Malware exists in different flavors. Most of the time, malware consists of malicious files stored in computers operating systems, just like any other file, and running as software with or without high privileges. When found, they generally can be easily deleted from the file system or removed when the operating system is being reinstalled. However, rootkits are yet different malware.
What are rootkits?
Rootkits are designed to provide access to a computer and possibly mask other malicious software running on it. Some rootkits also do not reside in the usual file system from the operating system, but in other places, like firmware. Rootkits often also run at kernel level, instead of the usual software level.
Such a piece of malware needs a lot more effort to be developed, compared to usual malware, because it faces many more technical and programming challenges.
According to Kaspersky, the rootkit is located in the firmware images of Gigabyte or ASUS motherboards. The infected firmware images are related to designs using the H81 chipset, suggesting that a common vulnerability may exist, which allowed the attackers to inject the rootkit into the firmwares image.
How does CosmicStrand work?
Affected firmware images have been altered to run the malicious code at system startup. A long execution chain is triggered to download and deploy malicious content inside the Windows operating systems kernel on the affected machine. The initial entry point for the firmware has been patched to redirect to code execution added in the .reloc section.
The firmware is being modified with an automated patcher, according to the researchers, which means the attackers had prior access to the victim’s computer in order to extract the firmware, inject the malicious code then overwrite the motherboard’s firmware.
Since the goal of this rootkit is to allow the running of malicious code at the kernel level of the operating system, the infection chain is highly complex, a lot more than for any usual malware infection. The UEFI code runs before the Windows system is loaded, which means the attacker has to somehow find a way to pass the malicious code to the operating system before it is launched, while the UEFI code will have been terminated.
The attacker achieves this by setting multiple hooks in succession, allowing the malicious code to be executed after the operating system has been launched (Figure A).
During the infection chain, the rootkit takes care of disabling Kernel Patch Protection (KPP) , also known as PatchGuard, a 64-bit Windows security mechanism preventing modifications in key structures of the Windows kernel in memory.
At the end of the operating system boot, the CosmicStrand rootkit allocates a buffer in the kernel’s address space and maps a shellcode there, before executing it.
SEE: Mobile device security policy (TechRepublic Premium)
The kernel level malicious payload
The shellcode run by the rootkit waits for a new thread in winlogon.exe and then executes a callback in this context, which is high-privileged. It then sleeps for 10 minutes before testing internet connectivity. That test is done via the Transport Device Interface instead of using the usual high-level API functions, and sends a DNS request to Google’s DNS server or to a custom one located in China.
If internet connectivity is available, the shellcode retrieves the final payload at a C2 server update.bokts[.]com. The payload is expected from CosmicStrand to be received in chunks of 528 bytes following a particular structure, probably to defeat automated analysis tools.
That last payload could not be retrieved by Kaspersky, but the researchers instead found a user-mode sample in the memory of one of the infected computers they could analyze. That sample, which is believed to be linked with CosmicStrand, creates a user named “aaaabbbb” on the targeted machine and adds that user to the local administrators group.
A long-running threat targeting individuals
Kaspersky discovered older versions of the rootkit that reached another C2 server to obtain additional shellcode. These older versions might have been used between the end of 2016 and mid-2017, while the latest version was active in 2020. An earlier version of the rootkit has also been analyzed by Qihoo360 in 2017.
Analysis of data related to both C2 servers found by the researchers indicate that the domains had a long lifetime and resolved to different IP addresses during limited timeframes, outside of which the rootkit would have been inoperative.
Regarding the targets of the CosmicStrand threat, Kaspersky noted that all victims in their telemetry appear to be private individuals using the free version of their product, located in China, Vietnam, Iran and Russia.
Probable Chinese threat actor
According to Kaspersky, several data leads to believe that “CosmicStrand was developed by a Chinese-speaking threat actor, or by leveraging common resources shared among Chinese-speaking threat actors.”
MyKings botnet uses a number of code patterns also observed in CosmicStrand, which is believed to have been developed by Chinese-speaking threat actors. Both threats also share identical tags when allocating memory in kernel mode and generate network packets the same way. The API hashing code used in both is also identical and has only been found in two other rootkits according to Kaspersky, MoonBounce and xTalker, also tied to Chinese-speaking threat actors.
SEE: Password breach: Why pop culture and passwords don’t mix (free PDF) (TechRepublic)
How to detect rootkits?
Rootkits are particularly difficult to detect, especially when they use hardware capabilities that are out of the operating system, which is the case for the CosmicStrand rootkit.
Security software scanning computer activity at the lowest levels might detect unusual activity from rootkits and successfully detect it.
Another way to detect it is via all systems that are not infected by the rootkit but connected to the same network: it is possible to detect the malicious network activity just as for any other piece of malware by using Intrusion Detection Systems/Prevention Detection Systems (IDS/IPS).
If a computer is suspected of being running an UEFI rootkit, incident responders might check the firmware for anomalies. A firmware that shows a different hash than the one provided by the vendor is probably compromised.
Finally, it needs to be understood that even if malicious files are removed from the Windows operating system, they will be reinstalled by the rootkit at every boot. A clean and safe version of the firmware needs to be installed to replace the malicious one.
Disclosure: I work for Trend Micro, but the views expressed in this article are mine.