A newly documented Linux rootkit is giving attackers a dangerous edge: the ability to make their malware functionally invisible to the very tools administrators rely on to check what’s running on a system. Named GhostMap by the Aqua Security researchers who found it, the rootkit was discovered hiding IronWorm, a Rust-based malware strain previously delivered through compromised npm packages, and has since turned up in a second, possibly unrelated campaign targeting Arch Linux users.
Hiding in Plain Sight, Selectively
GhostMap isn’t the malware itself — it’s the concealment layer bolted onto it. Once installed, it can make specific processes, files, and network sockets disappear from the standard listings that administrators and security tools check, without actually removing anything from the system. In testing, Aqua researchers watched a malware listener vanish entirely from /proc/net/tcp, one of the most common places to check for suspicious network activity — yet the same listener remained fully visible to ss, a separate Linux utility that reads socket data through a different kernel interface entirely.
That inconsistency is itself a clue for defenders: GhostMap’s concealment isn’t uniform across every tool, which means cross-checking multiple utilities against each other can reveal exactly the kind of mismatch that signals tampering.
Abusing eBPF, a Legitimate Kernel Feature
The rootkit works by abusing eBPF (extended Berkeley Packet Filter), a legitimate Linux kernel feature that allows small, sandboxed programs to run at specific points inside the kernel for purposes like performance monitoring and network filtering. Aqua’s researchers, Dolev Mizrahi and Ivan Kosarev, counted 13 separate eBPF programs and 10 supporting data maps making up GhostMap’s implementation — a reminder that the same kernel capability prized by performance engineers and observability platforms can be repurposed for exactly the opposite goal.
In directory listings, the researchers found that altering just a single byte was enough to make a reader silently skip selected entries. Critically, the underlying data was never deleted — it remained fully resident in memory and the total amount of data returned didn’t even change, making the tampering invisible unless you already knew what you were looking for. Test processes dropped out of standard ps output but could still be reached directly through their known /proc/PID paths, and reappeared the instant the rootkit was deactivated.
Two Campaigns, One Concealment Tool
IronWorm first came to attention through a supply-chain attack distributed via the npm package registry, where it focused on stealing developer secrets such as credentials and API keys from infected machines. Aqua’s new research picks up the story after installation, examining what happens once a Linux service launches the implant and GhostMap begins altering what local tools can see. In their tests, a sample bundled inside the npm package weavedb-sdk@0.45.3 copied itself to /var/lib and established a persistent systemd service, which then loaded the rootkit — a process chain investigators need to follow carefully, since tracking only the original installer’s child processes would miss the step where concealment actually begins.
Separately, Aqua found matching eBPF program and data-map names inside a sample tied to the “Atomic Arch” campaign, which has been traced to compromised Arch Linux User Repository (AUR) build scripts. The researchers were careful to note that shared tooling doesn’t necessarily mean a shared operator, but the overlap gives defenders a strong reason to look beyond npm packages alone when hunting for GhostMap.
Detection Is Possible, But Not a Silver Bullet
Encouragingly, Aqua’s own open-source runtime security tool, Tracee, successfully flagged three of GhostMap’s eBPF programs the moment they attached to the kernel, based on their memory-write capabilities, and retained a complete record of directory-reader activity throughout testing. That demonstrates detection is achievable — but researchers stress it doesn’t equal prevention, and a clean-looking system scan is no guarantee that nothing malicious is running underneath.
For organizations with potentially exposed Linux hosts, Aqua recommends a layered response:
- Cross-check process and socket listings using multiple independent tools, since GhostMap’s concealment is not applied consistently everywhere.
- Review systemd services and eBPF program attachments for anything unexpected, especially shortly after installing new packages.
- Preserve executable metadata, including extended file attributes, before removing any suspect component — a matching cryptographic hash confirms a specific artifact, while a suspicious path or attribute name alone does not.
- Extend supply-chain scrutiny beyond npm to other package ecosystems, given GhostMap’s apparent reuse across unrelated delivery channels.
With eBPF increasingly embedded in both legitimate security tooling and now in rootkits like GhostMap, the arms race over what Linux administrators can actually trust their own monitoring commands to show them looks set to intensify.
Leave a Reply
You must be logged in to post a comment.