[v5] driver core: Fix use-after-free and double free on glue directory
diff mbox series

Message ID 20190718111958.17336-1-smuchun@gmail.com
State Superseded
Headers show
Series
  • [v5] driver core: Fix use-after-free and double free on glue directory
Related show

Commit Message

Muchun Song July 18, 2019, 11:19 a.m. UTC
There is a race condition between removing glue directory and adding a new
device under the glue directory. It can be reproduced in following test:

path 1: Add the child device under glue dir
device_add()
    get_device_parent()
        mutex_lock(&gdp_mutex);
        ....
        /*find parent from glue_dirs.list*/
        list_for_each_entry(k, &dev->class->p->glue_dirs.list, entry)
            if (k->parent == parent_kobj) {
                kobj = kobject_get(k);
                break;
            }
        ....
        mutex_unlock(&gdp_mutex);
        ....
    ....
    kobject_add()
        kobject_add_internal()
            create_dir()
                sysfs_create_dir_ns()
                    if (kobj->parent)
                        parent = kobj->parent->sd;
                    ....
                    kernfs_create_dir_ns(parent)
                        kernfs_new_node()
                            kernfs_get(parent)
                        ....
                        /* link in */
                        rc = kernfs_add_one(kn);
                        if (!rc)
                            return kn;

                        kernfs_put(kn)
                            ....
                            repeat:
                            kmem_cache_free(kn)
                            kn = parent;

                            if (kn) {
                                if (atomic_dec_and_test(&kn->count))
                                    goto repeat;
                            }
                        ....

path2: Remove last child device under glue dir
device_del()
    cleanup_glue_dir()
        mutex_lock(&gdp_mutex);
        if (!kobject_has_children(glue_dir))
            kobject_del(glue_dir);
        kobject_put(glue_dir);
        mutex_unlock(&gdp_mutex);

Before path2 remove last child device under glue dir, If path1 add a new
device under glue dir, the glue_dir kobject reference count will be
increase to 2 via kobject_get(k) in get_device_parent(). And path1 has
been called kernfs_new_node(), but not call kernfs_get(parent).
Meanwhile, path2 call kobject_del(glue_dir) beacause 0 is returned by
kobject_has_children(). This result in glue_dir->sd is freed and it's
reference count will be 0. Then path1 call kernfs_get(parent) will trigger
a warning in kernfs_get()(WARN_ON(!atomic_read(&kn->count))) and increase
it's reference count to 1. Because glue_dir->sd is freed by path2, the next
call kernfs_add_one() by path1 will fail(This is also use-after-free)
and call atomic_dec_and_test() to decrease reference count. Because the
reference count is decremented to 0, it will also call kmem_cache_free()
to free glue_dir->sd again. This will result in double free.

In order to avoid this happening, we also should make sure that kernfs_node
for glue_dir is released in path2 only when refcount for glue_dir kobj is
1 to fix this race.

The following calltrace is captured in kernel 4.14 with the following patch
applied:

commit 726e41097920 ("drivers: core: Remove glue dirs from sysfs earlier")

--------------------------------------------------------------------------
[    3.633703] WARNING: CPU: 4 PID: 513 at .../fs/kernfs/dir.c:494
                Here is WARN_ON(!atomic_read(&kn->count) in kernfs_get().
....
[    3.633986] Call trace:
[    3.633991]  kernfs_create_dir_ns+0xa8/0xb0
[    3.633994]  sysfs_create_dir_ns+0x54/0xe8
[    3.634001]  kobject_add_internal+0x22c/0x3f0
[    3.634005]  kobject_add+0xe4/0x118
[    3.634011]  device_add+0x200/0x870
[    3.634017]  _request_firmware+0x958/0xc38
[    3.634020]  request_firmware_into_buf+0x4c/0x70
....
[    3.634064] kernel BUG at .../mm/slub.c:294!
                Here is BUG_ON(object == fp) in set_freepointer().
....
[    3.634346] Call trace:
[    3.634351]  kmem_cache_free+0x504/0x6b8
[    3.634355]  kernfs_put+0x14c/0x1d8
[    3.634359]  kernfs_create_dir_ns+0x88/0xb0
[    3.634362]  sysfs_create_dir_ns+0x54/0xe8
[    3.634366]  kobject_add_internal+0x22c/0x3f0
[    3.634370]  kobject_add+0xe4/0x118
[    3.634374]  device_add+0x200/0x870
[    3.634378]  _request_firmware+0x958/0xc38
[    3.634381]  request_firmware_into_buf+0x4c/0x70
--------------------------------------------------------------------------

