There are 3 open security issues in bookworm.
1 important issue:
- CVE-2026-41565:
CryptX versions before 0.088_001 for Perl have a stack buffer overflow in four AEAD decrypt_verify helpers. The gcm_decrypt_verify, ccm_decrypt_verify, chacha20poly1305_decrypt_verify and eax_decrypt_verify XS routines copied the caller-supplied authentication tag into a fixed 144-byte stack buffer (MAXBLOCKSIZE) without checking the supplied length. A longer tag overwrites the stack past the buffer. Version 0.088 added the clamp to gcm_decrypt_verify, and 0.088_001 added it to the other three. Any caller of an affected helper that forwards an attacker-controlled tag longer than the buffer can trigger the overflow.
2 issues left for the package maintainer to handle:
- CVE-2025-40914:
(needs triaging)
Perl CryptX before version 0.087 contains a dependency that may be susceptible to an integer overflow. CryptX embeds a version of the libtommath library that is susceptible to an integer overflow associated with CVE-2023-36328.
- CVE-2026-41564:
(needs triaging)
CryptX versions before 0.088 for Perl do not reseed the Crypt::PK PRNG state after forking. The Crypt::PK::RSA, Crypt::PK::DSA, Crypt::PK::DH, Crypt::PK::ECC, Crypt::PK::Ed25519 and Crypt::PK::X25519 modules seed a per-object PRNG state in their constructors and reuse it without fork detection. A Crypt::PK::* object created before `fork()` shares byte-identical PRNG state with every child process, and any randomized operation they perform can produce identical output, including key generation. Two ECDSA or DSA signatures from different processes are enough to recover the signing private key through nonce-reuse key recovery. This affects preforking services such as the Starman web server, where a Crypt::PK::* object loaded at startup is inherited by every worker process.
You can find information about how to handle these issues in the security team's documentation.
![[bug history graph]](https://qa.debian.org/data/bts/graphs/libc/libcryptx-perl.png){kind=link}