cgroup: fix cgroup_sk_alloc() for sk_clone_lock()

When we clone a socket in sk_clone_lock(), its sk_cgrp_data is
copied, so the cgroup refcnt must be taken too. And, unlike the
sk_alloc() path, sock_update_netprioidx() is not called here.
Therefore, it is safe and necessary to grab the cgroup refcnt
even when cgroup_sk_alloc is disabled.

sk_clone_lock() is in BH context anyway, the in_interrupt()
would terminate this function if called there. And for sk_alloc()
skcd->val is always zero. So it's safe to factor out the code
to make it more readable.

The global variable 'cgroup_sk_alloc_disabled' is used to determine
whether to take these reference counts. It is impossible to make
the reference counting correct unless we save this bit of information
in skcd->val. So, add a new bit there to record whether the socket
has already taken the reference counts. This obviously relies on
kmalloc() to align cgroup pointers to at least 4 bytes,
ARCH_KMALLOC_MINALIGN is certainly larger than that.

This bug seems to be introduced since the beginning, commit
d979a39 ("cgroup: duplicate cgroup reference when cloning sockets")
tried to fix it but not compeletely. It seems not easy to trigger until
the recent commit 090e28b
("netprio_cgroup: Fix unlimited memory leak of v2 cgroups") was merged.

Fixes: bd1060a ("sock, cgroup: add sock->sk_cgroup")
Reported-by: Cameron Berkenpas <[email protected]>
Reported-by: Peter Geis <[email protected]>
Reported-by: Lu Fengqi <[email protected]>
Reported-by: Daniël Sonck <[email protected]>
Reported-by: Zhang Qiang <[email protected]>
Tested-by: Cameron Berkenpas <[email protected]>
Tested-by: Peter Geis <[email protected]>
Tested-by: Thomas Lamprecht <[email protected]>
Cc: Daniel Borkmann <[email protected]>
Cc: Zefan Li <[email protected]>
Cc: Tejun Heo <[email protected]>
Cc: Roman Gushchin <[email protected]>
Signed-off-by: Cong Wang <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

include/linux/cgroup-defs.h

@@ -790,7 +790,8 @@ struct sock_cgroup_data {
 	union {
 #ifdef __LITTLE_ENDIAN
 		struct {
-			u8	is_data;
+			u8	is_data : 1;
+			u8	no_refcnt : 1;
 			u8	padding;
 			u16	prioidx;
 			u32	classid;
@@ -800,7 +801,8 @@ struct sock_cgroup_data {
 			u32	classid;
 			u16	prioidx;
 			u8	padding;
-			u8	is_data;
+			u8	no_refcnt : 1;
+			u8	is_data : 1;
 		} __packed;
 #endif
 		u64		val;

include/linux/cgroup.h

@@ -822,6 +822,7 @@ extern spinlock_t cgroup_sk_update_lock;
 
 void cgroup_sk_alloc_disable(void);
 void cgroup_sk_alloc(struct sock_cgroup_data *skcd);
+void cgroup_sk_clone(struct sock_cgroup_data *skcd);
 void cgroup_sk_free(struct sock_cgroup_data *skcd);
 
 static inline struct cgroup *sock_cgroup_ptr(struct sock_cgroup_data *skcd)
@@ -835,7 +836,7 @@ static inline struct cgroup *sock_cgroup_ptr(struct sock_cgroup_data *skcd)
 	 */
 	v = READ_ONCE(skcd->val);
 
-	if (v & 1)
+	if (v & 3)
 		return &cgrp_dfl_root.cgrp;
 
 	return (struct cgroup *)(unsigned long)v ?: &cgrp_dfl_root.cgrp;
@@ -847,6 +848,7 @@ static inline struct cgroup *sock_cgroup_ptr(struct sock_cgroup_data *skcd)
 #else	/* CONFIG_CGROUP_DATA */
 
 static inline void cgroup_sk_alloc(struct sock_cgroup_data *skcd) {}
+static inline void cgroup_sk_clone(struct sock_cgroup_data *skcd) {}
 static inline void cgroup_sk_free(struct sock_cgroup_data *skcd) {}
 
 #endif	/* CONFIG_CGROUP_DATA */

kernel/cgroup/cgroup.c

@@ -6439,18 +6439,8 @@ void cgroup_sk_alloc_disable(void)
 
 void cgroup_sk_alloc(struct sock_cgroup_data *skcd)
 {
-	if (cgroup_sk_alloc_disabled)
-		return;
-
-	/* Socket clone path */
-	if (skcd->val) {
-		/*
-		 * We might be cloning a socket which is left in an empty
-		 * cgroup and the cgroup might have already been rmdir'd.
-		 * Don't use cgroup_get_live().
-		 */
-		cgroup_get(sock_cgroup_ptr(skcd));
-		cgroup_bpf_get(sock_cgroup_ptr(skcd));
+	if (cgroup_sk_alloc_disabled) {
+		skcd->no_refcnt = 1;
 		return;
 	}
 
@@ -6475,10 +6465,27 @@ void cgroup_sk_alloc(struct sock_cgroup_data *skcd)
 	rcu_read_unlock();
 }
 
+void cgroup_sk_clone(struct sock_cgroup_data *skcd)
+{
+	if (skcd->val) {
+		if (skcd->no_refcnt)
+			return;
+		/*
+		 * We might be cloning a socket which is left in an empty
+		 * cgroup and the cgroup might have already been rmdir'd.
+		 * Don't use cgroup_get_live().
+		 */
+		cgroup_get(sock_cgroup_ptr(skcd));
+		cgroup_bpf_get(sock_cgroup_ptr(skcd));
+	}
+}
+
 void cgroup_sk_free(struct sock_cgroup_data *skcd)
 {
 	struct cgroup *cgrp = sock_cgroup_ptr(skcd);
 
+	if (skcd->no_refcnt)
+		return;
 	cgroup_bpf_put(cgrp);
 	cgroup_put(cgrp);
 }

net/core/sock.c

@@ -1926,7 +1926,7 @@ struct sock *sk_clone_lock(const struct sock *sk, const gfp_t priority)
 		/* sk->sk_memcg will be populated at accept() time */
 		newsk->sk_memcg = NULL;
 
-		cgroup_sk_alloc(&newsk->sk_cgrp_data);
+		cgroup_sk_clone(&newsk->sk_cgrp_data);
 
 		rcu_read_lock();
 		filter = rcu_dereference(sk->sk_filter);