[PATCH security v1] apparmor: fix a use-after-free in aa_lookupn_ns()

[PATCH security v1] apparmor: fix a use-after-free in aa_lookupn_ns()

[Binbin Deng]

aa_lookupn_ns() takes a reference on the namespace it finds with
aa_get_ns(),which is kref_get() on the reference count that the
namespace is released through, without checking whether the namespace
is still in use.  A concurrent removal of a ":"-prefixed namespace
looks the same namespace up and only then takes the parent lock to
remove it, so the last reference can be dropped between the lookup and
the aa_get_ns() call: the reference count then reports the increment on
a zero count and the kernel classifies the following use of the object
as a use-after-free.  A user with CAP_MAC_ADMIN can trigger this by
writing a ":<name>" removal and a ":<name>/<profile>" removal to
/sys/kernel/security/apparmor/.remove concurrently.

refcount_t: addition on 0; use-after-free.
WARNING: lib/refcount.c:25 at refcount_warn_saturate+0xc0/0xe0, CPU#6
Call Trace:
 <TASK>
 aa_lookupn_ns+0xa9/0xc0
 aa_remove_profiles+0x4fb/0x730
 ? __kvmalloc_node_noprof+0x219/0x5c0
 ? __pfx_aa_remove_profiles+0x10/0x10
 ? aa_loaddata_alloc+0x65/0x1f0
 ? _copy_from_user+0x2d/0x80
 ? aa_loaddata_alloc+0x111/0x1f0
 profile_remove+0x2ee/0x4a0
 vfs_write+0x21e/0xcf0
 ? rcu_core+0x29c/0x1740
 ? timerqueue_linked_add+0x1f4/0x3d0
 ? __pfx_vfs_write+0x10/0x10
 ? mutex_lock+0x81/0xe0
 ? __pfx_mutex_lock+0x10/0x10
 ? __pfx_rcu_core+0x10/0x10
 ? __pfx_hrtimer_update_next_event+0x10/0x10
 ? mutex_unlock+0x7b/0xd0
 ? fdget_pos+0x24d/0x4b0
 ksys_write+0xf7/0x1c0
 ? __pfx_ksys_write+0x10/0x10
 ? restore_fpregs_from_fpstate+0x37/0xc0
 do_syscall_64+0xf9/0x540
 entry_SYSCALL_64_after_hwframe+0x77/0x7f
 </TASK>

Fix by rejecting a namespace whose reference count.

Fixes: 3664268f19ea0 ("apparmor: add namespace lookup fns()")
Signed-off-by: Binbin Deng <18983559317@163.com>
---
 security/apparmor/include/policy_ns.h | 16 ++++++++++++++++
 security/apparmor/policy_ns.c         |  2 +-
 2 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/security/apparmor/include/policy_ns.h b/security/apparmor/include/policy_ns.h
index cc6e84151812..4f8359875476 100644
--- a/security/apparmor/include/policy_ns.h
+++ b/security/apparmor/include/policy_ns.h
@@ -117,6 +117,22 @@ static inline struct aa_ns *aa_get_ns(struct aa_ns *ns)
 	return ns;
 }
 
+/**
+ * aa_get_ns_not0 - increment references count on @ns found via lookup
+ * @ns: namespace to increment reference count of (MAYBE NULL)
+ *
+ * Returns: pointer to @ns, if @ns is NULL returns NULL,
+ *          NULL if @ns is being freed
+ * Requires: @ns must be held with valid refcount when called
+ */
+static inline struct aa_ns *aa_get_ns_not0(struct aa_ns *ns)
+{
+	if (ns && aa_get_profile_not0(ns->unconfined))
+		return ns;
+
+	return NULL;
+}
+
 /**
  * aa_put_ns - decrement refcount on @ns
  * @ns: namespace to put reference of
diff --git a/security/apparmor/policy_ns.c b/security/apparmor/policy_ns.c
index 5a907a875d8f..ae9168c1a85e 100644
--- a/security/apparmor/policy_ns.c
+++ b/security/apparmor/policy_ns.c
@@ -207,7 +207,7 @@ struct aa_ns *aa_lookupn_ns(struct aa_ns *view, const char *name, size_t n)
 	struct aa_ns *ns = NULL;
 
 	rcu_read_lock();
-	ns = aa_get_ns(__aa_lookupn_ns(view, name, n));
+	ns = aa_get_ns_not0(__aa_lookupn_ns(view, name, n));
 	rcu_read_unlock();
 
 	return ns;
-- 
2.43.0

Re: [PATCH security v1] apparmor: fix a use-after-free in aa_lookupn_ns()

