Microsoft informed users on Wednesday that an update for the Windows Host Compute Service Shim library patches a critical remote code execution vulnerability.
Introduced in January 2017, the Windows Host Compute Service (HCS) is a low level container management API for Microsoft’s Hyper-V hypervisor. The tech giant has made available two open source wrappers that allow users to call the HCS from higher level programming languages instead of the C API directly.
One of these wrappers is the Windows Host Compute Service Shim (hcsshim), which supports launching Windows Server containers from the Go language. Hcsshim is mainly used in the Docker Engine project, but Microsoft says it can be freely used by others as well.
Swiss developer and security researcher Michael Hanselmann discovered that hcsshim fails to properly validate input when importing a container image, allowing a malicious actor to remotely execute arbitrary code on the host operating system.
“To exploit the vulnerability, an attacker would place malicious code in a specially crafted container image which, if an authenticated administrator imported (pulled), could cause a container management service utilizing the Host Compute Service Shim library to execute malicious code on the Windows host,” Microsoft said in its advisory.
The vulnerability, tracked as CVE-2018-8115, has been classified as critical, but Microsoft believes it is unlikely to be exploited for malicious purposes. Technical details of the issue have not been made public.
The flaw has been fixed with the release of hcsshim 0.6.10, which can be obtained from GitHub. US-CERT has also released an alert advising users to apply the update.
This is not the only out-of-band update released by Microsoft recently. Last month, the company updated its Malware Protection Engine to patch a vulnerability that can be exploited to take control of a system by placing a malicious file in a location where it would be scanned.
UPDATE. Hanselmann says he reported the vulnerability to both Microsoft and Docker in February. The researcher will release technical details and a proof-of-concept (PoC) exploit on May 9.