Threat Intelligence

PHP2Shell: How Attackers Turned a Web Server Bug Into Root Access on Firewalls

The Hacker News · 19 Nov 2026 · updated 21 Nov 2026
Key Takeaway Make sure the management pages of your firewalls and other network devices are not reachable from the internet, and restrict them to trusted internal addresses.

On October 28, Palo Alto Networks confirmed that attackers were exploiting two PHP vulnerabilities, CVE-2026-109618 and CVE-2026-109619, against PAN-OS management interfaces exposed to the internet. Both flaws let anyone run code on the web server without logging in. Details leaked on October 20, ahead of the coordinated release, and the PHP project shipped an emergency patch the next day. The two reports differ on the date of the confirmation: The Hacker News reports October 28, while Sophos reports 29 October.

The flaws are not easy to exploit reliably, but appliances change the equation. Every device on a given software version runs the same binaries, so an attacker can study them in advance. Dream, working with Palo Alto Networks, found the attacker had obtained PAN-OS firmware, decrypted it, extracted the PHP libraries and built memory profiles for nine software versions. The attacker then chained PHP2Shell with a Linux kernel flaw (Fragnesia, CVE-2026-46300) to gain root on 25 systems, 24 firewalls and one management server, in the U.S., Japan, Taiwan and South Korea. According to Sophos, the activity is described as targeted and does not suggest a capability for broad exploitation of other platforms. PAN-OS also ships end-of-life PHP versions (8.0.30 and 7.4.33).

The bugs sit in how PHP handles a crafted file upload that pushes it past its memory limit before any application code runs. Small requests can trigger this, so a request well under the default 8 MB post size limit can still exhaust the 128 MB memory limit. Web application firewalls are often not adequate here, and blocking uploads is not realistic for many applications. Intruder found that nginx in front of PAN-OS buffered chunked request bodies, which accidentally broke public exploit chains.

Risk was also hard to judge. Scores ranged from 9.2 (CVSS 4.0, PHP project) to 9.8 (Fortinet, CVSS v3), while Palo Alto first scored it 8.7 at moderate urgency before raising it to 9.2 Critical on October 27. Many public proofs of concept did not work, so noisy logs said little about real exposure. According to Sophos, proof-of-concept code is public and exploitation is viable, but whether an attacker gets remote code execution depends on deployment architecture, PHP version and configuration. Sophos adds that building a working exploit is difficult, even with AI help, and some researcher claims may only work in labs.

Sophos telemetry is quieter. It says an early report of a successful compromise was corrected, as its data showed scanning and a few failed attempts. One blocked attempt carried a payload that appears to be Potassium, a Mirai-derived botnet variant that can run DDoS attacks and act as a downloader. Sophos has seen no successful compromise, no exploitation in its customer telemetry, and no link to ransomware, data theft or supply chain compromise. It also notes a separate PHP advisory on 23 October for another remote code execution flaw (CVE-2026-115560); the mitigations for CVE-2026-109619 also address it. Palo Alto's guidance is to remove internet access to management interfaces and limit them to trusted internal IP addresses.

Why it matters

Organisations that treat firewall admin pages as network gear rather than web applications leave their most privileged devices exposed to web-style attacks. The gap between Palo Alto and Dream's confirmed root compromises and Sophos's customer telemetry, which shows no successful exploitation, suggests risk is concentrated in exposed management interfaces rather than evenly spread. Log noise and a single CVSS number are poor guides, so exposure should drive priority.

What to do

  • Remove internet access to PAN-OS management interfaces and limit them to trusted internal IP addresses.
  • Inventory every management web interface across your estate and confirm none is exposed to the internet.
  • Apply the PHP mitigations for CVE-2026-109619, which Sophos says also address CVE-2026-115560.
  • Review logs for failed attempts and unexpected downloads, rather than relying on log volume alone to judge exposure.
Also reported by Sophos

Summarised by CISO AI from The Hacker News, with the same story also reported by Sophos, written with Claude Sonnet 5.5. We link back to every original so you can read it yourself.