In the Linux kernel, the following vulnerability has been resolved:
ntfs: bound the free-cluster bitmap scan to the volume
vol->lcn_empty_bits_per_page is sized from vol->nr_clusters at mount, but
ntfs_cluster_alloc() bounds its scan of that array by the size of $Bitmap.
Those are independent on-disk quantities and the mount-time check only
rejects a $Bitmap that is too small, so an image whose $Bitmap covers more
clusters than the volume has lets the scan index past the array. A run
whose LCN lies in that gap takes the allocator straight there, since the
caller passes the file's own last LCN as its locality hint. KASAN reports
a slab out-of-bounds read when a file on such a volume is extended.
Clamp the scan to what that array covers, mirroring the max_index
calculation the mount-time scan already uses, and reject a decoded LCN
at or beyond nr_clusters in the mapping pairs decoder. Conforming
volumes are unaffected.
CVSS Vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
CVSS Score: 9.8
AV:N - fs/ntfs registers ntfs_export_ops, so nfsd/ksmbd VFS writes on an exported NTFS volume reach ntfs_file_write_iter->iomap->ntfs_attr_expand->ntfs_cluster_alloc, and reads reach ntfs_mapping_pairs_decompress, without local shell access.
AC:L - A crafted image whose $Bitmap covers more clusters than nr_clusters, plus a file run whose LCN sits in that gap, deterministically drives the allocator to index past lcn_empty_bits_per_page and the decoder to accept out-of-volume LCNs, with no race or victim-specific layout.
PR:N - Triggering needs only permission to cause server-side parse or extend of mounted NTFS metadata (guest/anonymous SMB, a permissive NFS export, or an unprivileged local open/write on an already-mounted volume), not init-namespace root or CAP_SYS_ADMIN at trigger time.
UI:N - After the crafted volume is mounted and optionally exported, ordinary remote or local write/truncate/fallocate, and reads that decompress mapping pairs, trigger the bug; no extra victim mount dialog, USB insertion, or confirmation is required during the attack.
S:U - Slab out-of-bounds access to vol->lcn_empty_bits_per_page and misdirected block I/O remain in the host kernel/OS security domain and do not cross a VM, container, or IOMMU boundary to another authority.
C:H - ntfs_cluster_alloc performs a KASAN-confirmed slab out-of-bounds read of lcn_empty_bits_per_page at an attacker-influenced page index, and decoded LCNs past nr_clusters make iomap reads disclose data beyond the NTFS volume, a kernel heap/OOB-read disclosure primitive.
I:H - With zone_end set from the poisoned LCN, the bitmap scan calls ntfs_set_lcn_empty_bits() on out-of-range indices (heap OOB write of empty-bit counts) and iomap can write past the volume using those LCNs, a kernel memory-corruption primitive consistent with privilege escalation.
A:H - KASAN reports a slab out-of-bounds access when a file on such a volume is extended; indexing past the kvmalloc array can oops, and the OOB write of lcn_empty_bits_per_page can panic or hang the host, so even failed exploitation denies service.
| Attack Vector |
Network |
Scope |
Unchanged |
| Attack Complexity |
Low |
Confidentiality Impact |
High |
| Privileges Required |
None |
Integrity Impact |
High |
| User Interaction |
None |
Availability Impact |
High |
AV:N - fs/ntfs registers ntfs_export_ops, so nfsd/ksmbd VFS writes on an exported NTFS volume reach ntfs_file_write_iter->iomap->ntfs_attr_expand->ntfs_cluster_alloc, and reads reach ntfs_mapping_pairs_decompress, without local shell access.
AC:L - A crafted image whose $Bitmap covers more clusters than nr_clusters, plus a file run whose LCN sits in that gap, deterministically drives the allocator to index past lcn_empty_bits_per_page and the decoder to accept out-of-volume LCNs, with no race or victim-specific layout.
PR:N - Triggering needs only permission to cause server-side parse or extend of mounted NTFS metadata (guest/anonymous SMB, a permissive NFS export, or an unprivileged local open/write on an already-mounted volume), not init-namespace root or CAP_SYS_ADMIN at trigger time.
UI:N - After the crafted volume is mounted and optionally exported, ordinary remote or local write/truncate/fallocate, and reads that decompress mapping pairs, trigger the bug; no extra victim mount dialog, USB insertion, or confirmation is required during the attack.
S:U - Slab out-of-bounds access to vol->lcn_empty_bits_per_page and misdirected block I/O remain in the host kernel/OS security domain and do not cross a VM, container, or IOMMU boundary to another authority.
C:H - ntfs_cluster_alloc performs a KASAN-confirmed slab out-of-bounds read of lcn_empty_bits_per_page at an attacker-influenced page index, and decoded LCNs past nr_clusters make iomap reads disclose data beyond the NTFS volume, a kernel heap/OOB-read disclosure primitive.
I:H - With zone_end set from the poisoned LCN, the bitmap scan calls ntfs_set_lcn_empty_bits() on out-of-range indices (heap OOB write of empty-bit counts) and iomap can write past the volume using those LCNs, a kernel memory-corruption primitive consistent with privilege escalation.
A:H - KASAN reports a slab out-of-bounds access when a file on such a volume is extended; indexing past the kvmalloc array can oops, and the OOB write of lcn_empty_bits_per_page can panic or hang the host, so even failed exploitation denies service.
CVSS 3.1