{"dataType":"CVE_RECORD","dataVersion":"5.2","cveMetadata":{"cveId":"CVE-2026-99478","assignerOrgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","state":"PUBLISHED","assignerShortName":"Linux","dateReserved":"2026-09-25T10:25:14.417Z","datePublished":"2026-11-19T14:43:12.810Z","dateUpdated":"2026-11-22T11:08:34.874Z"},"containers":{"cna":{"providerMetadata":{"orgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","shortName":"Linux","dateUpdated":"2026-11-22T11:08:34.874Z"},"descriptions":[{"lang":"en","value":"In the Linux kernel, the following vulnerability has been resolved:\n\nfilelock: keep superblock active through delegation wait\n\nA caller may drop its path after try_break_deleg() returns a delegated\ninode but before break_deleg_wait() finishes. The extra inode reference\nheld in struct delegated_inode does not keep the superblock active.\n\nAfter a lazy unmount, closing the delegation holder can release the last\nmount reference and shut down the superblock while another task waits for\nthe delegation break. The waiter's final iput() can then enter the\nfilesystem's eviction code after its superblock-private state was freed.\nThis has been observed as a KASAN use-after-free in shmem_evict_inode().\n\nWhen try_break_deleg() saves a delegated inode, take an active reference\nto its superblock while the caller still pins the mount. Keep that\nreference through break_deleg_wait() and the final iput(), then release\nit with deactivate_super(). This protects every caller of the helpers,\nincluding the mknod, mkdir, symlink and rmdir paths that drop their path\nbefore waiting."}],"metrics":[{"cvssV3_1":{"version":"3.1","vectorString":"CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","baseScore":7.8,"baseSeverity":"HIGH"},"scenarios":[{"lang":"en","value":"AV:L - filename_mkdirat, filename_mknodat, filename_symlinkat and filename_rmdir call try_break_deleg on the parent directory, then end_creating_path or path_put, then break_deleg_wait. That is a local syscall path; nfsd and ksmbd pass a NULL delegated_inode and never reach break_deleg_wait, and no wire message carries the triggering state.\nAC:L - The attacker mounts their own tmpfs, takes a directory delegation with fcntl F_SETDELEG, blocks in break_deleg_wait via mkdirat, then umount2(MNT_DETACH) and close() of the lease fd from another thread they control. The iput in break_deleg_wait races shmem_put_super only against that close, so the interleaving is attacker-controlled.\nPR:L - vfs_setlease allows F_SETDELEG for the directory owner without CAP_LEASE, and mkdirat only needs write and execute on that directory. Mounting tmpfs and umount2(MNT_DETACH) need CAP_SYS_ADMIN in the mount namespace user_ns (may_mount); tmpfs sets FS_USERNS_MOUNT, so an unprivileged user gets that by unsharing a user and mount namespace.\nUI:N - The same attacker opens the directory, sets the delegation, issues mkdirat, mknodat, symlinkat or rmdir, lazy-unmounts the tmpfs, and closes the lease fd. No other user has to mount an image or perform a filesystem operation.\nS:U - break_deleg_wait's final iput corrupts the calling kernel in shmem_evict_inode. The bug does not cross a KVM/Xen guest-host boundary or an IOMMU/DMA boundary.\nC:H - After shmem_put_super frees the shmem_sb_info, shmem_evict_inode still loads it through SHMEM_SB. simple_xattrs_free then dereferences the freed xa_cache.ht via rhashtable_remove_fast when the directory has xattrs, and shmem_free_inode reads max_inodes from that slab object. Reclaiming it yields an arbitrary read.\nI:H - shmem_free_inode raw_spin_locks the freed sbinfo->stat_lock and adds into free_ispace, and simple_xattrs_free unlinks xattrs through the freed xa_cache. Both write attacker-reclaimable heap, which is enough to build a kernel write primitive from this use-after-free.\nA:H - The waiter's iput runs shmem_evict_inode on sbinfo that shmem_put_super already freed, which is the reported KASAN use-after-free. If the inode is still on sb->s_inodes, generic_shutdown_super instead poisons i_sb and the same iput faults. Either way the kernel oopses."}]}],"affected":[{"product":"Linux","vendor":"Linux","defaultStatus":"unaffected","repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["include/linux/filelock.h"],"versions":[{"version":"e12d203b8c880061c0bf0339cad51e5851a33442","lessThan":"a6e222f1d75bce8de430ba4d0f8c6a1e923f9acd","status":"affected","versionType":"git"},{"version":"e12d203b8c880061c0bf0339cad51e5851a33442","lessThan":"c206cdfee6c037b829178bb7e17e6e37ddc9aa35","status":"affected","versionType":"git"}]},{"product":"Linux","vendor":"Linux","defaultStatus":"affected","repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","programFiles":["include/linux/filelock.h"],"versions":[{"version":"6.19","status":"affected"},{"version":"0","lessThan":"6.19","status":"unaffected","versionType":"semver"},{"version":"7.3.2","lessThanOrEqual":"7.3.*","status":"unaffected","versionType":"semver"},{"version":"7.4-rc1","lessThanOrEqual":"*","status":"unaffected","versionType":"original_commit_for_fix"}]}],"cpeApplicability":[{"nodes":[{"operator":"OR","negate":false,"cpeMatch":[{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.19","versionEndExcluding":"7.3.2"},{"vulnerable":true,"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionStartIncluding":"6.19","versionEndExcluding":"7.4-rc1"}]}]}],"references":[{"url":"https://git.kernel.org/stable/c/a6e222f1d75bce8de430ba4d0f8c6a1e923f9acd"},{"url":"https://git.kernel.org/stable/c/c206cdfee6c037b829178bb7e17e6e37ddc9aa35"}],"title":"filelock: keep superblock active through delegation wait","x_generator":{"engine":"bippy-1.2.0"}}}}