecurity alert systems flagged what appeared to be a false positive, but was not—a Microsoft signed driver called "Netfilter."

The driver in question was seen communicating with China-based C&C IPs providing no legitimate functionality and as such raised suspicions.

This is when G Data's malware analyst Karsten Hahn shared this publicly and simultaneously contacted Microsoft:
netfilter_signature.jpeg

"Since Windows Vista, any code that runs in kernel mode is required to be tested and signed before public release to ensure stability for the operating system."

"Drivers without a Microsoft certificate cannot be installed by default," states Hahn.

At the time, BleepingComputer began observing the behavior of C2 URLs and also contacted Microsoft for a statement.

The first C2 URL returns a set of more routes (URLs) separated by the pipe ("|") symbol:
first-c2.jpg

Each of these serves a purpose, according to Hahn:
  • The URL ending in "/p" is associated with proxy settings,
  • "/sneed" provides encoded redirection IPs,
  • "/h?" is for receiving CPU-ID,
  • "/c" provided a root certificate, and
  • "/v?" is related to the malware's self-update functionality.
As seen by BleepingComputer, for example, the "/v?" path provided URL to the malicious Netfilter driver in question itself (living at "/d3"):
path-to-d3-netfilter.jpg

The G Data researcher spent some time sufficiently analyzing the driver and concluded it to be malware.

The researcher has analyzed the driver, its self-update functionality, and Indicators of Compromise (IOCs) in a detailed blog post.

"The sample has a self-update routine that sends its own MD5 hash to the server via hxxp://110.42.4.180:2081/v?v=6&m=," says Hahn.

An example request would look like this:
hxxp://110.42.4.180:2081/v?v=6&m=921fa8a5442e9bf3fe727e770cded4ab
"The server then responds with the URL for the latest sample, e.g. hxxp://110.42.4.180:2081/d6 or with 'OK' if the sample is up-to-date. The malware replaces its own file accordingly," further explained the researcher.
netfilter_selfupdate.png

During the course of his analysis, Hahn was joined by other malware researchers including Johann Aydinbas, Takahiro Haruyama, and Florian Roth.

Roth was able to gather the list of samples in a spreadsheet and has provided YARA rules for detecting these in your network environments.

Notably, the C2 IP 110.42.4.180 that the malicious Netfilter driver connects to belonged to Ningbo Zhuo Zhi Innovation Network Technology Co., Ltd, according to WHOIS records:
whois-apnic.jpg


Microsoft admits to signing the malicious driver

Microsoft is actively investigating this incident, although thus far, there is no evidence that stolen code-signing certificates were used.

The mishap seems to have resulted from the threat actor following Microsoft's process to submit the malicious Netfilter drivers, and managing to acquire the Microsoft-signed binary in a legitimate manner:

"Microsoft is investigating a malicious actor distributing malicious drivers within gaming environments."

"The actor submitted drivers for certification through the Windows Hardware Compatibility Program. The drivers were built by a third party."

"We have suspended the account and reviewed their submissions for additional signs of malware," said Microsoft yesterday.

According to Microsoft, the threat actor has mainly targeted the gaming sector specifically in China with these malicious drivers, and there is no indication of enterprise environments having been affected so far.

Microsoft has refrained from attributing this incident to nation-state actors just yet.

Falsely signed binaries can be abused by sophisticated threat actors to facilitate large-scale software supply-chain attacks.

The multifaceted Stuxnet attack that targeted Iran's nuclear program marks a well-known incident in which code-signing certificates were stolen from Realtek and JMicron to facilitate the attack.

This particular incident, however, has exposed weaknesses in a legitimate code-signing process, exploited by threat actors to acquire Microsoft-signed code without compromising any certificates.

Updates:
Jun 26th 12:26 PM ET: Clarified that BleepingComputer did not see the DoD list explicitly mentioning the alleged Chinese company, contrary to the details in the researcher's report. Also reached out to Hahn for clarification.

Jun 27th, 04:58 AM ET: A previous version of the blog post mentioned another researcher @cowonaut alleging that the aforementioned company has previously been marked by the U.S. Department of Defense (DoD) as a "Communist Chinese military" company. However, BleepingComputer did not see Ningbo Zhuo Zhi Innovation Network Technology Co., Ltd. present on any of the DoD lists available. The claim has since been retracted from the original blog post, and we have updated our article to reflect the same.

----------------------------------------------------------------------------
11421ba6d12f48e72d2ba1813e8893d29cf5c242ad43c78487ccd9968a77a209.png
 
Sometimes I do wonder how many layers of backdoors can be in a modern computer both on a hardware and software level. Will spooks from different countries be able to close to doors for the others? Personal (not really) computers!
 
