Format String Bug Found in SurrealDB
A critical vulnerability identified in SurrealDB's rquickjs component allows for potential memory access or code execution.
A newly disclosed security vulnerability has emerged within the SurrealDB platform, centered on its handling of internal scripting errors. The flaw resides in the rquickjs component, specifically within the Exception::throw_type function, creating a condition that could have significant implications for system integrity.
The Nature of the Flaw
Identified as CVE-2024-58366, this security issue is classified as a format string vulnerability. The bug is present in all versions of the database platform preceding 1.1.1. When scripting functionality is active, the database's error-handling mechanism fails to sanitize inputs correctly.
Impact on System Security
The technical defect allows for the submission of specific format string sequences within error inputs. According to the advisory, this provides a mechanism for an attacker to potentially read arbitrary memory segments. Furthermore, the vulnerability may permit the execution of code, operating under the same permissions as the SurrealDB process itself.
Technical Specifications
- Advisory ID: GHSA-qcfm-cmfw-f4x4
- Affected Versions: Before 1.1.1
- Severity Rating: Critical
- CVSS Score: 8.5
Understanding the Risk
The discovery of this vulnerability highlights a technical risk for users of the database who have enabled scripting features. Because the flaw allows for unauthorized memory access and potential code execution, the primary concern centers on the scope of the process privileges held by the database instance. Given that these permissions are tied to the execution environment of the software, any unauthorized interaction with the Exception::throw_type function could result in a breach of the expected isolation between user input and system processes.
Sources
- GitHub Security Advisories Original source
Continue Reading
LXD, MicroCloud flaw lets clients be impersonated
Canonical patched a maximum-severity LXD and MicroCloud bug that let attackers assume a trusted client's identity without its private key.
A hands-on way to gauge AI ransomware risk
OWASP volunteers released RAID-AI, a free manual self-assessment that maps AI controls to NIST and CISA frameworks.
AWS AgentCore Flaw Let One Prompt Hijack All Agents
Researchers say a single chat prompt could extract cloud credentials and let an attacker take over every AgentCore agent in an AWS account and region.