ntfs_attr_find_in_attrdef() walks the in-memory $AttrDef table, but the
loop condition bounds only the start of each entry, not the whole entry:

	for (ad = vol->attrdef; (u8 *)ad - (u8 *)vol->attrdef <
			vol->attrdef_size && ad->type; ++ad)

struct attr_def is 160 bytes; the guard reads ad->type at offset 128 and
the loop body reads further fields. vol->attrdef is kvzalloc(i_size),
where i_size is the on-disk $AttrDef data size, checked in
load_and_init_attrdef() only as 0 < i_size <= 0x7fffffff. A volume whose
$AttrDef data size is smaller than one entry (e.g. 120 bytes) makes the
read of ad->type run past the allocation. Creating a file reaches this
through ntfs_attr_size_bounds_check() and reads out of bounds:

  BUG: KASAN: slab-out-of-bounds in ntfs_attr_find_in_attrdef+0x66/0xa0
  Read of size 4 at addr ffff888005833280 by task init/1
   ntfs_attr_find_in_attrdef
   ntfs_attr_size_bounds_check
   ntfs_attr_can_be_non_resident
   ntfs_attr_add

Require the whole entry to lie within attrdef_size in the loop guard, and
reject at mount a $AttrDef too small to hold one attr_def entry.

Fixes: 1e9ea7e04472 ("Revert "fs: Remove NTFS classic"")
Assisted-by: Claude:claude-opus-4-8
Signed-off-by: Dennis Tighe <dennis.tighe@gmail.com>
---
Changes in v2:
 - Rewrite the loop guard to bound the entry offset without forming a
   pointer past the end of the buffer. Good call by Namjae Jeon.

 fs/ntfs/attrib.c | 4 ++--
 fs/ntfs/super.c  | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/fs/ntfs/attrib.c b/fs/ntfs/attrib.c
index 990c3937a..7d0816de7 100644
--- a/fs/ntfs/attrib.c
+++ b/fs/ntfs/attrib.c
@@ -1737,8 +1737,8 @@ static struct attr_def *ntfs_attr_find_in_attrdef(const struct ntfs_volume *vol,
 	struct attr_def *ad;
 
 	WARN_ON(!type);
-	for (ad = vol->attrdef; (u8 *)ad - (u8 *)vol->attrdef <
-			vol->attrdef_size && ad->type; ++ad) {
+	for (ad = vol->attrdef; (u8 *)ad - (u8 *)vol->attrdef <=
+	     vol->attrdef_size - (s32)sizeof(*ad) && ad->type; ++ad) {
 		/* We have not found it yet, carry on searching. */
 		if (likely(le32_to_cpu(ad->type) < le32_to_cpu(type)))
 			continue;
diff --git a/fs/ntfs/super.c b/fs/ntfs/super.c
index cd8fa2c13..2fd7db672 100644
--- a/fs/ntfs/super.c
+++ b/fs/ntfs/super.c
@@ -1241,9 +1241,9 @@ static bool load_and_init_attrdef(struct ntfs_volume *vol)
 		goto failed;
 	}
 	NInoSetSparseDisabled(NTFS_I(ino));
-	/* The size of FILE_AttrDef must be above 0 and fit inside 31 bits. */
+	/* FILE_AttrDef must hold at least one entry and fit inside 31 bits. */
 	i_size = i_size_read(ino);
-	if (i_size <= 0 || i_size > 0x7fffffff)
+	if (i_size < (s64)sizeof(struct attr_def) || i_size > 0x7fffffff)
 		goto iput_failed;
 	vol->attrdef = kvzalloc(i_size, GFP_NOFS);
 	if (!vol->attrdef)

On Wed, Aug 26, 2026 at 1:44 PM Dennis Tighe <dennis.tighe@gmail.com> wrote:
>
> ntfs_attr_find_in_attrdef() walks the in-memory $AttrDef table, but the
> loop condition bounds only the start of each entry, not the whole entry:
>
>         for (ad = vol->attrdef; (u8 *)ad - (u8 *)vol->attrdef <
>                         vol->attrdef_size && ad->type; ++ad)
>
> struct attr_def is 160 bytes; the guard reads ad->type at offset 128 and
> the loop body reads further fields. vol->attrdef is kvzalloc(i_size),
> where i_size is the on-disk $AttrDef data size, checked in
> load_and_init_attrdef() only as 0 < i_size <= 0x7fffffff. A volume whose
> $AttrDef data size is smaller than one entry (e.g. 120 bytes) makes the
> read of ad->type run past the allocation. Creating a file reaches this
> through ntfs_attr_size_bounds_check() and reads out of bounds:
>
>   BUG: KASAN: slab-out-of-bounds in ntfs_attr_find_in_attrdef+0x66/0xa0
>   Read of size 4 at addr ffff888005833280 by task init/1
>    ntfs_attr_find_in_attrdef
>    ntfs_attr_size_bounds_check
>    ntfs_attr_can_be_non_resident
>    ntfs_attr_add
>
> Require the whole entry to lie within attrdef_size in the loop guard, and
> reject at mount a $AttrDef too small to hold one attr_def entry.
>
> Fixes: 1e9ea7e04472 ("Revert "fs: Remove NTFS classic"")
> Assisted-by: Claude:claude-opus-4-8
> Signed-off-by: Dennis Tighe <dennis.tighe@gmail.com>
Applied it to #ntfs-next.
Thanks!