Re: [PATCH] loop: don't change loop device under exclusive opener in loop_set_status
From: Yongpeng Yang
Date: Tue Dec 02 2025 - 04:03:41 EST
On 12/1/25 20:38, Jan Kara wrote:
On Tue 18-11-25 15:10:20, Yongpeng Yang wrote:
On 11/14/25 22:42, Raphael Pinsonneault-Thibeault wrote:
loop_set_status() is allowed to change the loop device while there
are other openers of the device, even exclusive ones.
In this case, it causes a KASAN: slab-out-of-bounds Read in
ext4_search_dir(), since when looking for an entry in an inlined
directory, e_value_offs is changed underneath the filesystem by
loop_set_status().
Fix the problem by forbidding loop_set_status() from modifying the loop
device while there are exclusive openers of the device. This is similar
to the fix in loop_configure() by commit 33ec3e53e7b1 ("loop: Don't
change loop device under exclusive opener") alongside commit ecbe6bc0003b
("block: use bd_prepare_to_claim directly in the loop driver").
Reported-by: syzbot+3ee481e21fd75e14c397@xxxxxxxxxxxxxxxxxxxxxxxxx
Closes: https://syzkaller.appspot.com/bug?extid=3ee481e21fd75e14c397
Tested-by: syzbot+3ee481e21fd75e14c397@xxxxxxxxxxxxxxxxxxxxxxxxx
Signed-off-by: Raphael Pinsonneault-Thibeault <rpthibeault@xxxxxxxxx>
---
ML thread for previous, misguided patch idea:
https://lore.kernel.org/all/20251112185712.2031993-2-rpthibeault@xxxxxxxxx/t/
drivers/block/loop.c | 41 ++++++++++++++++++++++++++++++-----------
1 file changed, 30 insertions(+), 11 deletions(-)
diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index 053a086d547e..756ee682e767 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -1222,13 +1222,24 @@ static int loop_clr_fd(struct loop_device *lo)
}
static int
-loop_set_status(struct loop_device *lo, const struct loop_info64 *info)
+loop_set_status(struct loop_device *lo, blk_mode_t mode,
+ struct block_device *bdev, const struct loop_info64 *info)
{
int err;
bool partscan = false;
bool size_changed = false;
unsigned int memflags;
+ /*
+ * If we don't hold exclusive handle for the device, upgrade to it
+ * here to avoid changing device under exclusive owner.
+ */
+ if (!(mode & BLK_OPEN_EXCL)) {
+ err = bd_prepare_to_claim(bdev, loop_set_status, NULL);
+ if (err)
+ goto out_reread_partitions;
+ }
+
+ if (mode & BLK_OPEN_EXCL) {
+ struct block_device *whole = bdev_whole(bdev);
+
+ BUG_ON(whole->bd_claiming == NULL);
+ }
I add the above code and do the following test:
# losetup -f data.1g
# echo "0 `blockdev --getsz /dev/loop0` linear /dev/loop0 0" | dmsetup
create my-linear
# ./ioctl-test /dev/mapper/my-linear // trigger BUG_ON, ioctl-test.c is
in attachment.
The root causes of BUG_ON:
1. When creating 'my-linear' device, the mode for opening /dev/loop0
does not include the BLK_OPEN_EXCL flag.
table_load
- dm_table_create // get_mode() never assign BLK_OPEN_EXCL to {struct
dm_table *t}->mode
- populate_table
- dm_table_add_target
- linear_ctr
- dm_get_device // mode = {struct dm_table *t}->mode, never open
loop0 with BLK_OPEN_EXCL mode.
BLK_OPEN_EXCL is added by bdev_open() whenever it is called with non-NULL
holder. And DM code (open_table_device()) calls bdev_file_open_by_dev() with
_dm_claim_ptr as the holder. So all opens from DM should be exclusive ones.
The question obviously is what is broken in this that your reproducer still
works...
Honza
Yes, I was mistaken. When loop0's whole->bd_claiming is NULL,
bdev->bd_holder is set to _dm_claim_ptr by bd_finish_claiming. When the
my-linear device is opened with BLK_OPEN_EXCL, its ioctls are handled
normally. If the my-linear device is opened without BLK_OPEN_EXCL, the
ioctl call returns -EBUSY, which is the expected behavior.
bdev_open
- bd_prepare_to_claim //whole->bd_claiming = _dm_claim_ptr;
- bd_finish_claiming // whole->bd_holder = bd_may_claim;
bdev->bd_holder = _dm_claim_ptr;
- bd_clear_claiming // whole->bd_claiming = NULL;
Tested-by: Yongpeng Yang <yangyongpeng@xxxxxxxxxx>
Yongpeng,
2. When 'my-linear' device is opened with the O_EXCL flag, and an ioctl
is issued to it. The dm_blk_ioctl function calls bdev->bd_disk->fops-
ioctl(bdev, mode, cmd, arg), which passes the mode with BLK_OPEN_EXCL
flag to lo_ioctl.
