Research detects dangerous malware hiding in peripherals

A Berlin researcher has demonstrated the capability to detect previously undetectable stealthy malware that resides in graphics and network cards.
Patrick Stewin's proof of concept demonstrated that a detector could be built to find the sophisticated malware that ran on dedicated devices and attacked direct memory access (DMA).
The attacks launched by the malware dubbed DAGGER targeted host runtime memory using DMA provided to hardware devices. These attacks were not within scope of antimalware systems and therefore not detected.

DAGGER, also developed by Stewin and Iurii Bystrov of the FGSect Technical University of Berlin research group, attacked 32bit and 64bit Windows and Linux systems and could bypass memory address randomisation.
After beginning life last year as a keylogger, DAGGER has recently been upgraded with new functionality and now included the ability to update its attack behaviour during runtime via an out-of-band channel.
"DMA malware is stealthy to a point where the host cannot detect its presence," Stewin said.
In a paper Stewin will present next month, he said the DMA attacks were both dangerous and undetectable. (pdf)

"DMA-based attacks launched from peripherals are capable of compromising the host without exploiting vulnerabilities present in the operating system running on the host.
"Therefore they present a highly critical threat to system security and integrity. Unfortunately,to date no OS (operating system) implements security mechanisms that can detect DMA-based attacks. Furthermore, attacks against memory management units have been demonstrated in the past and therefore cannot be considered trustworthy."

The German Government funded research was closing in on its aim to develop a reliable detector for DMA malware.
"At the moment we have a proof-of-concept that proves that a detector is possible," Stewin said in an email to SC. "It can find DAGGER."
The proof-of-concept was based on a runtime monitor dubbed BARM which modelled and compared expected memory bus activity to the resulting activity, meaning malware residing on peripherals would be detected.
Stewin said the detector would not significantly drain compute resources.
Some detectors had been previously developed but they required that peripherals be modified or that a special debug feature exist.
The researchers aimed to develop the proof of concept into a detector that did not require modification.
The pair would present the research paper "A Primitive for Revealing Stealthy Peripheral-based Attacks on the Computing Platform's Main Memory" at the 16th International Symposium on Research in Attacks, Intrusions and Defenses in October in Saint Lucia.