Fixes: 726e41097920 ("drivers: core: Remove glue dirs from sysfs earlier")

Signed-off-by: Muchun Song <smuchun@gmail.com>
---

Change in v5:
       1. Revert to the v1 fix.
       2. Add some comment to explain why we need do this in
          cleanup_glue_dir().
Change in v4:
       1. Add some kerneldoc comment.
       2. Remove unlock_if_glue_dir().
       3. Rename get_device_parent_locked_if_glue_dir() to
          get_device_parent_locked.
       4. Update commit message.
Change in v3:
       Add change log.
Change in v2:
       Fix device_move() also.

 drivers/base/core.c | 54 ++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 53 insertions(+), 1 deletion(-)

Comments

Mukesh Ojha July 24, 2019, 1:41 p.m. UTC | #1
On 7/18/2019 4:49 PM, Muchun Song wrote:
> There is a race condition between removing glue directory and adding a new
> device under the glue directory. It can be reproduced in following test:
>
> path 1: Add the child device under glue dir
> device_add()
>      get_device_parent()
>          mutex_lock(&gdp_mutex);
>          ....
>          /*find parent from glue_dirs.list*/
>          list_for_each_entry(k, &dev->class->p->glue_dirs.list, entry)
>              if (k->parent == parent_kobj) {
>                  kobj = kobject_get(k);
>                  break;
>              }
>          ....
>          mutex_unlock(&gdp_mutex);
>          ....
>      ....
>      kobject_add()
>          kobject_add_internal()
>              create_dir()
>                  sysfs_create_dir_ns()
>                      if (kobj->parent)
>                          parent = kobj->parent->sd;
>                      ....
>                      kernfs_create_dir_ns(parent)
>                          kernfs_new_node()
>                              kernfs_get(parent)
>                          ....
>                          /* link in */
>                          rc = kernfs_add_one(kn);
>                          if (!rc)
>                              return kn;
>
>                          kernfs_put(kn)
>                              ....
>                              repeat:
>                              kmem_cache_free(kn)
>                              kn = parent;
>
>                              if (kn) {
>                                  if (atomic_dec_and_test(&kn->count))
>                                      goto repeat;
>                              }
>                          ....
>
> path2: Remove last child device under glue dir
> device_del()
>      cleanup_glue_dir()
>          mutex_lock(&gdp_mutex);
>          if (!kobject_has_children(glue_dir))
>              kobject_del(glue_dir);
>          kobject_put(glue_dir);
>          mutex_unlock(&gdp_mutex);
>
> Before path2 remove last child device under glue dir, If path1 add a new
> device under glue dir, the glue_dir kobject reference count will be
> increase to 2 via kobject_get(k) in get_device_parent(). And path1 has
> been called kernfs_new_node(), but not call kernfs_get(parent).
> Meanwhile, path2 call kobject_del(glue_dir) beacause 0 is returned by
> kobject_has_children(). This result in glue_dir->sd is freed and it's
> reference count will be 0. Then path1 call kernfs_get(parent) will trigger
> a warning in kernfs_get()(WARN_ON(!atomic_read(&kn->count))) and increase
> it's reference count to 1. Because glue_dir->sd is freed by path2, the next
> call kernfs_add_one() by path1 will fail(This is also use-after-free)
> and call atomic_dec_and_test() to decrease reference count. Because the
> reference count is decremented to 0, it will also call kmem_cache_free()
> to free glue_dir->sd again. This will result in double free.
>
> In order to avoid this happening, we also should make sure that kernfs_node
> for glue_dir is released in path2 only when refcount for glue_dir kobj is
> 1 to fix this race.
>
> The following calltrace is captured in kernel 4.14 with the following patch
> applied:
>
> commit 726e41097920 ("drivers: core: Remove glue dirs from sysfs earlier")
>
> --------------------------------------------------------------------------
> [    3.633703] WARNING: CPU: 4 PID: 513 at .../fs/kernfs/dir.c:494
