JWX vulnerable to a denial of service attack using compressed JWE message

Description

Summary

This vulnerability allows an attacker with a trusted public key to cause a Denial-of-Service (DoS) condition by crafting a malicious JSON Web Encryption (JWE) token with an exceptionally high compression ratio. When this token is processed by the recipient, it results in significant memory allocation and processing time during decompression.

Details

The attacker needs to obtain a valid public key to compress the payload. It needs to be valid so that the recipient can use to successfully decompress the payload. Furthermore in context JWT processing in the v2 versions, the recipient must explicitly allow JWE handling .

The attacker then crafts a message with high compression ratio, e.g. a payload with very high frequency of repeating patterns that can decompress to a much larger size. If the payload is large enough, recipient who is decompressing the data will have to allocate a large amount of memory, which then can lead to a denial of service.

The original report includes a reference to [1], but there are some very subtle differences between this library and the aforementioned issue. The most important aspect is that the referenced issue focuses on JWT processing, whereas this library is intentionally divided into parts that comprise JOSE, i.e. JWT, JWS, JWE, JWK. In particular, v2 of this library does not attempt to handle JWT payload enveloped in a JWE message automatically (v1 attempted to do this automatically, but it was never stable).

Reflecting this subtle difference, the approach taken to mitigate this vulnerability is slightly different from the referenced issue. The referenced issue limits the size of JWT when parsing, but the fixes for this library limits the maximum size of the decompressed data when decrypting JWE messages. Therefore the fix in this library is applicable regardless of the usage context, and a limit is now imposed on the size of the message that our JWE implementation can handle.

Proof of Concept

Modified from the original report to fit the vulnerability better:

// The value below just needs to be "large enough" so that the it puts enough strain on the
// recipient's environment. The value below is a safe size on my machine to run the test
// without causing problems. When you increase the payload size, at some point the processing
// will be slow enough to virtually freeze the program or cause a memory allocation error
const payloadSize = 1 << 31

privkey, err := rsa.GenerateKey(rand.Reader, 2048)
require.NoError(t, err, `rsa.GenerateKey should succeed`)
pubkey := &privkey.PublicKey
payload := strings.Repeat("x", payloadSize)

encrypted, err := jwe.Encrypt([]byte(payload), jwe.WithKey(jwa.RSA_OAEP, pubkey), jwe.WithContentEncryption("A128CBC-HS256"), jwe.WithCompress(jwa.Deflate))
require.NoError(t, err, `jwe.Encrypt should succeed`)
_, err = jwe.Decrypt(encrypted, jwe.WithKey(jwa.RSA_OAEP, privkey)) // Will be allocating large amounts of memory
require.Error(t, err, `jwe.Decrypt should fail`)

References

[1] CVE-2024-21319

Basic information

Type
reviewed
Severity
medium
Advisory on GitHub
Open advisory ↗
Repository advisory
Open repository advisory ↗
Source code
Browse source ↗
Published (advisory)
2024-03-08 15:06:53 UTC
Updated
2024-03-11 03:37:16 UTC
GitHub reviewed
2024-03-08 15:06:53 UTC
NVD published
2024-03-08

EPSS Score

Score Percentile
0.15% 35.65%

CVSS Scores

Base score Version Severity Vector
6.8 3.1
CVSS:3.1/AV:N/AC:L/PR:H/UI:N/S:C/C:N/I:N/A:H Click to expand
Attack vector (AV:N)
Could be attacked over the internet or any normal routed network—not just someone sitting at the machine.
Attack complexity (AC:L)
Once they can reach the bug, pulling it off is straightforward—no weird race conditions or rare setup.
Privileges required (PR:H)
They need powerful rights—admin, root, or similar—before this pays off.
User interaction (UI:N)
Nobody has to click “OK” or open a trap file; it can work without a victim helping.
Scope (S:C)
Breaking this can reach past the original component and bite other resources—bigger blast radius.
Confidentiality (C:N)
Doesn’t really leak secrets in a meaningful way.
Integrity (I:N)
Data isn’t meaningfully altered or forged.
Availability (A:H)
Could take the service down hard or make it unusable for people who depend on it.

Identifiers

CWEs

CWE id Name
CWE-400 Uncontrolled Resource Consumption

Credits

  • zer0yu (reporter)

Affected packages (2)

Vulnerable version ranges and first patched releases as published by GitHub.

Ecosystem Package Vulnerable range First patched Vulnerable functions
go github.com/lestrrat-go/jwx/v2 < 2.0.21 2.0.21
go github.com/lestrrat-go/jwx < 1.2.29 1.2.29

References

cvelogic Threat Intelligence