CVE-2022-22817: Python Pillow RCE Vulnerability

CVE-2022-22817 Overview

CVE-2022-22817 is a critical code injection vulnerability in the Python Pillow library, specifically affecting the PIL.ImageMath.eval function. In versions prior to 9.0.0, this function allows evaluation of arbitrary Python expressions, including those that use dangerous built-in methods like exec. Attackers can exploit this flaw to execute arbitrary code by supplying malicious expressions, potentially through lambda expressions or direct Python code injection.

Critical Impact

Remote attackers can achieve full system compromise by injecting arbitrary Python code through the ImageMath.eval function, leading to complete loss of confidentiality, integrity, and availability.

Affected Products

• Python Pillow versions prior to 9.0.0
• Debian Linux 9.0 (Stretch)
• Debian Linux 10.0 (Buster)
• Debian Linux 11.0 (Bullseye)

Discovery Timeline

• 2022-01-10 - CVE CVE-2022-22817 published to NVD
• 2024-11-21 - Last updated in NVD database

Technical Details for CVE-2022-22817

Vulnerability Analysis

The PIL.ImageMath.eval function in Pillow is designed to evaluate mathematical expressions on images for pixel manipulation and image processing operations. However, the implementation failed to properly restrict the Python built-in functions available during expression evaluation. This oversight allows attackers to leverage dangerous Python built-ins such as exec, eval, and __import__ within the expression context.

The vulnerability is particularly severe because the function accepts string input that is directly evaluated as Python code. When applications pass user-controlled input to ImageMath.eval, attackers can craft malicious expressions that escape the intended mathematical context and execute arbitrary Python code with the same privileges as the running application.

Root Cause

The root cause of this vulnerability lies in the insufficient restriction of Python built-in functions exposed to the ImageMath.eval evaluation environment. Prior to version 9.0.0, Pillow did not implement adequate sandboxing or filtering of dangerous built-ins, allowing the full Python runtime to be accessible through crafted expressions. This represents a classic case of unsafe code evaluation where input validation was insufficient to prevent code injection attacks.

Attack Vector

This vulnerability is exploitable over the network when applications accept untrusted input and pass it to the PIL.ImageMath.eval function. The attack requires no authentication and no user interaction, making it particularly dangerous for web applications that process user-supplied image manipulation parameters. An attacker can craft a malicious expression containing Python code using the exec method or lambda expressions to achieve arbitrary code execution.

The exploitation technique involves injecting Python expressions that invoke dangerous built-in functions. For example, an attacker could inject an expression using exec() to import system modules and execute shell commands, or use lambda functions to bypass simple keyword filtering. The attack surface extends to any application that uses ImageMath.eval with user-controllable input.

Detection Methods for CVE-2022-22817

Indicators of Compromise

• Unusual process spawning from Python applications that use Pillow for image processing
• Unexpected network connections originating from image processing services
• Log entries showing errors related to PIL.ImageMath.eval with suspicious expression strings
• Presence of shell commands or Python code in application logs where image parameters are expected

Detection Strategies

• Monitor Python application logs for expressions containing dangerous keywords like exec, eval, __import__, or lambda in image processing contexts
• Implement application-level input validation to detect and block potentially malicious mathematical expressions
• Deploy runtime application self-protection (RASP) solutions to detect code injection attempts in Python applications
• Use SentinelOne's behavioral AI to detect anomalous process behavior from Python applications

Monitoring Recommendations

• Enable detailed logging for all applications utilizing the Pillow library's ImageMath module
• Set up alerts for process execution chains that originate from Python web applications or image processing services
• Monitor file system changes and network activity from processes handling image manipulation
• Implement canary tokens in sensitive directories to detect lateral movement following exploitation

How to Mitigate CVE-2022-22817

Immediate Actions Required

• Upgrade Python Pillow to version 9.0.1 or later immediately, as this version includes comprehensive fixes for the vulnerability
• Audit all applications using the PIL.ImageMath.eval function to identify potential exposure to untrusted input
• Implement input validation to sanitize any user-supplied data before passing it to image processing functions
• Consider temporarily disabling features that rely on ImageMath.eval if immediate patching is not possible

Patch Information

The Pillow development team addressed this vulnerability in version 9.0.0 by restricting the built-in functions available to ImageMath.eval. Additional security hardening was included in version 9.0.1. Organizations should update to the latest stable Pillow release to ensure complete protection. Detailed information about the security fixes is available in the Pillow Release Notes 9.0.0 and Pillow Release Notes 9.0.1.

For Debian systems, security updates are available through the standard package repositories. Refer to Debian Security Advisory DSA-5053 for platform-specific guidance.

Workarounds

• Avoid passing user-controllable input to PIL.ImageMath.eval until the library can be updated
• Implement a whitelist-based input validation that only allows known-safe mathematical expressions
• Run image processing services in isolated containers with restricted system access to limit the impact of potential exploitation
• Use network segmentation to prevent lateral movement if image processing services are compromised

# Upgrade Pillow to patched version
pip install --upgrade pillow>=9.0.1

# Verify installed version
pip show pillow | grep Version