>                  Here is WARN_ON(!atomic_read(&kn->count) in kernfs_get().
> ....
> [    3.633986] Call trace:
> [    3.633991]  kernfs_create_dir_ns+0xa8/0xb0
> [    3.633994]  sysfs_create_dir_ns+0x54/0xe8
> [    3.634001]  kobject_add_internal+0x22c/0x3f0
> [    3.634005]  kobject_add+0xe4/0x118
> [    3.634011]  device_add+0x200/0x870
> [    3.634017]  _request_firmware+0x958/0xc38
> [    3.634020]  request_firmware_into_buf+0x4c/0x70
> ....
> [    3.634064] kernel BUG at .../mm/slub.c:294!
>                  Here is BUG_ON(object == fp) in set_freepointer().
> ....
> [    3.634346] Call trace:
> [    3.634351]  kmem_cache_free+0x504/0x6b8
> [    3.634355]  kernfs_put+0x14c/0x1d8
> [    3.634359]  kernfs_create_dir_ns+0x88/0xb0
> [    3.634362]  sysfs_create_dir_ns+0x54/0xe8
> [    3.634366]  kobject_add_internal+0x22c/0x3f0
> [    3.634370]  kobject_add+0xe4/0x118
> [    3.634374]  device_add+0x200/0x870
> [    3.634378]  _request_firmware+0x958/0xc38
> [    3.634381]  request_firmware_into_buf+0x4c/0x70
> --------------------------------------------------------------------------
>
> Fixes: 726e41097920 ("drivers: core: Remove glue dirs from sysfs earlier")
>
> Signed-off-by: Muchun Song <smuchun@gmail.com>
> ---
>
> Change in v5:
>         1. Revert to the v1 fix.
>         2. Add some comment to explain why we need do this in
>            cleanup_glue_dir().
> Change in v4:
>         1. Add some kerneldoc comment.
>         2. Remove unlock_if_glue_dir().
>         3. Rename get_device_parent_locked_if_glue_dir() to
>            get_device_parent_locked.
>         4. Update commit message.
> Change in v3:
>         Add change log.
> Change in v2:
>         Fix device_move() also.
>
>   drivers/base/core.c | 54 ++++++++++++++++++++++++++++++++++++++++++++-
>   1 file changed, 53 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/base/core.c b/drivers/base/core.c
> index 4aeaa0c92bda..1a67ee325584 100644
> --- a/drivers/base/core.c
> +++ b/drivers/base/core.c
> @@ -1825,7 +1825,59 @@ static void cleanup_glue_dir(struct device *dev, struct kobject *glue_dir)
>   		return;
>   
>   	mutex_lock(&gdp_mutex);
> -	if (!kobject_has_children(glue_dir))
> +	/**
> +	 * There is a race condition between removing glue directory
> +	 * and adding a new device under the glue directory.
> +	 *
> +	 * path 1: Add the child device under glue dir
> +	 * device_add()
> +	 *	get_device_parent()
> +	 *		mutex_lock(&gdp_mutex);
> +	 *		....
> +	 *		list_for_each_entry(k, &dev->class->p->glue_dirs.list,
> +	 *				    entry)
> +	 *		if (k->parent == parent_kobj) {
> +	 *			kobj = kobject_get(k);
> +	 *			break;
> +	 *		}
> +	 *		....
> +	 *		mutex_unlock(&gdp_mutex);
> +	 *		....
> +	 *	....
> +	 *	kobject_add()
> +	 *		kobject_add_internal()
> +	 *		create_dir()
> +	 *			....
> +	 *			if (kobj->parent)
> +	 *				parent = kobj->parent->sd;
> +	 *			....
> +	 *			kernfs_create_dir_ns(parent)
> +	 *				....
> +	 *
> +	 * path2: Remove last child device under glue dir
> +	 * device_del()
> +	 *	cleanup_glue_dir()
> +	 *		....
> +	 *		mutex_lock(&gdp_mutex);
> +	 *		if (!kobject_has_children(glue_dir))
> +	 *			kobject_del(glue_dir);
> +	 *		....
> +	 *		mutex_unlock(&gdp_mutex);
> +	 *
> +	 * Before path2 remove last child device under glue dir, if path1 add
> +	 * a new device under glue dir, the glue_dir kobject reference count
> +	 * will be increase to 2 via kobject_get(k) in get_device_parent().
> +	 * And path1 has been called kernfs_create_dir_ns(). Meanwhile,
> +	 * path2 call kobject_del(glue_dir) beacause 0 is returned by
> +	 * kobject_has_children(). This result in glue_dir->sd is freed.
> +	 * Then the path1 will see a stale "empty" but still potentially used
> +	 * glue dir around.
> +	 *
> +	 * In order to avoid this happening, we also should make sure that
> +	 * kernfs_node for glue_dir is released in path2 only when refcount
> +	 * for glue_dir kobj is 1.
> +	 */
> +	if (!kobject_has_children(glue_dir) && kref_read(&glue_dir->kref) == 1)
Instead of hardcoding "1 " , can't we check against zero Like Prateek 
sood did in his patch.https://lkml.org/lkml/2019/5/1/3

