Threat Research

Analysis of A New Golang Ransomware Targeting Linux Systems

By Kai Lu | October 08, 2019

A FortiGuard Labs Threat Analysis Report

During the past two months, I have been working on reverse engineering malware written in Golang. Go, also known as Golang, is a statically typed, compiled programming language designed at Google that is becoming more popular within the malware development community. In this blog, I will analyze a newly found Golang ransomware targeting Linux systems.

0x01 Overview of Go Binary

The sample being analyzed is a stripped ELF executable. A stripped executable can make reversing harder because you need to do extra work to restore symbols in the stripped binary. Thankfully, the redress tool can help us with this. Redress software is a tool for analyzing stripped Go binaries compiled with the Go compiler. It extracts data from the binary, then uses it to reconstruct symbols and perform analysis. 

The following is the output of analyzing this sample with the parameter “-src”. 

Figure 1. Output of source code after analyzing malware in redress

As shown in Figure 1, we can see the malware’s source code includes three Go files, all implemented functions, as well as their code line numbers. From the name of some functions, we can guess this malware should be ransomware. However, the number of lines in the source code is just over 300. So this ransomware isn’t complicated and might be in its initial development stage.

Next, let’s start dynamic debugging in debugger. Here, I use Radare2 as the debugger. Radare2 is able to analyze stripped Go binary and restore symbols by issuing an analysis command – which is why I chose Radare2 as the debugger for this project instead of GDB in Linux.

0x02 Dynamic Analysis of Go Binary

I issued the command “aaa” in Radare2 to perform an automatic analysis, and in Figure 2 we can see that Radare2 restores and recognizes the function names and symbols names very well. It really helped me debug the Go binary more efficiently.

Figure 2. Restored function names and symbols using Radare2

As we can see, the function init() is executed before the main function. The function check() is called in the function init(). In function check(), the malware first obtains the location information of the infected machine by sending an http request to hxxps://ipapi.co/json/. It then filters out Belarus (BY), Russia (RU), and Ukraine (UA) in order to prevent itself from running if the malware is being executed in one of those countries.

Figure 3. Filtering out Belarus (BY), Russia (RU), and Ukraine (UA)

In the main() function, it first deletes the Go binary. Then it calls the function randSeq() to generate a random AES key where the size is 0x20 in bytes, like the following:

Figure 4. Generating a random AES key

Next, it calls the function makesecret(), which is used to encrypt the AES key with an RSA public key hard-coded in the binary. Inside this function, it calls the function EncryptPKCS1v15 to encrypt the given AES key using RSA encryption and the padding scheme from PKCS#1 v1.5.

Figure 5. The hard-coded RSA public key in the Go binary

The following is the data after RSA encryption.

Figure 6. The Encrypted AES key after using RSA encryption

Next, it calls the function EncodeToString in the Golang package encoding/base64 to encode the previously encrypted data with a base64 algorithm. 

Figure 7. Encoded encrypted AES key with Base64

It then forms a buffer for the decrypted README file, shown in Figure 8.

Figure 8. The buffer of the decrypted README file

We can see that the encrypted AES key is written into the decrypted readme file with Base64 encoding.

Before the ransomware encrypts files, it kills the following list of processes by issuing the commands “service stop [pname]” or “systemctl stop [pname]”.

Figure 9. The services to be stopped

When it attempts to stop apache2.service, it prompts a dialog titled “Authentication Required” to instruct the user to input the system password to finish this operation.

Figure 10. apache2.service authentication dialog prompt

Finally, the malware starts to traverse the root directory “/” by calling the function Walk(root string, walkFn WalkFunc) in the Golang package “path/filepath” and then encrypting files.

Figure 11. Traverse root directory and encrypt files

The malware also has a blacklist of directories for encryption. The following is the directory blacklist.

Figure 12. The blacklist of directories for encryption

The malware encrypts files using the AES-256-CFB algorithm, and the encrypted files have a name that concatenates the original name with an “.encrypted” extension. The README file for decryption is shown in Figure 13.

Figure 13. The README file for decryption

The function EncFile() is used to encrypt files. It first gets the size of the file to be encrypted. If the file size is less than 0x986880(1,000,000) in bytes, it encrypts all of the file data with an AES-256-CFB algorithm. Otherwise, it reads the first 0x986880(1,000,000) bytes of data and encrypts them, then copies the remaining data of the original file to the end of the encrypted file.

Figure 14. Checking the size of the file to be encrypted

Figure 15. The data after encryption for files larger than 0x989680 in bytes

0x02 Conclusion

According to our analysis, we can see that this ransomware isn’t complicated and might be in an initial development stage. We should be aware that more and more malware is being developed with Golang, and I will constantly monitor and filter malware written in this new programming language.

0x03 Solution

This malicious elf file has been detected as “ELF/Cryptor.B!tr” by the FortiGuard AntiVirus service.

0x04 Reference

SHA256: 50470f94e7d65b50bf00d7416a9634d9e4141c5109a78f5769e4204906ab5f0b

IoC: fullofdeep<at>protonmail.com

REDRESS: https://go-re.tk/redress/

Radare2: https://rada.re/r/

Learn more about FortiGuard Labs and the FortiGuard Security Services portfolioSign up for our weekly FortiGuard Threat Brief.

Read about the FortiGuard Security Rating Service, which provides security audits and best practices.