CVE-2026-90472
Medium · CVSS 6.9Published 2026-09-12
CVSS
6.9
Medium
EPSS
—
exploit probability
Active Exploit
No
not yet listed
What is CVE-2026-90472?
msgpack-java through 0.9.12 contains a stack overflow vulnerability in MessageUnpacker.unpackValue() that recursively deserializes arrays and maps without nesting depth limits. Attackers can craft payloads with deeply nested arrays to exhaust the deserializing thread's stack and trigger StackOverflowError, causing per-request deserialization failures.
CVSS Score Explained
CVSS 6.9 is Medium severity. Exploitation usually requires some specific conditions (local access, user interaction, or a particular config). Still worth patching in your next maintenance window.
EPSS Exploitation Probability
No EPSS score yet — model needs more data on this CVE.
References & Patches
- ›https://github.com/msgpack/msgpack-java
- ›https://github.com/msgpack/msgpack-java/blob/v0.9.12/msgpack-core/src/main/java/org/msgpack/core/MessageUnpacker.java#L646-L664
- ›https://github.com/msgpack/msgpack-java/issues/1015
- ›https://www.vulncheck.com/advisories/msgpack-java-through-0.9.12-stack-overflow-via-nested-arrays
Track vulnerabilities for your stack
Tell us what tech you run — Node.js, Python, Java, nginx, whatever — and get a weekly email when new CVEs affect it.
Recent CVEs
- CVE-2026-90554vLLM versions >=0.10.2 and <0.28.0 do not apply any audio decode-size or duration limit when extracting audio from video input for NanoNemotronVL models. In nano_nemotron_vl.py, _extract_audio_from_videos calls load_audio_pyav(BytesIO(video_bytes)) without the max_duration_s or max_decode_bytes parameters, so neither VLLM_MAX_AUDIO_DECODE_DURATION_S nor VLLM_MAX_AUDIO_DECODE_BYTES is enforced (unlike the direct audio upload path in AudioMediaIO). When a NanoNemotronVL model is served with use_audio_in_video=True, an attacker who supplies a small, highly compressed video as multimodal input can force the server to allocate gigabytes of memory during audio decoding, resulting in a denial of service. Fixed in vLLM 0.28.0.6.9
- CVE-2026-90555vLLM versions before 0.28.0 fail to validate audio sample rate headers in the transcription endpoint, allowing authenticated clients to bypass duration checks. Attackers can submit forged FLAC headers with inflated sample rates to trigger excessive memory allocation and crash the API server process affecting all tenants.7.1
- CVE-2026-90547WWBN AVideo through commit c3edcc274c389816d434acadac07ee78eaf330c1 fails to validate user permissions in the Bookmark plugin getBookmarks.json.php endpoint, allowing unauthenticated attackers to read chapter names from password-protected videos. Attackers can query the endpoint with a video ID parameter to retrieve sensitive chapter metadata without authentication or password verification.6.9
- CVE-2026-90548WWBN AVideo through commit c3edcc274c389816d434acadac07ee78eaf330c1 fails to validate user permissions in the ImageGallery list.json.php endpoint, allowing unauthenticated access to list gallery files. Attackers can retrieve filenames and URLs of password-protected image galleries by directly accessing the endpoint, then fetch the exposed files without authentication.6.9
- CVE-2026-90549WWBN AVideo through commit c3edcc274c389816d434acadac07ee78eaf330c1 fails to properly authorize access to the videosAndroid.json.php endpoint, allowing unauthenticated guests to list password-protected videos with sensitive owner information. Attackers can retrieve video metadata including owner email, lastLogin, filename, and hashId by sending an unauthenticated GET request to the endpoint.6.9
- CVE-2026-90550WWBN AVideo through commit c3edcc274c389816d434acadac07ee78eaf330c1 fails to check user authorization in the PlayerSkins mediaSession.json.php endpoint before returning video metadata. Unauthenticated attackers can request the endpoint with a video ID parameter to retrieve password-protected video titles and owner email addresses without authentication.6.9