+ref = kref_read(&glue_dir->kref);
+if  (!kobject_has_children(glue_dir) && --ref)


We have seen many pattern of this issue which is coming from different 
driver e.g uinput as well in 4.14 
kernel.(https://lkml.org/lkml/2019/4/10/146)
The reason why it  started coming after

726e41097920 ("drivers: core: Remove glue dirs from sysfs earlier")

Looks good to me

Reviewed-by: Mukesh Ojha  <mojha@codeaurora.org>


-Mukesh


>   		kobject_del(glue_dir);
>   	kobject_put(glue_dir);
>   	mutex_unlock(&gdp_mutex);
Mukesh Ojha July 24, 2019, 2:23 p.m. UTC | #2
On 7/24/2019 7:11 PM, Mukesh Ojha wrote:
>
> On 7/18/2019 4:49 PM, Muchun Song wrote:
>> There is a race condition between removing glue directory and adding 
>> a new
>> device under the glue directory. It can be reproduced in following test:
>>
>> path 1: Add the child device under glue dir
>> device_add()
>>      get_device_parent()
>>          mutex_lock(&gdp_mutex);
>>          ....
>>          /*find parent from glue_dirs.list*/
>>          list_for_each_entry(k, &dev->class->p->glue_dirs.list, entry)
>>              if (k->parent == parent_kobj) {
>>                  kobj = kobject_get(k);
>>                  break;
>>              }
>>          ....
>>          mutex_unlock(&gdp_mutex);
>>          ....
>>      ....
>>      kobject_add()
>>          kobject_add_internal()
>>              create_dir()
>>                  sysfs_create_dir_ns()
>>                      if (kobj->parent)
>>                          parent = kobj->parent->sd;
>>                      ....
>>                      kernfs_create_dir_ns(parent)
>>                          kernfs_new_node()
>>                              kernfs_get(parent)
>>                          ....
>>                          /* link in */
>>                          rc = kernfs_add_one(kn);
>>                          if (!rc)
>>                              return kn;
>>
>>                          kernfs_put(kn)
>>                              ....
>>                              repeat:
>>                              kmem_cache_free(kn)
>>                              kn = parent;
>>
>>                              if (kn) {
>>                                  if (atomic_dec_and_test(&kn->count))
>>                                      goto repeat;
>>                              }
>>                          ....
>>
>> path2: Remove last child device under glue dir
>> device_del()
>>      cleanup_glue_dir()
>>          mutex_lock(&gdp_mutex);
>>          if (!kobject_has_children(glue_dir))
>>              kobject_del(glue_dir);
>>          kobject_put(glue_dir);
>>          mutex_unlock(&gdp_mutex);
>>
>> Before path2 remove last child device under glue dir, If path1 add a new
>> device under glue dir, the glue_dir kobject reference count will be
>> increase to 2 via kobject_get(k) in get_device_parent(). And path1 has
>> been called kernfs_new_node(), but not call kernfs_get(parent).
>> Meanwhile, path2 call kobject_del(glue_dir) beacause 0 is returned by
>> kobject_has_children(). This result in glue_dir->sd is freed and it's
>> reference count will be 0. Then path1 call kernfs_get(parent) will 
>> trigger
>> a warning in kernfs_get()(WARN_ON(!atomic_read(&kn->count))) and 
>> increase
>> it's reference count to 1. Because glue_dir->sd is freed by path2, 
>> the next
>> call kernfs_add_one() by path1 will fail(This is also use-after-free)
>> and call atomic_dec_and_test() to decrease reference count. Because the
>> reference count is decremented to 0, it will also call kmem_cache_free()
>> to free glue_dir->sd again. This will result in double free.
>>
>> In order to avoid this happening, we also should make sure that 
>> kernfs_node
>> for glue_dir is released in path2 only when refcount for glue_dir 
>> kobj is
>> 1 to fix this race.
>>
>> The following calltrace is captured in kernel 4.14 with the following 
>> patch
>> applied:
>>
>> commit 726e41097920 ("drivers: core: Remove glue dirs from sysfs 
>> earlier")
>>
>> -------------------------------------------------------------------------- 
>>
>> [    3.633703] WARNING: CPU: 4 PID: 513 at .../fs/kernfs/dir.c:494
>>                  Here is WARN_ON(!atomic_read(&kn->count) in 
>> kernfs_get().
>> ....
>> [    3.633986] Call trace:
>> [    3.633991]  kernfs_create_dir_ns+0xa8/0xb0
>> [    3.633994]  sysfs_create_dir_ns+0x54/0xe8
>> [    3.634001]  kobject_add_internal+0x22c/0x3f0
>> [    3.634005]  kobject_add+0xe4/0x118
>> [    3.634011]  device_add+0x200/0x870
>> [    3.634017]  _request_firmware+0x958/0xc38
>> [    3.634020]  request_firmware_into_buf+0x4c/0x70
>> ....
>> [    3.634064] kernel BUG at .../mm/slub.c:294!
>>                  Here is BUG_ON(object == fp) in set_freepointer().
