CVE-2023-45820

Provides defined handling (alert and additional actions) for the exceptional condition of audit logging failure.

Supplies a concrete handling action (safe mode) for exceptional conditions, mitigating risks from improper or absent handling that could allow continued attacks.

By preparing users for contingency scenarios, the control promotes proper handling of exceptional conditions instead of default or unsafe behaviors.

An updated contingency plan defines current actions for exceptional conditions, reducing the window for attackers to exploit improper handling leading to system failure.

Procedures ensure proper handling of exceptional conditions to support effective incident response.

Incident response testing confirms proper handling of exceptional conditions to limit exploit impact.

Gives users guidance on incident handling, reducing improper handling of exceptional conditions that could stem from exploited weaknesses.

Enforces structured response to exceptional conditions so the system cannot remain in an unsafe state.

Documented operating procedures may specify exception handling but do not guarantee implementation.

Logging captures unhandled exceptions, aiding detection but not preventing the weakness.

Monitoring can surface unhandled exceptions but does not enforce proper handling.

Secure SDLC mandates exception-handling requirements and testing that directly prevent improper handling of exceptional conditions.

A.8.26 explicitly lists resilience against malicious attacks including SQL injections, input validation, error message handling, and several integrity/authorization requirements that address exceptional conditions in a minority of cases, but most CWE-755 instances (e.g. unhandled nulls, resource errors, unexpected states) fall outside its named topics.

A.8.27 mandates security-by-design principles (fail securely, least privilege, defense-in-depth, distrust external input, assume-breach) plus design reviews that can catch and stop many but not all mishandlings of exceptional conditions before they reach production code; a bounded remainder (e.g. subtle runtime error paths, library-level faults, or post-review regressions) can still be present and exploitable.