Description

In the Linux kernel, the following vulnerability has been resolved: NFSD: check truncate permission under inode lock nfsd_setattr() checks whether a size update needs NFSD_MAY_TRUNC before it takes inode_lock(). The comparison uses the file size sampled by that unlocked read, but the actual ATTR_SIZE update is applied later under inode_lock() by notify_change(). This leaves a TOCTOU window for append-only files. If a client sends a SETATTR that does not shrink the file at the time of the unlocked sample, a concurrent append can extend the file before nfsd_setattr() takes inode_lock(). notify_change() then applies a real truncation without the NFSD_MAY_TRUNC check that rejects IS_APPEND(inode). The VFS truncate syscall paths perform their own append-only checks before calling notify_change(), so NFSD must make this decision against the locked size it is about to change. Split the write-count acquisition from the truncation permission check. Keep get_write_access() before the locked setattr work, then recheck whether the requested size is below i_size_read(inode) after inode_lock() has been acquired and before notify_change(ATTR_SIZE). This also avoids the plain unlocked inode->i_size load.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "9.1",
  "pubDate": "2026-09-11T20:19:58.743Z",
  "pubdate": "2026-09-11T20:19:58.743Z",
  "executiveSummary": "A Time-of-Check to Time-of-Use (TOCTOU) vulnerability exists in the Linux kernel's NFSD component during the processing of SETATTR operations.\nThe vulnerability occurs because the NFSD_MAY_TRUNC permission check is performed using an unlocked file size value before acquiring the inode lock.\nAn attacker can exploit this race condition to bypass append-only file restrictions by extending the file after the initial permission check but before the inode is locked.\nThis flaw allows unauthorized truncation of append-only files, potentially compromising data integrity and security policies enforced by the filesystem.\nThe vulnerability affects systems utilizing NFSD to handle file attributes; successful exploitation requires a local or remote user with the ability to issue SETATTR requests to the affected NFS export.\nThe risk is categorized as a security bypass, as the kernel fails to properly enforce mandatory access controls regarding file truncation on constrained files.",
  "technicalDetails": "The root cause of this vulnerability lies in an improper sequence of operations within nfsd_setattr(). The function historically performed a check for NFSD_MAY_TRUNC based on a cached or unlocked sample of the file size (i_size) retrieved before the inode_lock() was acquired.\nBecause the final size change is applied by notify_change() only after the inode lock is taken, a TOCTOU window is opened between the check and the actual mutation.\nAn attacker can exploit this window by monitoring or timing a SETATTR operation. Specifically, if a client sends a SETATTR request that does not trigger a truncation (size >= current size) at the time of the initial check, the check passes. If the attacker then performs a concurrent append operation to extend the file before nfsd_setattr() reaches the notify_change() call, the actual size applied by the kernel will result in a truncation relative to the new, extended size.\nThis circumvents the IS_APPEND(inode) protection check, which is designed to prevent modifications to files flagged with the append-only attribute. In standard VFS syscall paths, append-only checks are performed in conjunction with lock acquisition, ensuring the state cannot change between validation and execution.\nBy failing to re-verify the necessity of the truncation check after acquiring the inode lock, NFSD allows the notify_change() function to proceed with an operation that would otherwise be rejected by the VFS append-only constraint.\nThe patch resolves this by decoupling write-count acquisition from the permission logic. The NFSD_MAY_TRUNC decision is deferred until after the inode_lock() has been acquired, ensuring that the size comparison is performed against a stable, locked representation of the inode's current size. This eliminates the race condition by ensuring the check and the mutation are atomic relative to the state of the inode."
}