Linux-PAM EUVD-2026-36662

Three specific conditions must all be true simultaneously for exploitation. First and most critically, the pam_userdb module must be configured with crypt=none, an unrecognized or misspelled crypt= value, or no crypt= argument at all - only these configurations enable the plaintext storage and comparison path; any recognized hash method (SHA512, md5, etc.) renders this path unreachable and the vulnerability unexploitable. Second, the attacker must be able to repeatedly submit authentication requests to a service that calls pam_userdb for credential verification - this requires either network reachability to a network-exposed service (SSH, FTP, custom daemon) or local execution on a system with a locally-bound service. Third, the attacker requires a sufficiently stable and low-jitter communication path to extract statistically meaningful timing deltas from response latency measurements across many attempts. What limits exploitation: the crypt=none configuration is non-default, explicitly insecure, and should not appear in hardened deployments; authentication rate limiting or lockout policies at the service layer directly impede the required volume of attempts; and the attack leaves a high-volume authentication log trail detectable by standard SIEM alerting.

The CVSS 4.0 score of 6.9 with vector AV:N/AC:H/AT:P/PR:N/UI:N captures the real-world risk profile accurately. AC:H reflects that timing attacks require precise measurement infrastructure, many authentication attempts, and a low-noise network path. AT:P (Attack Requirements: Present) explicitly encodes the non-default configuration prerequisite, limiting the exposed population to deployments where an administrator has configured crypt=none or equivalent. The supplemental AU:N (Automatable: No) and V:D (Value density: Diffuse) further confirm this is targeted rather than opportunistic - large-scale automated exploitation is not practical. However, VC:H (Vulnerable system Confidentiality: High) is well-justified because success yields full plaintext credential recovery. The E:P metric (Exploit maturity: Proof-of-Concept) elevates priority above theoretical vulnerabilities - methodology is demonstrably feasible. No KEV listing exists, distinguishing this from confirmed active exploitation. The description's framing of 'local or network-adjacent' as the attacker position contrasts somewhat with the provided AV:N assignment; defenders should verify whether their pam_userdb-consuming service is exposed to untrusted network segments, as this significantly changes actual attack surface. Missing data: no EPSS score was provided; no vendor advisory URL was included in the references beyond upstream source pointers.

An attacker with access to a network service backed by pam_userdb in crypt=none mode submits thousands of authentication attempts with systematically crafted candidate passwords - first varying length to identify the stored password's exact character count via timing differentials, then iterating each byte position with candidate ASCII values to recover the password character by character. Because proof-of-concept exploit methodology exists (CVSS 4.0 E:P), the statistical analysis approach to extracting timing signal is publicly documented. The high attack complexity (AC:H, AT:P) makes this a deliberate, targeted campaign requiring a stable low-latency path and extended observation time rather than an opportunistic automated attack.