  • Like
Reactions: Falcos_Commisar
Sometimes I do wonder how many layers of backdoors can be in a modern computer both on a hardware and software level. Will spooks from different countries be able to close to doors for the others? Personal (not really) computers!
Practically infinite. The whole hardware/software stack is fucking swiss cheese. It's a bloody miracle every computer in the world hasn't already been altered to just display goatse or meatspin and do nothing else.

First there's intentional hardware backdoors. Baked right into modern CPUs (like Intel Management Engine), TPM (that Microsoft is supposedly going to start requiring it for Windows 11), LOM (which is legitimately useful and intentional, but still technically a backdoor since it can basically see and control everything a system is doing and you're up shit creek if someone gains access to it who isn't supposed to).

Then come intentional software backdoors. BIOS, bootloader, kernel, init/systemd (or whatever garbage Windows uses as its equivalent), compilers and userspace programs have all had unpleasant backdoors inserted into them at various points.

Finally there are unintentional backdoors -- vulnerabilities in everything I just listed that get exploited by nefarious third parties.

As for spooks intentionally closing backdoors? They'll only ever do that for backdoors that could negatively impact them, and they'll only ever do it for themselves (without disclosing it to the public).

Shit sucks, man. The only real "security" any of us truly has is being so uninteresting and boring in the grand scheme of things that we're not generally worth snooping on.
 
Practically infinite. The whole hardware/software stack is fucking swiss cheese. It's a bloody miracle every computer in the world hasn't already been altered to just display goatse or meatspin and do nothing else.

First there's intentional hardware backdoors. Baked right into modern CPUs (like Intel Management Engine), TPM (that Microsoft is supposedly going to start requiring it for Windows 11), LOM (which is legitimately useful and intentional, but still technically a backdoor since it can basically see and control everything a system is doing and you're up shit creek if someone gains access to it who isn't supposed to).

Then come intentional software backdoors. BIOS, bootloader, kernel, init/systemd (or whatever garbage Windows uses as its equivalent), compilers and userspace programs have all had unpleasant backdoors inserted into them at various points.

Finally there are unintentional backdoors -- vulnerabilities in everything I just listed that get exploited by nefarious third parties.

As for spooks intentionally closing backdoors? They'll only ever do that for backdoors that could negatively impact them, and they'll only ever do it for themselves (without disclosing it to the public).

Shit sucks, man. The only real "security" any of us truly has is being so uninteresting and boring in the grand scheme of things that we're not generally worth snooping on.
100% correct

Any computing device that can accept input can get owned. Anything connected to the internet can be owners of someone REALLY wants it.

Either you go full DPRK or you take measures to mitigate damage.
 
What exactly is the Intel Management Engine?

Basically a little tiny embedded computer that runs a lite linux distro which is embedded into every Intel motherboard since 2006ish that has full debugger-level access to the system even when it's off. It's got direct unrestrained access to all the hardware too so it can just grab files and memory if it wants to and shuttle it off willy nilly to parts unknown on the internet.

Some versions have horrendous things like built-in wifi chips that talk on out-of-band wifi as well, so good luck airgapping when your computer is physically hosting an invisible hotspot when it's got any power whatsoever even if the computer itself is off.

https://hackaday.com/2017/12/11/what-you-need-to-know-about-the-intel-management-engine/ has a more complete summary and more info.
 
Last edited:
Basically a little tiny embedded computer that runs a lite linux distro
Almost, it's actually minix. People have tried crippling it by extracting it out of the firmware images of computers with varying success. If you remove it fully it does things like making your computer shut down every half hour, so the best software like me_cleaner can do is making it get stuck in a boot loop. What people need to understand is that it basically sits a level above everything else in the computer and the OS you run cannot even be aware of it. It's also always running. As long as the power supply supplies +5V standby power the management engine is running so with modern ATX computers the only way to turn it off is by flipping the switch of the power supply or disconnecting the computer from power. In notebooks it's running as long as a battery charge remains.

AMD has kind of a similar thing, the AMD Platform Security processor, which is an in the CPU embedded ARM core running a custom OS. Contrary to the ME It can mostly be turned off though, or well, so they claim. Same thing here, if you remove the software parts from the firmware, the computer will not boot. What is known through analysis though is that it has by default no network stack and in most default configurations can't easily be accessed from the "outside world", contrary to ME, it's more of a CPU-facing thing with functionality like fTPM.

Both also have a history of exploitable flaws being found with them. Intel leans very hard with it's gaggle of lawyers on anyone who tries to publish reverse-engineered findings on it through regular channels.

If you want a safe computer these days you either go 25-40 years back and use a vintage system when computers were a fringe thing nobody even thought of putting backdoors into (also systems that are safer by virtue of their simplicity/reduced attack surface) or you have to build it yourself out of primitive CPUs like the 68k or Z80/180 where you can always be very sure what it is doing and not doing. The good news is it has never been easier to do that. Of course these won't be very capable computers but hey. I'm kidding but only a little.
 
Back