3. loop0 was not opened by dm_get_device() in BLK_OPEN_EXCL mode. As a
result, whole->bd_claiming is NULL.
Thus, the BLK_OPEN_EXCL flag in the mode passed to lo_ioctl doesn't
guarantee the loop device was opened with BLK_OPEN_EXCL.
How about use per-device rw_semaphore instead of 'bd_prepare_to_claim/
bd_abort_claiming'?
Yongpeng,
err = mutex_lock_killable(&lo->lo_mutex);
if (err)
return err;
@@ -1270,6 +1281,9 @@ loop_set_status(struct loop_device *lo, const struct loop_info64 *info)
}
out_unlock:
mutex_unlock(&lo->lo_mutex);
+ if (!(mode & BLK_OPEN_EXCL))
+ bd_abort_claiming(bdev, loop_set_status);
+out_reread_partitions:
if (partscan)
loop_reread_partitions(lo);
@@ -1349,7 +1363,9 @@ loop_info64_to_old(const struct loop_info64 *info64, struct loop_info *info)
}
static int
-loop_set_status_old(struct loop_device *lo, const struct loop_info __user *arg)
+loop_set_status_old(struct loop_device *lo, blk_mode_t mode,
+ struct block_device *bdev,
+ const struct loop_info __user *arg)
{
struct loop_info info;
struct loop_info64 info64;
@@ -1357,17 +1373,19 @@ loop_set_status_old(struct loop_device *lo, const struct loop_info __user *arg)
if (copy_from_user(&info, arg, sizeof (struct loop_info)))
return -EFAULT;
loop_info64_from_old(&info, &info64);
- return loop_set_status(lo, &info64);
+ return loop_set_status(lo, mode, bdev, &info64);
}
static int
-loop_set_status64(struct loop_device *lo, const struct loop_info64 __user *arg)
+loop_set_status64(struct loop_device *lo, blk_mode_t mode,
+ struct block_device *bdev,
+ const struct loop_info64 __user *arg)
{
struct loop_info64 info64;
if (copy_from_user(&info64, arg, sizeof (struct loop_info64)))
return -EFAULT;
- return loop_set_status(lo, &info64);
+ return loop_set_status(lo, mode, bdev, &info64);
}
static int
@@ -1546,14 +1564,14 @@ static int lo_ioctl(struct block_device *bdev, blk_mode_t mode,
case LOOP_SET_STATUS:
err = -EPERM;
if ((mode & BLK_OPEN_WRITE) || capable(CAP_SYS_ADMIN))
- err = loop_set_status_old(lo, argp);
+ err = loop_set_status_old(lo, mode, bdev, argp);
break;
case LOOP_GET_STATUS:
return loop_get_status_old(lo, argp);
case LOOP_SET_STATUS64:
err = -EPERM;
if ((mode & BLK_OPEN_WRITE) || capable(CAP_SYS_ADMIN))
- err = loop_set_status64(lo, argp);
+ err = loop_set_status64(lo, mode, bdev, argp);
break;
case LOOP_GET_STATUS64:
return loop_get_status64(lo, argp);
@@ -1647,8 +1665,9 @@ loop_info64_to_compat(const struct loop_info64 *info64,
}
static int
-loop_set_status_compat(struct loop_device *lo,
- const struct compat_loop_info __user *arg)
+loop_set_status_compat(struct loop_device *lo, blk_mode_t mode,
+ struct block_device *bdev,
+ const struct compat_loop_info __user *arg)
{
struct loop_info64 info64;
int ret;
@@ -1656,7 +1675,7 @@ loop_set_status_compat(struct loop_device *lo,
ret = loop_info64_from_compat(arg, &info64);
if (ret < 0)
return ret;
- return loop_set_status(lo, &info64);
+ return loop_set_status(lo, mode, bdev, &info64);
}
static int
@@ -1682,7 +1701,7 @@ static int lo_compat_ioctl(struct block_device *bdev, blk_mode_t mode,
switch(cmd) {
case LOOP_SET_STATUS:
- err = loop_set_status_compat(lo,
+ err = loop_set_status_compat(lo, mode, bdev,
(const struct compat_loop_info __user *)arg);
break;
case LOOP_GET_STATUS: