netfilter: nft_socket: make cgroupsv2 matching work with namespaces

When running in container environmment, /sys/fs/cgroup/ might not be
the real root node of the sk-attached cgroup.

Example:

In container:
% stat /sys//fs/cgroup/
Device: 0,21    Inode: 2214  ..
% stat /sys/fs/cgroup/foo
Device: 0,21    Inode: 2264  ..

The expectation would be for:

  nft add rule .. socket cgroupv2 level 1 "foo" counter

to match traffic from a process that got added to "foo" via
"echo $pid > /sys/fs/cgroup/foo/cgroup.procs".

However, 'level 3' is needed to make this work.

Seen from initial namespace, the complete hierarchy is:

% stat /sys/fs/cgroup/system.slice/docker-.../foo
  Device: 0,21    Inode: 2264 ..

i.e. hierarchy is
0    1               2              3
/ -> system.slice -> docker-1... -> foo

... but the container doesn't know that its "/" is the "docker-1.."
cgroup.  Current code will retrieve the 'system.slice' cgroup node
and store its kn->id in the destination register, so compare with
2264 ("foo" cgroup id) will not match.

Fetch "/" cgroup from ->init() and add its level to the level we try to
extract.  cgroup root-level is 0 for the init-namespace or the level
of the ancestor that is exposed as the cgroup root inside the container.

In the above case, cgrp->level of "/" resolved in the container is 2
(docker-1...scope/) and request for 'level 1' will get adjusted
to fetch the actual level (3).

v2: use CONFIG_SOCK_CGROUP_DATA, eval function depends on it.
    (kernel test robot)

Cc: cgroups@vger.kernel.org
Fixes: e0bb96db96 ("netfilter: nft_socket: add support for cgroupsv2")
Reported-by: Nadia Pinaeva <n.m.pinaeva@gmail.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>

net/netfilter/nft_socket.c

@@ -9,7 +9,8 @@
 
 struct nft_socket {
 	enum nft_socket_keys		key:8;
-	u8				level;
+	u8				level;		/* cgroupv2 level to extract */
+	u8				level_user;	/* cgroupv2 level provided by userspace */
 	u8				len;
 	union {
 		u8			dreg;
@@ -53,6 +54,28 @@ nft_sock_get_eval_cgroupv2(u32 *dest, struct sock *sk, const struct nft_pktinfo
 	memcpy(dest, &cgid, sizeof(u64));
 	return true;
 }
+
+/* process context only, uses current->nsproxy. */
+static noinline int nft_socket_cgroup_subtree_level(void)
+{
+	struct cgroup *cgrp = cgroup_get_from_path("/");
+	int level;
+
+	if (!cgrp)
+		return -ENOENT;
+
+	level = cgrp->level;
+
+	cgroup_put(cgrp);
+
+	if (WARN_ON_ONCE(level > 255))
+		return -ERANGE;
+
+	if (WARN_ON_ONCE(level < 0))
+		return -EINVAL;
+
+	return level;
+}
 #endif
 
 static struct sock *nft_socket_do_lookup(const struct nft_pktinfo *pkt)
@@ -174,9 +197,10 @@ static int nft_socket_init(const struct nft_ctx *ctx,
 	case NFT_SOCKET_MARK:
 		len = sizeof(u32);
 		break;
-#ifdef CONFIG_CGROUPS
+#ifdef CONFIG_SOCK_CGROUP_DATA
 	case NFT_SOCKET_CGROUPV2: {
 		unsigned int level;
+		int err;
 
 		if (!tb[NFTA_SOCKET_LEVEL])
 			return -EINVAL;
@@ -185,6 +209,17 @@ static int nft_socket_init(const struct nft_ctx *ctx,
 		if (level > 255)
 			return -EOPNOTSUPP;
 
+		err = nft_socket_cgroup_subtree_level();
+		if (err < 0)
+			return err;
+
+		priv->level_user = level;
+
+		level += err;
+		/* Implies a giant cgroup tree */
+		if (WARN_ON_ONCE(level > 255))
+			return -EOPNOTSUPP;
+
 		priv->level = level;
 		len = sizeof(u64);
 		break;
@@ -209,7 +244,7 @@ static int nft_socket_dump(struct sk_buff *skb,
 	if (nft_dump_register(skb, NFTA_SOCKET_DREG, priv->dreg))
 		return -1;
 	if (priv->key == NFT_SOCKET_CGROUPV2 &&
-	    nla_put_be32(skb, NFTA_SOCKET_LEVEL, htonl(priv->level)))
+	    nla_put_be32(skb, NFTA_SOCKET_LEVEL, htonl(priv->level_user)))
 		return -1;
 	return 0;
 }