>> ....
>> [    3.634346] Call trace:
>> [    3.634351]  kmem_cache_free+0x504/0x6b8
>> [    3.634355]  kernfs_put+0x14c/0x1d8
>> [    3.634359]  kernfs_create_dir_ns+0x88/0xb0
>> [    3.634362]  sysfs_create_dir_ns+0x54/0xe8
>> [    3.634366]  kobject_add_internal+0x22c/0x3f0
>> [    3.634370]  kobject_add+0xe4/0x118
>> [    3.634374]  device_add+0x200/0x870
>> [    3.634378]  _request_firmware+0x958/0xc38
>> [    3.634381]  request_firmware_into_buf+0x4c/0x70
>> -------------------------------------------------------------------------- 
>>
>>
>> Fixes: 726e41097920 ("drivers: core: Remove glue dirs from sysfs 
>> earlier")
>>
>> Signed-off-by: Muchun Song <smuchun@gmail.com>
>> ---
>>
>> Change in v5:
>>         1. Revert to the v1 fix.
>>         2. Add some comment to explain why we need do this in
>>            cleanup_glue_dir().
>> Change in v4:
>>         1. Add some kerneldoc comment.
>>         2. Remove unlock_if_glue_dir().
>>         3. Rename get_device_parent_locked_if_glue_dir() to
>>            get_device_parent_locked.
>>         4. Update commit message.
>> Change in v3:
>>         Add change log.
>> Change in v2:
>>         Fix device_move() also.
>>
>>   drivers/base/core.c | 54 ++++++++++++++++++++++++++++++++++++++++++++-
>>   1 file changed, 53 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/base/core.c b/drivers/base/core.c
>> index 4aeaa0c92bda..1a67ee325584 100644
>> --- a/drivers/base/core.c
>> +++ b/drivers/base/core.c
>> @@ -1825,7 +1825,59 @@ static void cleanup_glue_dir(struct device 
>> *dev, struct kobject *glue_dir)
>>           return;
>>         mutex_lock(&gdp_mutex);
>> -    if (!kobject_has_children(glue_dir))
>> +    /**
>> +     * There is a race condition between removing glue directory
>> +     * and adding a new device under the glue directory.
>> +     *
>> +     * path 1: Add the child device under glue dir
>> +     * device_add()
>> +     *    get_device_parent()
>> +     *        mutex_lock(&gdp_mutex);
>> +     *        ....
>> +     *        list_for_each_entry(k, &dev->class->p->glue_dirs.list,
>> +     *                    entry)
>> +     *        if (k->parent == parent_kobj) {
>> +     *            kobj = kobject_get(k);
>> +     *            break;
>> +     *        }
>> +     *        ....
>> +     *        mutex_unlock(&gdp_mutex);
>> +     *        ....
>> +     *    ....
>> +     *    kobject_add()
>> +     *        kobject_add_internal()
>> +     *        create_dir()
>> +     *            ....
>> +     *            if (kobj->parent)
>> +     *                parent = kobj->parent->sd;
>> +     *            ....
>> +     *            kernfs_create_dir_ns(parent)
>> +     *                ....
>> +     *
>> +     * path2: Remove last child device under glue dir
>> +     * device_del()
>> +     *    cleanup_glue_dir()
>> +     *        ....
>> +     *        mutex_lock(&gdp_mutex);
>> +     *        if (!kobject_has_children(glue_dir))
>> +     *            kobject_del(glue_dir);
>> +     *        ....
>> +     *        mutex_unlock(&gdp_mutex);
>> +     *
>> +     * Before path2 remove last child device under glue dir, if 
>> path1 add
>> +     * a new device under glue dir, the glue_dir kobject reference 
>> count
>> +     * will be increase to 2 via kobject_get(k) in get_device_parent().
>> +     * And path1 has been called kernfs_create_dir_ns(). Meanwhile,
>> +     * path2 call kobject_del(glue_dir) beacause 0 is returned by
>> +     * kobject_has_children(). This result in glue_dir->sd is freed.
>> +     * Then the path1 will see a stale "empty" but still potentially 
>> used
>> +     * glue dir around.
>> +     *
>> +     * In order to avoid this happening, we also should make sure that
>> +     * kernfs_node for glue_dir is released in path2 only when refcount
>> +     * for glue_dir kobj is 1.
>> +     */
>> +    if (!kobject_has_children(glue_dir) && 
>> kref_read(&glue_dir->kref) == 1)
> Instead of hardcoding "1 " , can't we check against zero Like Prateek 
> sood did in his patch.https://lkml.org/lkml/2019/5/1/3
>
> +ref = kref_read(&glue_dir->kref);
> +if  (!kobject_has_children(glue_dir) && --ref)
>
I meant if  (!kobject_has_children(glue_dir) && ! --ref)
>
>
> We have seen many pattern of this issue which is coming from different 
> driver e.g uinput as well in 4.14 
> kernel.(https://lkml.org/lkml/2019/4/10/146)
> The reason why it  started coming after
>
> 726e41097920 ("drivers: core: Remove glue dirs from sysfs earlier")
>
> Looks good to me
>
> Reviewed-by: Mukesh Ojha  <mojha@codeaurora.org>
>
>
> -Mukesh
>
>
>>           kobject_del(glue_dir);
>>       kobject_put(glue_dir);
>>       mutex_unlock(&gdp_mutex);

