HighGHSA-6qrw-m8cv-hcxfCVE-2026-119192

Forge: RSA PKCS#1 v1.5 signature forgery via non-canonical DigestInfo encoding

Published 10 Nov 2026 · Updated 10 Nov 2026

Is your project on an affected node-forge?

Private repo? Upload its lockfile instead, or connect GitHub to check every repo.

Affected package

Affected
<1.4.1
Fixed in
1.4.1
Update to the fixed version for your release line, for example:npm install node-forge@1.4.1

Details

Summary

RSASSA-PKCS1-v1_5 signature verification (publicKey.verify() with the default scheme, also used by X.509 certificate and PKCS#7 verification) accepted non-canonical BER encodings of the DigestInfo, such as NULL parameters with content and non-minimal lengths. These leave attacker-controlled bytes inside an accepted DigestInfo, which allows signatures to be forged without the private key for RSA keys with a low public exponent (e=3). Fixed in node-forge 1.4.1.

Details

In lib/rsa.js, key.verify() with the default RSASSA-PKCS1-V1_5 scheme removes the padding, parses the recovered DigestInfo with the lenient BER parser asn1.fromDer() and checks its structure with asn1.validate() against digestInfoValidator. Neither requires a canonical DER encoding, so verification accepted, among others:

  • NULL parameters with content bytes (e.g. 05 40 <64 bytes>)
  • non-NULL or constructed values in the parameters position
  • non-minimal (long form) and indefinite lengths
  • non-canonical OID encodings (padded subidentifiers, trailing bytes)

These leave attacker-controlled bytes inside an accepted DigestInfo. For RSA keys with a low public exponent (e=3) this allows signatures to be forged without the private key (BERserk / Bleichenbacher-style forgery, see CVE-2014-1568).

RFC 8017 section 8.2.2 verifies a signature by encoding the expected EMSA-PKCS1-v1_5 message and comparing it with the recovered one; section 9.2 defines the DigestInfo encoding.

The fix adds _encodeDigestInfo(), which DER-encodes the expected DigestInfo from the digest algorithm OID and the digest (with and without NULL parameters). Verification now requires the recovered DigestInfo to match one of these encodings byte for byte. The same change fixes GHSA-2g46-22mc-6gwp (CVE-2026-85393).

Patches

Fixed in node-forge 1.4.1. Verification now re-encodes the expected DigestInfo and requires a byte-for-byte match (RFC 8017 section 8.2.2).

Workarounds

Do not use RSA keys with a low public exponent (e=3) for signature verification.

References

  • https://www.rfc-editor.org/rfc/rfc8017#section-8.2.2
  • https://www.rfc-editor.org/rfc/rfc8017#section-9.2
  • CVE-2026-85393 / GHSA-2g46-22mc-6gwp (related, same fix)
  • CVE-2026-33894 / https://github.com/digitalbazaar/forge/security/advisories/GHSA-ppp5-5v6c-4jwp
  • https://www.cve.org/CVERecord?id=CVE-2014-1568 (BERserk)

Severity

CVSS 3.1: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:N (7.5, High) CVSS 4.0 (informational): CVSS:4.0/AV:N/AC:L/AT:P/PR:N/UI:N/VC:N/VI:H/VA:N/SC:N/SI:N/SA:N (8.2) Both versions rate this High. The severity rating follows CVSS 3.1.

Credits

Independently reported by Oren Yomtov (@orenyomtov, Fireblocks), Bronson Yen (Calif.io, in collaboration with Claude and Anthropic Research), Kamil Hałajda, @pzartech, @antonlissone and @Brechard.

Proof of concept

DigestInfo encodings accepted before 1.4.1 (SHA-256; OID = 06 09 60 86 48 01 65 03 04 02 01, H = the 32-byte digest):

NULL parameters with content: 30 33 30 0F <OID> 05 02 BE EF 04 20 <H> non-minimal DigestInfo length: 30 81 31 30 0D <OID> 05 00 04 20 <H> (canonical form: 30 31 30 0D <OID> 05 00 04 20 <H>)

Long-form lengths are accepted the same way for the DigestAlgorithm, OID, NULL and digest fields. A signature whose EMSA-PKCS1-v1_5 block (00 01 FF..FF 00 || DigestInfo) contains such an encoding verified as valid: publicKey.verify(digest, signature) returned true. Content bytes in the NULL parameters (reports used e.g. 05 40 <64 bytes>) and similar slack give an attacker room to construct a forged signature for an e=3 key. With 1.4.1, verification throws "ASN.1 object does not contain a valid RSASSA-PKCS1-v1_5 DigestInfo value."

Regression tests: tests/unit/rsa.js, "canonical DigestInfo checks". They use an e=1 key, so the signature equals the encoded block, with one test per non-canonical shape (also indefinite length, constructed and non-NULL parameters, padded OID subidentifiers) and two tests that the canonical encodings with and without NULL parameters still verify. Included in the 1.4.1 release.

Impact

Applications that verify RSASSA-PKCS1-v1_5 signatures with node-forge: publicKey.verify() with the default scheme, and X.509 certificate and PKCS#7 signature verification, which use it. If the verifying RSA public key has a low public exponent (e=3), an attacker can forge a signature that verifies as valid, without the private key. Only keys with a small public exponent such as e=3 are practically affected.

Is GHSA-6qrw-m8cv-hcxf in any of your client repos?

Checking one package by hand is fine. Checking it across forty client projects is not. NPMscan reads the lockfiles in every GitHub repo you pick, finds each project with an affected node-forge, and gives you the command that fixes it.

More advisories for node-forge

  • LowCVE-2026-125335Forge: Ed25519 signature verification accepts small-order public keys · 10 Nov 2026
  • MediumCVE-2026-125332Forge: TLS client crash on an unsupported ServerHello compression method · 10 Nov 2026
  • MediumCVE-2026-125333Forge: ReDoS in forge.http response header and cookie trimming · 10 Nov 2026
  • MediumCVE-2026-125334Forge: TLS client crash on a handshake record received without a session · 10 Nov 2026
  • LowCVE-2026-118774Forge: Certificate signature verified with the issuer's RSA signature scheme · 10 Nov 2026
  • LowCVE-2026-118776Forge: Incomplete PKCS#7 padding validation in AES/DES CBC and ECB decryption · 10 Nov 2026
All 36 advisories for node-forge

References

Advisory data from OSV.dev (opens OSV.dev in a new tab), refreshed daily.