[sashiko-bot]

> aa_lookupn_ns() takes a reference on the namespace it finds with
> aa_get_ns(),which is kref_get() on the reference count that the
> namespace is released through, without checking whether the namespace
> is still in use.  A concurrent removal of a ":"-prefixed namespace
> looks the same namespace up and only then takes the parent lock to
> remove it, so the last reference can be dropped between the lookup and
> the aa_get_ns() call: the reference count then reports the increment on
> a zero count and the kernel classifies the following use of the object
> as a use-after-free.  A user with CAP_MAC_ADMIN can trigger this by
> writing a ":<name>" removal and a ":<name>/<profile>" removal to
> /sys/kernel/security/apparmor/.remove concurrently.
> 
> refcount_t: addition on 0; use-after-free.
> WARNING: lib/refcount.c:25 at refcount_warn_saturate+0xc0/0xe0, CPU#6
> Call Trace:
> [ ... ]
> 
> Fix by rejecting a namespace whose reference count.
> 
> Fixes: 3664268f19ea0 ("apparmor: add namespace lookup fns()")
> Signed-off-by: Binbin Deng <18983559317@163.com>

Sashiko has reviewed this patch and found no issues. It looks great!

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260923054623.247634-1-18983559317@163.com?part=1

[PATCH security v1] apparmor: fix a use-after-free in aa_lookupn_ns()

[Binbin Deng]

aa_lookupn_ns() takes a reference on the namespace it finds with
aa_get_ns(),which is kref_get() on the reference count that the
namespace is released through, without checking whether the namespace
is still in use.  A concurrent removal of a ":"-prefixed namespace
looks the same namespace up and only then takes the parent lock to
remove it, so the last reference can be dropped between the lookup and
the aa_get_ns() call: the reference count then reports the increment on
a zero count and the kernel classifies the following use of the object
as a use-after-free.  A user with CAP_MAC_ADMIN can trigger this by
writing a ":<name>" removal and a ":<name>/<profile>" removal to
/sys/kernel/security/apparmor/.remove concurrently.

refcount_t: addition on 0; use-after-free.
WARNING: lib/refcount.c:25 at refcount_warn_saturate+0xc0/0xe0, CPU#6
Call Trace:
 <TASK>
 aa_lookupn_ns+0xa9/0xc0
 aa_remove_profiles+0x4fb/0x730
 ? __kvmalloc_node_noprof+0x219/0x5c0
 ? __pfx_aa_remove_profiles+0x10/0x10
 ? aa_loaddata_alloc+0x65/0x1f0
 ? _copy_from_user+0x2d/0x80
 ? aa_loaddata_alloc+0x111/0x1f0
 profile_remove+0x2ee/0x4a0
 vfs_write+0x21e/0xcf0
 ? rcu_core+0x29c/0x1740
 ? timerqueue_linked_add+0x1f4/0x3d0
 ? __pfx_vfs_write+0x10/0x10
 ? mutex_lock+0x81/0xe0
 ? __pfx_mutex_lock+0x10/0x10
 ? __pfx_rcu_core+0x10/0x10
 ? __pfx_hrtimer_update_next_event+0x10/0x10
 ? mutex_unlock+0x7b/0xd0
 ? fdget_pos+0x24d/0x4b0
 ksys_write+0xf7/0x1c0
 ? __pfx_ksys_write+0x10/0x10
 ? restore_fpregs_from_fpstate+0x37/0xc0
 do_syscall_64+0xf9/0x540
 entry_SYSCALL_64_after_hwframe+0x77/0x7f
 </TASK>

Fix by rejecting a namespace whose reference count.

