containers/storage CVE-2026-114236

Exploitation requires that the application invokes the containers/storage archive functions ApplyLayer or ApplyUncompressedLayer on attacker-influenced input - i.e., unpacking corrupt or malicious container image layers/tar streams - and that this processing can be driven repeatedly (for example via an image pull, build, or sync loop) until the process file-descriptor limit is exhausted. Limiting factors: the CVSS vector is local (AV:L) and requires user/operator interaction (UI:R), so the attacker must get a victim process to actually process the malicious layers; impact is availability-only (C:N/I:N/A:L) and is confined to the affected process (S:U), with no code execution or data exposure. A single malformed input does not immediately deny service - sustained repeated processing is needed - which further bounds real-world severity.

This is a genuinely low-priority issue despite affecting a widely-embedded component. The CVSS 3.1 base score is 3.3 (Low) with vector AV:L/AC:L/PR:N/UI:R/S:U/C:N/I:N/A:L - the only impact is a partial availability (A:L) hit from file-descriptor exhaustion (CWE-772), with no confidentiality or integrity consequence. Exploitation requires local vector (AV:L) and user interaction (UI:R), meaning an operator or process must be induced to repeatedly process attacker-supplied corrupt/malicious layer archives before the fd limit is reached; this is a gradual degradation, not an immediate crash. No EPSS score, CISA KEV listing, POC, or SSVC decision data was provided in the input. There is no public exploit identified at time of analysis and no evidence of active exploitation. The signals are internally consistent (low score, local vector, availability-only, resource-exhaustion CWE), so there is no conflict to flag - the main caveat is that A:L likely understates nothing and the realistic outcome is a denial of service confined to the affected process. This is a hardening/defense-in-depth fix, not an emergency; prioritize via normal patch cycles.

An attacker crafts a corrupt or malicious container image layer and causes a victim system to pull and unpack it (e.g., repeatedly via an automated build or image-sync pipeline). Each call to ApplyLayer/ApplyUncompressedLayer on the bad input leaks a file descriptor, and sustained repeated processing eventually exhausts the process's fd limit, degrading or denying service. No public exploit code is identified at time of analysis, and the local vector plus required user/operator interaction make this a slow-burn DoS rather than a one-shot remote crash.