Skip to contentSkip to conversation

ntfs: reject invalid sectors_per_cluster in the boot sector

PATCH Applied Dennis Tighe 2 participants 2 messages Aug 26

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 ntfs: reject invalid sectors_per_cluster in the boot sector

is_boot_sector_ntfs() checks the boot sector's sectors_per_cluster field with a range test that rejects 0x81..0xf3 but accepts 0 and other non-power-of-two counts. A zero value reaches parse_ntfs_boot_sector():

	sectors_per_cluster_bits = ffs(sectors_per_cluster) - 1;
	...
	vol->cluster_size = vol->sector_size << sectors_per_cluster_bits;

ffs(0) is 0, so sectors_per_cluster_bits becomes (unsigned)-1 and the shift is undefined:

  UBSAN: shift-out-of-bounds in fs/ntfs/super.c:673:39
  shift exponent 4294967295 is too large for 32-bit type 'int'

This change rejects any non-power-of-two value, since it feeds the aforementioned shift via ffs() - 1, which only yields the correct shift for a power of two.

Fixes 6251f0b0de7d ("ntfs: update super block operations")
Assisted-by Claude:claude-opus-4-8

---
Changes in v2:
- Check that sectors_per_cluster is a power of two rather than only rejecting zero, per Hyunchul Lee's review.

is_boot_sector_ntfs() is where the driver decides an image is NTFS, so rejecting a bad geometry there stops it before parse_ntfs_boot_sector() computes ffs() - 1. Reached by mounting the image.

Built ntfs-next with KASAN+UBSAN and tested: sectors_per_cluster = 0 and a non-power-of-two value (e.g. 3) are both rejected with no UBSAN, and a valid volume still mounts (create/write/mkdir/rename/unlink smoke passes).

A reproducer is available on request.

 fs/ntfs/super.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
fs/ntfs/super.c +2 -2
index 2fd7db672..76f62dac2 100644
@@ -557,8 +557,8 @@ static bool is_boot_sector_ntfs(const struct super_block *sb,
 	 * Check sectors per cluster value is valid and the cluster size
 	 * is not above the maximum (2MB).
 	 */
-	if (b->bpb.sectors_per_cluster > 0x80 &&
-	    b->bpb.sectors_per_cluster < 0xf4)
+	if (b->bpb.sectors_per_cluster < 0xf4 &&
+	    !is_power_of_2(b->bpb.sectors_per_cluster))
 		goto not_ntfs;
 
 	/* Check reserved/unused fields are really zero. */
maintainer Applied
#1 lore raw
Quoted 15 lines · Dennis Tighe

is_boot_sector_ntfs() checks the boot sector's sectors_per_cluster field with a range test that rejects 0x81..0xf3 but accepts 0 and other non-power-of-two counts. A zero value reaches parse_ntfs_boot_sector():

        sectors_per_cluster_bits = ffs(sectors_per_cluster) - 1;
        ...
        vol->cluster_size = vol->sector_size << sectors_per_cluster_bits;

ffs(0) is 0, so sectors_per_cluster_bits becomes (unsigned)-1 and the shift is undefined:

  UBSAN: shift-out-of-bounds in fs/ntfs/super.c:673:39
  shift exponent 4294967295 is too large for 32-bit type 'int'
Quoted 7 lines

This change rejects any non-power-of-two value, since it feeds the aforementioned shift via ffs() - 1, which only yields the correct shift for a power of two.

Fixes 6251f0b0de7d ("ntfs: update super block operations")
Assisted-by Claude:claude-opus-4-8

Applied it to #ntfs-next.
Thanks!