Fixes: 3664268f19ea0 ("apparmor: add namespace lookup fns()")
Signed-off-by: Binbin Deng <18983559317@163.com>
---
 security/apparmor/include/policy_ns.h | 16 ++++++++++++++++
 security/apparmor/policy_ns.c         |  2 +-
 2 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/security/apparmor/include/policy_ns.h b/security/apparmor/include/policy_ns.h
index cc6e84151812..4f8359875476 100644
--- a/security/apparmor/include/policy_ns.h
+++ b/security/apparmor/include/policy_ns.h
@@ -117,6 +117,22 @@ static inline struct aa_ns *aa_get_ns(struct aa_ns *ns)
 	return ns;
 }
 
+/**
+ * aa_get_ns_not0 - increment references count on @ns found via lookup
+ * @ns: namespace to increment reference count of (MAYBE NULL)
+ *
+ * Returns: pointer to @ns, if @ns is NULL returns NULL,
+ *          NULL if @ns is being freed
+ * Requires: @ns must be held with valid refcount when called
+ */
+static inline struct aa_ns *aa_get_ns_not0(struct aa_ns *ns)
+{
+	if (ns && aa_get_profile_not0(ns->unconfined))
+		return ns;
+
+	return NULL;
+}
+
 /**
  * aa_put_ns - decrement refcount on @ns
  * @ns: namespace to put reference of
diff --git a/security/apparmor/policy_ns.c b/security/apparmor/policy_ns.c
index 5a907a875d8f..ae9168c1a85e 100644
--- a/security/apparmor/policy_ns.c
+++ b/security/apparmor/policy_ns.c
@@ -207,7 +207,7 @@ struct aa_ns *aa_lookupn_ns(struct aa_ns *view, const char *name, size_t n)
 	struct aa_ns *ns = NULL;
 
 	rcu_read_lock();
-	ns = aa_get_ns(__aa_lookupn_ns(view, name, n));
+	ns = aa_get_ns_not0(__aa_lookupn_ns(view, name, n));
 	rcu_read_unlock();
 
 	return ns;
-- 
2.43.0

Re: [PATCH security v1] apparmor: fix a use-after-free in aa_lookupn_ns()

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] apparmor: incorrect documentation for aa_get_ns_not0()
--

commit fda3772b41365651e2bf01f14c3712bb81c8786a
Author: Binbin Deng <18983559317@163.com>

apparmor: fix a use-after-free in aa_lookupn_ns()

This commit addresses a use-after-free bug in aa_lookupn_ns() caused by
concurrent removal of namespaces. It introduces aa_get_ns_not0() to safely
attempt incrementing the reference count only if the namespace is not already
being freed.

> diff --git a/security/apparmor/include/policy_ns.h b/security/apparmor/include/policy_ns.h
> index cc6e841518120..4f83598754764 100644
> --- a/security/apparmor/include/policy_ns.h
> +++ b/security/apparmor/include/policy_ns.h

[ ... ]

> @@ -117,6 +117,22 @@ static inline struct aa_ns *aa_get_ns(struct aa_ns *ns)
>  	return ns;
>  }
>  
> +/**
> + * aa_get_ns_not0 - increment references count on @ns found via lookup
> + * @ns: namespace to increment reference count of (MAYBE NULL)
> + *
> + * Returns: pointer to @ns, if @ns is NULL returns NULL,
> + *          NULL if @ns is being freed
> + * Requires: @ns must be held with valid refcount when called

[Severity: Low]
Does this documentation contradict the purpose of the newly introduced function?

The function aa_get_ns_not0() is explicitly designed to safely handle
namespaces where the reference count might have already dropped to zero, which
fixes the use-after-free bug. However, the documentation requires the caller
to already hold a valid refcount, which defeats the purpose of checking for
zero.

Should this state that an RCU read lock is required to protect the lookup
pointer instead?

> + */
> +static inline struct aa_ns *aa_get_ns_not0(struct aa_ns *ns)
> +{
> +	if (ns && aa_get_profile_not0(ns->unconfined))
> +		return ns;
> +
> +	return NULL;
> +}

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260926044724.4187-1-18983559317@163.com?part=1