CVE-2020-8929

### Impact Tink's Java version before 1.5 under some circumstances allowed attackers to change the key ID part of the ciphertext, resulting in the attacker creating a second ciphertext that will decrypt to the same plaintext. This can be a problem in particular in the case of encrypting with a deterministic AEAD with a single key, and relying on the fact that there is only a single valid ciphertext per plaintext. No loss of confidentiality or loss of plaintext integrity occurs due to this problem, only ciphertext integrity is compromised. ### Patches The issue was fixed in this [pull request](https://github.com/google/tink/commit/93d839a5865b9d950dffdc9d0bc99b71280a8899). ### Workarounds The only workaround is to backport the fixing [pull request](https://github.com/google/tink/commit/93d839a5865b9d950dffdc9d0bc99b71280a8899). ### Details Tink uses the first five bytes of a ciphertext for a version byte and a four byte key ID. Since each key has a well defined prefix, this extends non-malleability properties (but technically not indistinguishability). However, in the Java version this prefix lookup used a hash map indexed by unicode strings instead of the byte array, which means that invalid Unicode characters would be [replaced by U+FFFD](https://en.wikipedia.org/wiki/UTF-8#Invalid_sequences_and_error_handling) by the [Java API's default behavior](https://docs.oracle.com/javase/7/docs/api/java/lang/String.html#String(byte[],%20java.nio.charset.Charset)). This means several different values for the five bytes would result in the same hash table key, which allows an attacker to exchange one invalid byte sequence for another, creating a mutated ciphertext that still decrypts (to the same plaintext). ### Acknowledgements We'd like to thank Peter Esbensen for finding this issue and raising it internally. ### For more information If you have any questions or comments about this advisory: * Open an issue in [Tink](https://github.com/google/tink/issues)

A mis-handling of invalid unicode characters in the Java implementation of Tink versions prior to 1.5 allows an attacker to change the ID part of a ciphertext, which result in the creation of a second ciphertext that can decrypt to the same plaintext. This can be a problem with encrypting deterministic AEAD with a single key, and rely on a unique ciphertext-per-plaintext.