ns: add active reference count

The namespace tree is, among other things, currently used to support
file handles for namespaces. When a namespace is created it is placed on
the namespace trees and when it is destroyed it is removed from the
namespace trees.

While a namespace is on the namespace trees with a valid reference count
it is possible to reopen it through a namespace file handle. This is all
fine but has some issues that should be addressed.

On current kernels a namespace is visible to userspace in the
following cases:

(1) The namespace is in use by a task.
(2) The namespace is persisted through a VFS object (namespace file
    descriptor or bind-mount).
    Note that (2) only cares about direct persistence of the namespace
    itself not indirectly via e.g., file->f_cred file references or
    similar.
(3) The namespace is a hierarchical namespace type and is the parent of
    a single or multiple child namespaces.

Case (3) is interesting because it is possible that a parent namespace
might not fulfill any of (1) or (2), i.e., is invisible to userspace but
it may still be resurrected through the NS_GET_PARENT ioctl().

Currently namespace file handles allow much broader access to namespaces
than what is currently possible via (1)-(3). The reason is that
namespaces may remain pinned for completely internal reasons yet are
inaccessible to userspace.

For example, a user namespace my remain pinned by get_cred() calls to
stash the opener's credentials into file->f_cred. As it stands file
handles allow to resurrect such a users namespace even though this
should not be possible via (1)-(3). This is a fundamental uapi change
that we shouldn't do if we don't have to.

Consider the following insane case: Various architectures support the
CONFIG_MMU_LAZY_TLB_REFCOUNT option which uses lazy TLB destruction.
When this option is set a userspace task's struct mm_struct may be used
for kernel threads such as the idle task and will only be destroyed once
the cpu's runqueue switches back to another task. But because of ptrace()
permission checks struct mm_struct stashes the user namespace of the
task that struct mm_struct originally belonged to. The kernel thread
will take a reference on the struct mm_struct and thus pin it.

So on an idle system user namespaces can be persisted for arbitrary
amounts of time which also means that they can be resurrected using
namespace file handles. That makes no sense whatsoever. The problem is
of course excarabted on large systems with a huge number of cpus.

To handle this nicely we introduce an active reference count which
tracks (1)-(3). This is easy to do as all of these things are already
managed centrally. Only (1)-(3) will count towards the active reference
count and only namespaces which are active may be opened via namespace
file handles.

The problem is that namespaces may be resurrected. Which means that they
can become temporarily inactive and will be reactived some time later.
Currently the only example of this is the SIOGCSKNS socket ioctl. The
SIOCGSKNS ioctl allows to open a network namespace file descriptor based
on a socket file descriptor.

If a socket is tied to a network namespace that subsequently becomes
inactive but that socket is persisted by another process in another
network namespace (e.g., via SCM_RIGHTS of pidfd_getfd()) then the
SIOCGSKNS ioctl will resurrect this network namespace.

So calls to open_related_ns() and open_namespace() will end up
resurrecting the corresponding namespace tree.

Note that the active reference count does not regulate the lifetime of
the namespace itself. This is still done by the normal reference count.
The active reference count can only be elevated if the regular reference
count is elevated.

The active reference count also doesn't regulate the presence of a
namespace on the namespace trees. It only regulates its visiblity to
namespace file handles (and in later patches to listns()).

A namespace remains on the namespace trees from creation until its
actual destruction. This will allow the kernel to always reach any
namespace trivially and it will also enable subsystems like bpf to walk
the namespace lists on the system for tracing or general introspection
purposes.

Note that different namespaces have different visibility lifetimes on
current kernels. While most namespace are immediately released when the
last task using them exits, the user- and pid namespace are persisted
and thus both remain accessible via /proc/<pid>/ns/<ns_type>.

The user namespace lifetime is aliged with struct cred and is only
released through exit_creds(). However, it becomes inaccessible to
userspace once the last task using it is reaped, i.e., when
release_task() is called and all proc entries are flushed. Similarly,
the pid namespace is also visible until the last task using it has been
reaped and the associated pid numbers are freed.

The active reference counts of the user- and pid namespace are
decremented once the task is reaped.

Link: https://patch.msgid.link/20251029-work-namespace-nstree-listns-v4-11-2e6f823ebdc0@kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>

3 files changed, 145 insertions, 2 deletions

diff --git a/include/linux/ns_common.h b/include/linux/ns_common.h
index 5e09facafd93..bdd0df15ad9c 100644
--- a/include/linux/ns_common.h
+++ b/include/linux/ns_common.h
@@ -4,7 +4,9 @@
 
 #include <linux/refcount.h>
 #include <linux/rbtree.h>
+#include <linux/vfsdebug.h>
 #include <uapi/linux/sched.h>
+#include <uapi/linux/nsfs.h>
 
 struct proc_ns_operations;
 
@@ -37,6 +39,67 @@ extern const struct proc_ns_operations cgroupns_operations;
 extern const struct proc_ns_operations timens_operations;
 extern const struct proc_ns_operations timens_for_children_operations;
 
