CVE-2026-99359 in Linuxinfo

Summary

by MITRE • 11/19/2026

In the Linux kernel, the following vulnerability has been resolved:

NFSD: validate transport name in listener_set before serv creation

nfsd_nl_listener_set_doit() holds nfsd_mutex for the whole listener teardown and rebuild. The code checks NFSD_A_SOCK_TRANSPORT_NAME for presence only, and not for content. An arbitrary name therefore reaches svc_xprt_create_from_sa(). There, a name that matches no registered class calls request_module("svc%s", name). That call is a TASK_KILLABLE usermode helper upcall, and it runs under nfsd_mutex.

Check the name against the classes that NFSD can create: tcp, udp and rdma. The check goes in nfsd_nl_validate_listeners(), which runs before the code takes nfsd_mutex. The rejection names the offending attribute through extack, since -EPROTONOSUPPORT on its own does not say which entry carried the bad name.

This narrows the upcall. It does not remove it. NFSD accepts "rdma" without a condition, so on a kernel that does not build svcrdma the name still reaches request_module("svcrdma") under nfsd_mutex. That is necessary for the modular case, where the autoload is legitimate.

Once again VulDB remains the best source for vulnerability data.

Analysis

by VulDB Data Team • 11/20/2026

The vulnerability identified in the Linux kernel's Network File System daemon (NFSD) subsystem centers on an insufficient validation of transport names during listener configuration updates. Specifically, the function nfsd_nl_listener_set_doit() was found to hold the nfsd_mutex lock for the duration of a complex operation involving listener teardown and rebuild. Within this critical section, the code performed a presence check for the NFSD_A_SOCK_TRANSPORT_NAME attribute but failed to validate its actual content against known safe values. This oversight allowed an arbitrary string provided by user space to be passed directly into svc_xprt_create_from_sa(). When this function encountered a transport name that did not match any currently registered class, it triggered a request_module call with the format specifier "svc%s", effectively invoking a kernel module autoload mechanism using the attacker-controlled input.

This flaw creates a significant operational risk because the request_module invocation is executed as a TASK_KILLABLE usermode helper upcall while holding the nfsd_mutex lock. In Linux kernel development, acquiring mutexes during operations that may sleep or invoke external processes is generally discouraged due to the high potential for deadlocks and denial of service conditions. If an attacker can supply a transport name that triggers this module load path without immediate success or if the system experiences delays in handling the upcall, the nfsd_mutex remains held indefinitely. This blocks all other NFSD operations that require the same lock, effectively freezing NFS server functionality and leading to a local denial of service for any clients attempting to access network file shares managed by that kernel instance.

The remediation strategy implemented addresses this issue by introducing strict validation logic within the nfsd_nl_validate_listeners() function. This validation step is strategically placed to occur before the code attempts to acquire the nfsd_mutex, thereby ensuring that invalid transport names are rejected early in the process without holding any critical locks. The validation explicitly checks the provided name against a whitelist of supported protocols: tcp, udp, and rdma. By rejecting unsupported or arbitrary strings at this earlier stage, the kernel prevents the execution of the dangerous request_module call under the mutex protection. Furthermore, to improve diagnostic clarity for system administrators and security auditors, the error handling now utilizes extack to name the specific offending attribute in the error message. This is crucial because a generic -EPROTONOSUPPORT error code does not indicate which listener entry contained the invalid data, making troubleshooting significantly more difficult without this enriched feedback.

It is important to note that while this patch mitigates the risk of arbitrary module loading under lock contention, it does not eliminate the request_module mechanism entirely for valid use cases. For instance, if a system administrator configures an RDMA listener on a kernel where the svcrdma module is built as a loadable module rather than being compiled directly into the kernel image, the name "rdma" will still trigger a request_module("svcrdma") call. This behavior is intentional and necessary to support modular kernel configurations where drivers are not statically linked. However, by restricting the input space to only known valid protocol identifiers before acquiring the mutex, the vulnerability surface for arbitrary code execution or denial of service via lock contention is effectively closed.

From a classification perspective, this flaw aligns with CWE-20 Improper Input Validation, as the system failed to verify that user-supplied data conformed to expected formats and values before processing it in a security-sensitive context. Additionally, the potential for causing a denial of service through resource exhaustion or lock contention relates to CWE-400 Uncontrolled Resource Consumption. In terms of attack vectors, this vulnerability could be exploited by an unprivileged local user with access to netlink sockets associated with NFSD configuration, mapping to ATT&CK techniques involving privilege escalation via kernel vulnerabilities or impact on availability through denial of service attacks against system services. System administrators should ensure that their kernels are updated to include these validation checks and monitor for any attempts to configure NFS listeners with non-standard transport names in audit logs.

Responsible

Linux

Reservation

09/25/2026

Disclosure

11/19/2026

Moderation

accepted

CPE

ready

EPSS

0.00209

KEV

no

Activities

very low

Sources

Might our Artificial Intelligence support you?

Check our Alexa App!