[PATCH 07/18] Squashfs: fix fragment bytes overflow on 32-bit architectures

From: Phillip Lougher

Date: Sat Nov 07 2026 - 00:37:00 EST


In helper function squashfs_fragment_bytes() the u32 fragments
variable is multiplied by sizeof(struct squashfs_fragment_enty).

On a 32-bit architecture the type of sizeof() is a 32-bit unsigned
integer, which means the computation will overflow for fragments
values above 268435455.

The knock-on effect is the size of the in memory fragment index
table will be incorrectly computed, and a much smaller than expected
index table will be allocated and read into memory.

Later when a fragment index beyond this is looked up in this table,
this will produce an out of bounds access.

The fix is to explicitly widen the fragments variable to a long long
in the squashfs_fragment_bytes() helper function.

Reproducer
=========

The following script when run on a 32-bit x86 kernel (which lacks
KASAN support) will produce an I/O error. On a 32-bit ARM kernel
with KASAN enabled, the out of bounds access will also produce a
KASAN report.

[ 193.817260] ==================================================================
[ 193.818057] BUG: KASAN: slab-out-of-bounds in squashfs_frag_lookup+0x23c/0x264
[ 193.820408] Read of size 8 at addr cc126708 by task stat/357
[ 193.820720]
[ 193.821675] CPU: 3 UID: 0 PID: 357 Comm: stat Not tainted 7.2.0-13187-g66498c75b4f8 #113 VOLUNTARY

=================================================================
#!/bin/bash

ARGS="-no-duplicates -no-sparse -no-hardlinks -quiet -no-progress"

finish() {
popd 2>&1 > /dev/null
rm -rf $DIR
exit 1
}

# Create a Squashfs filesystem with 1026 files and 513 fragments.
# This will create a two entry Squashfs fragment index table.
#
DIR=$(mktemp -d)

pushd $DIR 2>&1 > /dev/null

# Create a Squashfs pseudo file which is fed to Mksquashfs
echo Create a Squashfs pseudo file which is fed to Mksquashfs

dd if=/dev/urandom of=urandom bs=1024 count=50 status=none
for i in $(seq 1 1026); do
echo "$i l urandom" >> pseudo
done

# Create the Squashfs filesystem
echo Create the Squashfs filesystem

if ! mksquashfs - img.sqsh -pf pseudo -p "/ d 0777 0 0" $ARGS; then
echo Failed to create Squashfs filesystem
finish

fi

# The fragments field (fragment count) in the superblock is at offset
# 16 bytes.
#
# Over-write the fragments count with 268,435,457 little-endian. On a
# 32-bit architecture, this when multiplied by 16 will overflow, giving
# the value 16.
#
# Ultimately this will cause the kernel code to create and read a single
# entry index table (where in fact the table has two entries).
#
# Then stating file 999 which has a fragment id in the second index,
# will cause an out of bounds access.

# Write 268,435,457 in little-endian format to file
/usr/bin/echo -en "\x01\x00\x00\x10" > file

# Over-write the fragments count in the superblock
echo Over-write the fragments count in the superblock

dd bs=1 seek=16 if=file of=img.sqsh count=4 conv=notrunc status=none

# Mount the filesystem to /mnt
echo Mount the filesystem to /mnt

if ! mount -t squashfs img.sqsh /mnt; then
echo Mount of Squashfs filesystem failed
finish
fi

# Stat file 999
echo Stat file 999

stat /mnt/999

popd 2>&1 > /dev/null

rm -rf $DIR
=================================================================

Fixes: ffae2cd73a9e ("Squashfs: header files")
Signed-off-by: Phillip Lougher <phillip@xxxxxxxxxxxxxxx>
---
fs/squashfs/squashfs_helpers.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/squashfs/squashfs_helpers.h b/fs/squashfs/squashfs_helpers.h
index 4088e598400f..524d4a77686d 100644
--- a/fs/squashfs/squashfs_helpers.h
+++ b/fs/squashfs/squashfs_helpers.h
@@ -55,7 +55,7 @@ static inline s64 squashfs_mkinode(u32 start, u16 offset)
/* fragment and fragment table defines */
static inline s64 squashfs_fragment_bytes(u32 fragments)
{
- return fragments * sizeof(struct squashfs_fragment_entry);
+ return (s64) fragments * sizeof(struct squashfs_fragment_entry);
}

static inline s32 squashfs_fragment_index(u32 fragments)
--
2.47.3