Hackers use Linux backdoor on compromised e-commerce sites with software skimmer

Security researchers have revealed a new hacking campaign that installs a Linux backdoor on compromised e-commerce sites after deploying a credit card skimmer on merchant websites.

Researchers from the Sansec Threat Research Team discovered a new malicious agent “linux_avp” that hides as a system process on e-commerce servers. They said hackers have been deploying this malware worldwide since last week, and it takes commands from a control server in Beijing.

In the campaign, hackers started automated e-commerce attack probes, testing for dozens of weaknesses in common online store platforms.

“After a day and a half, the attacker found a file upload vulnerability in one of the store’s plugins. S/he then uploaded a webshell and modified the server code to intercept customer data,” said researchers.

Researchers said hackers then uploaded the linux_avp malware, which is a Golang program that starts, removes itself from disk, and disguises as a fake ps -ef process.

“Analysis of linux_avp suggests that it serves as a backdoor, waiting for commands from a Beijing (Alibaba) hosted server,” said researchers. The backdoor also revealed where the user, known as "dob" built the backdoor in a project folder lin_avp, using code name GREECE.

The malware also injects a malicious crontab entry to ensure access in case that the process is removed or the server rebooted. The crontab downloads the Golang malware executable to a random writable directory and installs two configuration files. “One contains a public key, which is presumably used to ensure that no one, but the malware owner can launch commands,” researchers added.

This case has another Chinese connection, according to researchers, as a line was added to the e-commerce platform code called app/design/frontend/favicon_absolute_top.jpg, which contains PHP code to retrieve a fake payment form and inject it in the store. Researchers said the IP for this was hosted in Hong Kong and was previously observed as a skimming exfiltration endpoint in July and August of this year.

Researchers said, at the time of writing, no other antivirus vendor had recognized the malware.

“Curiously, one individual had submitted the same malware to Virustotal on Oct 8th with the comment “test”. This was just one day after the successful breach of our customer’s store,” said researchers.

They added that the person uploading the malware could very well be the malware author, who wanted to assert that common antivirus engines will not detect their creation.