+/*
+ * Namespace lifetimes are managed via a two-tier reference counting model:
+ *
+ * (1) __ns_ref (refcount_t): Main reference count tracking memory
+ *     lifetime. Controls when the namespace structure itself is freed.
+ *     It also pins the namespace on the namespace trees whereas (2)
+ *     only regulates their visibility to userspace.
+ *
+ * (2) __ns_ref_active (atomic_t): Reference count tracking active users.
+ *     Controls visibility of the namespace in the namespace trees.
+ *     Any live task that uses the namespace (via nsproxy or cred) holds
+ *     an active reference. Any open file descriptor or bind-mount of
+ *     the namespace holds an active reference. Once all tasks have
+ *     called exited their namespaces and all file descriptors and
+ *     bind-mounts have been released the active reference count drops
+ *     to zero and the namespace becomes inactive. IOW, the namespace
+ *     cannot be listed or opened via file handles anymore.
+ *
+ *     Note that it is valid to transition from active to inactive and
+ *     back from inactive to active e.g., when resurrecting an inactive
+ *     namespace tree via the SIOCGSKNS ioctl().
+ *
+ * Relationship and lifecycle states:
+ *
+ * - Active (__ns_ref_active > 0):
+ *   Namespace is actively used and visible to userspace. The namespace
+ *   can be reopened via /proc/<pid>/ns/<ns_type>, via namespace file
+ *   handles, or discovered via listns().
+ *
+ * - Inactive (__ns_ref_active == 0, __ns_ref > 0):
+ *   No tasks are actively using the namespace and it isn't pinned by
+ *   any bind-mounts or open file descriptors anymore. But the namespace
+ *   is still kept alive by internal references. For example, the user
+ *   namespace could be pinned by an open file through file->f_cred
+ *   references when one of the now defunct tasks had opened a file and
+ *   handed the file descriptor off to another process via a UNIX
+ *   sockets. Such references keep the namespace structure alive through
+ *   __ns_ref but will not hold an active reference.
+ *
+ * - Destroyed (__ns_ref == 0):
+ *   No references remain. The namespace is removed from the tree and freed.
+ *
+ * State transitions:
+ *
+ * Active -> Inactive:
+ *   When the last task using the namespace exits it drops its active
+ *   references to all namespaces. However, user and pid namespaces
+ *   remain accessible until the task has been reaped.
+ *
+ * Inactive -> Active:
+ *   An inactive namespace tree might be resurrected due to e.g., the
+ *   SIOCGSKNS ioctl() on a socket.
+ *
+ * Inactive -> Destroyed:
+ *   When __ns_ref drops to zero the namespace is removed from the
+ *   namespaces trees and the memory is freed (after RCU grace period).
+ *
+ * Initial namespaces:
+ *   Boot-time namespaces (init_net, init_pid_ns, etc.) start with
+ *   __ns_ref_active = 1 and remain active forever.
+ */
 struct ns_common {
 	u32 ns_type;
 	struct dentry *stashed;
@@ -48,6 +111,7 @@ struct ns_common {
 			u64 ns_id;
 			struct rb_node ns_tree_node;
 			struct list_head ns_list_node;
+			atomic_t __ns_ref_active; /* do not use directly */
 		};
 		struct rcu_head ns_rcu;
 	};
@@ -56,6 +120,13 @@ struct ns_common {
 int __ns_common_init(struct ns_common *ns, u32 ns_type, const struct proc_ns_operations *ops, int inum);
 void __ns_common_free(struct ns_common *ns);
 
+static __always_inline bool is_initial_namespace(struct ns_common *ns)
+{
+	VFS_WARN_ON_ONCE(ns->inum == 0);
+	return unlikely(in_range(ns->inum, MNT_NS_INIT_INO,
+				 IPC_NS_INIT_INO - MNT_NS_INIT_INO + 1));
+}
+
 #define to_ns_common(__ns)                                    \
 	_Generic((__ns),                                      \
 		struct cgroup_namespace *:       &(__ns)->ns, \
@@ -127,6 +198,7 @@ void __ns_common_free(struct ns_common *ns);
 	.ops			= to_ns_operations(&nsname),				\
 	.stashed		= NULL,							\
 	.__ns_ref		= REFCOUNT_INIT(refs),					\
+	.__ns_ref_active	= ATOMIC_INIT(1),					\
 	.ns_list_node		= LIST_HEAD_INIT(nsname.ns.ns_list_node),		\
 }
 
@@ -144,14 +216,26 @@ void __ns_common_free(struct ns_common *ns);
 
 #define ns_common_free(__ns) __ns_common_free(to_ns_common((__ns)))
 
