OpenSSL: Buffer overflow in EVP_EncodeBlock() (CVE-2026-121481, CVSS 9.8)

A newly discovered buffer overflow in OpenSSL 4’s base64 encoder (EVP_EncodeBlock, CVE-2026-121481, rated CVSS 9.8) is prompting concern, but many note that its real-world impact is limited because it only affects large inputs and OpenSSL 4 is not yet widely shipped by enterprise distros. Commenters use the case to criticize OpenSSL’s long‑standing C APIs as easy to misuse, debate whether CVSS scores meaningfully reflect practical risk, and contrast C’s memory‑safety issues with alternatives like Rust, LibreSSL, and BoringSSL. The flood of high‑severity vulnerabilities and AI-driven bug discovery is also raising worries about “vulnerability fatigue” and the need for better systemic defenses and more sustainable, well-funded security-critical infrastructure.

Scope and Affected Systems

  • Vulnerability is in OpenSSL 4.x’s EVP_EncodeBlock (Base64 encoder), not in 1.x/3.x.
  • Most enterprise/distros are still on OpenSSL 3.x; bleeding‑edge distros (e.g., Fedora, Arch) ship 4.x and are affected but expected to patch quickly.
  • Some commenters initially confused by lack of updates on Debian/enterprise; clarified that’s because they don’t ship 4.x yet.

Nature of the Vulnerability

  • Bug: for very large inputs (around 1.5 GB on 32‑bit int systems), length calculation overflows and data is written past the caller’s buffer.
  • API requires callers to pre‑compute output buffer size via a nontrivial Base64 length formula and does not take an output length parameter, so the function cannot validate buffer size.
  • Several participants call this “classic bad C API design” and note that OpenSSL also implemented it incorrectly in 4.x.

Exploitability and Real‑World Risk

  • No known attacker‑controlled path yet; a CVE note labels severity “low” in practice despite the 9.8 base score.
  • Likely only risky if applications pass extremely large, attacker‑controlled blobs directly into this encoder.
  • TLS itself is probably unaffected because its length fields are smaller and don’t use Base64 here; some argue callers must be audited to judge real risk.

CVSS 9.8 Debate

  • Strong skepticism about CVSS: scores are seen as “vibes” or “Ouija board,” often inflated and context‑free.
  • Concern that constant 9.x scores cause “boy who cried wolf” fatigue and unnecessary emergency patch cycles.
  • Others stress CVSS is only a baseline input to risk assessment; environment and temporal factors are rarely filled in but matter a lot.

OpenSSL Quality and Design

  • Longstanding criticism that OpenSSL APIs are low‑quality, hard to use correctly, and historically security‑fragile (Heartbleed cited).
  • Debate over C practices: size_t vs signed ints, passing explicit buffer lengths, better error returns, and lack of standard safe buffer types.
  • Some argue this is an example of why memory‑safe or higher‑level languages (Rust, Ada, etc.) and better APIs are needed; others note crypto still needs assembly and constant‑time guarantees.

Alternatives and Ecosystem

  • Alternatives mentioned: BoringSSL, LibreSSL, Rustls, AWS‑LC, s2n‑tls, Go’s crypto/tls, Conscrypt.
  • These are used in browsers, major platforms, or specific ecosystems, but OpenSSL remains the default in many distros.

Operational Fatigue, Tooling, and AI

  • Sysadmins and bank engineers report “vulnerability fatigue”: constant patching, CI blocking on CVEs, and fear of breakage.
  • Some see rising bug discovery (often AI‑assisted) as positive (bugs are being fixed); others worry about unstable infrastructure.
  • Discussion of AI both as source of more findings and as future tooling for remediation, monitoring, and possibly unsafe auto‑rewrites, with strong emphasis on the need for human review.