Patch
diff mbox series

diff --git a/drivers/base/core.c b/drivers/base/core.c
index 4aeaa0c92bda..1a67ee325584 100644
--- a/drivers/base/core.c
+++ b/drivers/base/core.c
@@ -1825,7 +1825,59 @@  static void cleanup_glue_dir(struct device *dev, struct kobject *glue_dir)
 		return;
 
 	mutex_lock(&gdp_mutex);
-	if (!kobject_has_children(glue_dir))
+	/**
+	 * There is a race condition between removing glue directory
+	 * and adding a new device under the glue directory.
+	 *
+	 * path 1: Add the child device under glue dir
+	 * device_add()
+	 *	get_device_parent()
+	 *		mutex_lock(&gdp_mutex);
+	 *		....
+	 *		list_for_each_entry(k, &dev->class->p->glue_dirs.list,
+	 *				    entry)
+	 *		if (k->parent == parent_kobj) {
+	 *			kobj = kobject_get(k);
+	 *			break;
+	 *		}
+	 *		....
+	 *		mutex_unlock(&gdp_mutex);
+	 *		....
+	 *	....
+	 *	kobject_add()
+	 *		kobject_add_internal()
+	 *		create_dir()
+	 *			....
+	 *			if (kobj->parent)
+	 *				parent = kobj->parent->sd;
+	 *			....
+	 *			kernfs_create_dir_ns(parent)
+	 *				....
+	 *
+	 * path2: Remove last child device under glue dir
+	 * device_del()
+	 *	cleanup_glue_dir()
+	 *		....
+	 *		mutex_lock(&gdp_mutex);
+	 *		if (!kobject_has_children(glue_dir))
+	 *			kobject_del(glue_dir);
+	 *		....
+	 *		mutex_unlock(&gdp_mutex);
+	 *
+	 * Before path2 remove last child device under glue dir, if path1 add
+	 * a new device under glue dir, the glue_dir kobject reference count
+	 * will be increase to 2 via kobject_get(k) in get_device_parent().
+	 * And path1 has been called kernfs_create_dir_ns(). Meanwhile,
+	 * path2 call kobject_del(glue_dir) beacause 0 is returned by
+	 * kobject_has_children(). This result in glue_dir->sd is freed.
+	 * Then the path1 will see a stale "empty" but still potentially used
+	 * glue dir around.
+	 *
+	 * In order to avoid this happening, we also should make sure that
+	 * kernfs_node for glue_dir is released in path2 only when refcount
+	 * for glue_dir kobj is 1.
+	 */
+	if (!kobject_has_children(glue_dir) && kref_read(&glue_dir->kref) == 1)
 		kobject_del(glue_dir);
 	kobject_put(glue_dir);
 	mutex_unlock(&gdp_mutex);