+static __always_inline __must_check int __ns_ref_active_read(const struct ns_common *ns)
+{
+	return atomic_read(&ns->__ns_ref_active);
+}
+
 static __always_inline __must_check bool __ns_ref_put(struct ns_common *ns)
 {
-	return refcount_dec_and_test(&ns->__ns_ref);
+	if (refcount_dec_and_test(&ns->__ns_ref)) {
+		VFS_WARN_ON_ONCE(__ns_ref_active_read(ns));
+		return true;
+	}
+	return false;
 }
 
 static __always_inline __must_check bool __ns_ref_get(struct ns_common *ns)
 {
-	return refcount_inc_not_zero(&ns->__ns_ref);
+	if (refcount_inc_not_zero(&ns->__ns_ref))
+		return true;
+	VFS_WARN_ON_ONCE(__ns_ref_active_read(ns));
+	return false;
 }
 
 static __always_inline __must_check int __ns_ref_read(const struct ns_common *ns)
@@ -166,4 +250,57 @@ static __always_inline __must_check int __ns_ref_read(const struct ns_common *ns
 #define ns_ref_put_and_lock(__ns, __lock) \
 	refcount_dec_and_lock(&to_ns_common((__ns))->__ns_ref, (__lock))
 
+#define ns_ref_active_read(__ns) \
+	((__ns) ? __ns_ref_active_read(to_ns_common(__ns)) : 0)
+
+void __ns_ref_active_get_owner(struct ns_common *ns);
+
+static __always_inline void __ns_ref_active_get(struct ns_common *ns)
+{
+	WARN_ON_ONCE(atomic_add_negative(1, &ns->__ns_ref_active));
+	VFS_WARN_ON_ONCE(is_initial_namespace(ns) && __ns_ref_active_read(ns) <= 0);
+}
+#define ns_ref_active_get(__ns) \
+	do { if (__ns) __ns_ref_active_get(to_ns_common(__ns)); } while (0)
+
+static __always_inline bool __ns_ref_active_get_not_zero(struct ns_common *ns)
+{
+	if (atomic_inc_not_zero(&ns->__ns_ref_active)) {
+		VFS_WARN_ON_ONCE(!__ns_ref_read(ns));
+		return true;
+	}
+	return false;
+}
+
+#define ns_ref_active_get_owner(__ns) \
+	do { if (__ns) __ns_ref_active_get_owner(to_ns_common(__ns)); } while (0)
+
+void __ns_ref_active_put_owner(struct ns_common *ns);
+
+static __always_inline void __ns_ref_active_put(struct ns_common *ns)
+{
+	if (atomic_dec_and_test(&ns->__ns_ref_active)) {
+		VFS_WARN_ON_ONCE(is_initial_namespace(ns));
+		VFS_WARN_ON_ONCE(!__ns_ref_read(ns));
+		__ns_ref_active_put_owner(ns);
+	}
+}
+#define ns_ref_active_put(__ns) \
+	do { if (__ns) __ns_ref_active_put(to_ns_common(__ns)); } while (0)
+
+static __always_inline struct ns_common *__must_check ns_get_unless_inactive(struct ns_common *ns)
+{
+	VFS_WARN_ON_ONCE(__ns_ref_active_read(ns) && !__ns_ref_read(ns));
+	if (!__ns_ref_active_read(ns))
+		return NULL;
+	if (!__ns_ref_get(ns))
+		return NULL;
+	return ns;
+}
+
+void __ns_ref_active_resurrect(struct ns_common *ns);
+
+#define ns_ref_active_resurrect(__ns) \
+	do { if (__ns) __ns_ref_active_resurrect(to_ns_common(__ns)); } while (0)
+
 #endif
diff --git a/include/linux/nsfs.h b/include/linux/nsfs.h
index e5a5fa83d36b..731b67fc2fec 100644
--- a/include/linux/nsfs.h
+++ b/include/linux/nsfs.h
@@ -37,4 +37,7 @@ void nsfs_init(void);
 
 #define current_in_namespace(__ns) (__current_namespace_from_type(__ns) == __ns)
 
+void nsproxy_ns_active_get(struct nsproxy *ns);
+void nsproxy_ns_active_put(struct nsproxy *ns);
+
 #endif /* _LINUX_NSFS_H */
diff --git a/include/linux/nsproxy.h b/include/linux/nsproxy.h
index 538ba8dba184..ac825eddec59 100644
--- a/include/linux/nsproxy.h
+++ b/include/linux/nsproxy.h
@@ -93,7 +93,10 @@ static inline struct cred *nsset_cred(struct nsset *set)
  */
 
 int copy_namespaces(u64 flags, struct task_struct *tsk);
+void switch_cred_namespaces(const struct cred *old, const struct cred *new);
 void exit_nsproxy_namespaces(struct task_struct *tsk);
+void get_cred_namespaces(struct task_struct *tsk);
+void exit_cred_namespaces(struct task_struct *tsk);
 void switch_task_namespaces(struct task_struct *tsk, struct nsproxy *new);
 int exec_task_namespaces(void);
 void free_nsproxy(struct nsproxy *ns);