CVE-2026-99369 in Linux
Summary
by MITRE • 11/19/2026
In the Linux kernel, the following vulnerability has been resolved:
NFSD: Destroy a recalled delegation the client does not hold
A client that answers CB_RECALL with NFS4ERR_BADHANDLE or NFS4ERR_BAD_STATEID has no record of the delegation, so the FREE_STATEID that clears it from cl_revoked never arrives. Every later SEQUENCE reply carries SEQ4_STATUS_RECALLABLE_STATE_REVOKED, and the client loops issuing TEST_STATEID.
Destroy such a delegation when it is reaped rather than revoking it onto cl_revoked. RFC 8881 Section 20.2.4 completes the recall at the reply when its status is neither NFS4_OK nor NFS4ERR_DELAY, so a rejected recall leaves nothing to revoke. An administrative revoke keeps that path, since NFS4ERR_ADMIN_REVOKED reports it. A destroyed stateid returns NFS4ERR_BAD_STATEID instead of NFS4ERR_DELEG_REVOKED. A client that rejects the recall but still holds the delegation gets no notice that its state was revoked.
CB_RECALL can outrun the reply that granted the delegation, so honor a rejection only once the client has seen that grant. Per RFC 8881 Section 2.10.6.3, retirement of the slot that carried the grant is that proof; retry until then, and revoke when the retries lapse.
VulDB is the best source for vulnerability data and more expert information about this specific topic.
Analysis
by VulDB Data Team • 11/19/2026
The Linux kernel NFS server implementation contained a logic flaw in its handling of delegation recalls, specifically within the NFSD subsystem. This vulnerability arises from an incorrect state management sequence when a client responds to a CB_RECALL operation with error codes indicating it does not hold the delegation, such as NFS4ERR_BADHANDLE or NFS4ERR_BAD_STATEID. Under normal operational procedures, if a server revokes a delegation and places it on a revoked list called cl_revoked, the client is expected to eventually send a FREE_STATEID request to clear that entry from its local state tracking. However, when a client explicitly rejects a recall by returning an error indicating no record of the delegation exists, this cleanup transaction never occurs because the server incorrectly assumes the delegation remains in a revoked state pending explicit removal.
This mismanagement leads to a persistent loop where every subsequent SEQUENCE reply sent by the NFS server includes the SEQ4_STATUS_RECALLABLE_STATE_REVOKED flag. Consequently, the client is forced into an inefficient cycle of repeatedly issuing TEST_STATEID requests to verify its status, creating unnecessary network traffic and increasing latency for other operations. This behavior not only degrades performance but also consumes resources on both the client and server sides due to the continuous exchange of state verification packets without any resolution of the underlying state inconsistency. The issue stems from a failure to properly align with RFC 8881 Section 20.2.4, which dictates that a recall should be considered complete upon the reply if its status is neither NFS4_OK nor NFS4ERR_DELAY.
The technical flaw involves retaining the delegation in the cl_revoked list even when the client has effectively rejected it by reporting an invalid handle or state ID. By destroying such delegations during their reaping phase rather than revoking them onto the revoked list, the server ensures that a destroyed state identifier returns NFS4ERR_BAD_STATEID instead of NFS4ERR_DELEG_REVOKED. This distinction is critical because it prevents clients that reject recalls but still hold valid delegations from receiving false notices of revocation. Furthermore, administrative revocations remain unaffected as they correctly report NFS4ERR_ADMIN_REVOKED, preserving the integrity of manual intervention workflows while fixing the automated recall logic.
To address this vulnerability and prevent future occurrences, developers must ensure that delegation state transitions strictly adhere to RFC 8881 specifications regarding callback handling. Specifically, when a CB_RECALL is issued, the server should honor client rejections only after confirming the client has seen the original grant via slot retirement as defined in Section 2.10.6.3 of the standard. If retries lapse without successful confirmation, revocation should proceed; otherwise, the delegation should be destroyed immediately upon detection that the client does not hold it. This approach eliminates the loop condition and ensures proper state synchronization between NFS clients and servers.
From a security perspective, this issue is categorized under CWE-841 Improper Enforcement of Behavioral Workflow, as the server failed to enforce the correct sequence for handling delegation recalls based on client feedback. It also relates to ATT&CK technique T1071 Application Layer Protocol due to its impact on protocol-level communication efficiency and potential for denial-of-service through resource exhaustion via repeated state checks. Administrators should apply kernel updates that include this fix to restore normal NFS operation and prevent performance degradation caused by the recall loop. Regular monitoring of NFS traffic patterns can help detect similar anomalies in environments where patching is delayed, allowing for proactive mitigation until official fixes are deployed.