cgroup: avoid accessing modular cgroup subsys structure without locking

subsys[i] is set to NULL in cgroup_unload_subsys() at modular unload,
and that's protected by cgroup_mutex, and then the memory *subsys[i]
resides will be freed.

So this is unsafe without any locking:

  if (!ss || ss->module)
  ...

v2:
- add a comment for enum cgroup_subsys_id
- simplify the comment in cgroup_exit()

Signed-off-by: Li Zefan <lizefan@huawei.com>
Signed-off-by: Tejun Heo <tj@kernel.org>

2 files changed, 28 insertions, 17 deletions

diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h
index 75c6ec1ba1ba..5f76829dd75e 100644
--- a/include/linux/cgroup.h
+++ b/include/linux/cgroup.h
@@ -44,14 +44,25 @@ extern void cgroup_unload_subsys(struct cgroup_subsys *ss);
 
 extern const struct file_operations proc_cgroup_operations;
 
-/* Define the enumeration of all builtin cgroup subsystems */
+/*
+ * Define the enumeration of all cgroup subsystems.
+ *
+ * We define ids for builtin subsystems and then modular ones.
+ */
 #define SUBSYS(_x) _x ## _subsys_id,
-#define IS_SUBSYS_ENABLED(option) IS_ENABLED(option)
 enum cgroup_subsys_id {
+#define IS_SUBSYS_ENABLED(option) IS_BUILTIN(option)
+#include <linux/cgroup_subsys.h>
+#undef IS_SUBSYS_ENABLED
+        CGROUP_BUILTIN_SUBSYS_COUNT,
+
+        __CGROUP_SUBSYS_TEMP_PLACEHOLDER = CGROUP_BUILTIN_SUBSYS_COUNT - 1,
+
+#define IS_SUBSYS_ENABLED(option) IS_MODULE(option)
 #include <linux/cgroup_subsys.h>
+#undef IS_SUBSYS_ENABLED
         CGROUP_SUBSYS_COUNT,
 };
-#undef IS_SUBSYS_ENABLED
 #undef SUBSYS
 
 /* Per-subsystem/per-cgroup state maintained by the system. */
diff --git a/kernel/cgroup.c b/kernel/cgroup.c
index 9df799d5d31c..7a6c4c72ca55 100644
--- a/kernel/cgroup.c
+++ b/kernel/cgroup.c
@@ -4940,17 +4940,17 @@ void cgroup_post_fork(struct task_struct *child)
          * and addition to css_set.
          */
         if (need_forkexit_callback) {
-                for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) {
+                /*
+                 * fork/exit callbacks are supported only for builtin
+                 * subsystems, and the builtin section of the subsys
+                 * array is immutable, so we don't need to lock the
+                 * subsys array here. On the other hand, modular section
+                 * of the array can be freed at module unload, so we
+                 * can't touch that.
+                 */
+                for (i = 0; i < CGROUP_BUILTIN_SUBSYS_COUNT; i++) {
                         struct cgroup_subsys *ss = subsys[i];
 
-                        /*
-                         * fork/exit callbacks are supported only for
-                         * builtin subsystems and we don't need further
-                         * synchronization as they never go away.
-                         */
-                        if (!ss || ss->module)
-                                continue;
-
                         if (ss->fork)
                                 ss->fork(child);
                 }
@@ -5015,13 +5015,13 @@ void cgroup_exit(struct task_struct *tsk, int run_callbacks)
         tsk->cgroups = &init_css_set;
 
         if (run_callbacks && need_forkexit_callback) {
-                for (i = 0; i < CGROUP_SUBSYS_COUNT; i++) {
+                /*
+                 * fork/exit callbacks are supported only for builtin
+                 * subsystems, see cgroup_post_fork() for details.
+                 */
+                for (i = 0; i < CGROUP_BUILTIN_SUBSYS_COUNT; i++) {
                         struct cgroup_subsys *ss = subsys[i];
 
-                        /* modular subsystems can't use callbacks */
-                        if (!ss || ss->module)
-                                continue;
-
                         if (ss->exit) {
                                 struct cgroup *old_cgrp =
                                         rcu_dereference_raw(cg->subsys[i])->cgroup;