soreuseport: setsockopt SO_ATTACH_REUSEPORT_[CE]BPF

Expose socket options for setting a classic or extended BPF program
for use when selecting sockets in an SO_REUSEPORT group.  These options
can be used on the first socket to belong to a group before bind or
on any socket in the group after bind.

This change includes refactoring of the existing sk_filter code to
allow reuse of the existing BPF filter validation checks.

Signed-off-by: Craig Gallek <kraig@google.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

3 files changed, 211 insertions, 27 deletions

diff --git a/net/core/filter.c b/net/core/filter.c
index c770196ae8d5..35e6fed28709 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -50,6 +50,7 @@
 #include <net/cls_cgroup.h>
 #include <net/dst_metadata.h>
 #include <net/dst.h>
+#include <net/sock_reuseport.h>
 
 /**
  *	sk_filter - run a packet through a socket filter
@@ -1167,17 +1168,32 @@ static int __sk_attach_prog(struct bpf_prog *prog, struct sock *sk)
 	return 0;
 }
 
-/**
- *	sk_attach_filter - attach a socket filter
- *	@fprog: the filter program
- *	@sk: the socket to use
- *
- * Attach the user's filter code. We first run some sanity checks on
- * it to make sure it does not explode on us later. If an error
- * occurs or there is insufficient memory for the filter a negative
- * errno code is returned. On success the return is zero.
- */
-int sk_attach_filter(struct sock_fprog *fprog, struct sock *sk)
+static int __reuseport_attach_prog(struct bpf_prog *prog, struct sock *sk)
+{
+	struct bpf_prog *old_prog;
+	int err;
+
+	if (bpf_prog_size(prog->len) > sysctl_optmem_max)
+		return -ENOMEM;
+
+	if (sk_unhashed(sk)) {
+		err = reuseport_alloc(sk);
+		if (err)
+			return err;
+	} else if (!rcu_access_pointer(sk->sk_reuseport_cb)) {
+		/* The socket wasn't bound with SO_REUSEPORT */
+		return -EINVAL;
+	}
+
+	old_prog = reuseport_attach_prog(sk, prog);
+	if (old_prog)
+		bpf_prog_destroy(old_prog);
+
+	return 0;
+}
+
+static
+struct bpf_prog *__get_filter(struct sock_fprog *fprog, struct sock *sk)
 {
 	unsigned int fsize = bpf_classic_proglen(fprog);
 	unsigned int bpf_fsize = bpf_prog_size(fprog->len);
@@ -1185,19 +1201,19 @@ int sk_attach_filter(struct sock_fprog *fprog, struct sock *sk)
 	int err;
 
 	if (sock_flag(sk, SOCK_FILTER_LOCKED))
-		return -EPERM;
+		return ERR_PTR(-EPERM);
 
 	/* Make sure new filter is there and in the right amounts. */
 	if (fprog->filter == NULL)
-		return -EINVAL;
+		return ERR_PTR(-EINVAL);
 
 	prog = bpf_prog_alloc(bpf_fsize, 0);
 	if (!prog)
-		return -ENOMEM;
+		return ERR_PTR(-ENOMEM);
 
 	if (copy_from_user(prog->insns, fprog->filter, fsize)) {
 		__bpf_prog_free(prog);
-		return -EFAULT;
+		return ERR_PTR(-EFAULT);
 	}
 
 	prog->len = fprog->len;
@@ -1205,13 +1221,30 @@ int sk_attach_filter(struct sock_fprog *fprog, struct sock *sk)
 	err = bpf_prog_store_orig_filter(prog, fprog);
 	if (err) {
 		__bpf_prog_free(prog);
-		return -ENOMEM;
+		return ERR_PTR(-ENOMEM);
 	}
 
 	/* bpf_prepare_filter() already takes care of freeing
 	 * memory in case something goes wrong.
 	 */
-	prog = bpf_prepare_filter(prog, NULL);
+	return bpf_prepare_filter(prog, NULL);
+}
+
+/**
+ *	sk_attach_filter - attach a socket filter
+ *	@fprog: the filter program
+ *	@sk: the socket to use
+ *
+ * Attach the user's filter code. We first run some sanity checks on
+ * it to make sure it does not explode on us later. If an error
+ * occurs or there is insufficient memory for the filter a negative
+ * errno code is returned. On success the return is zero.
+ */
+int sk_attach_filter(struct sock_fprog *fprog, struct sock *sk)
+{
+	struct bpf_prog *prog = __get_filter(fprog, sk);
+	int err;
+
 	if (IS_ERR(prog))
 		return PTR_ERR(prog);
 
@@ -1225,23 +1258,50 @@ int sk_attach_filter(struct sock_fprog *fprog, struct sock *sk)
 }
 EXPORT_SYMBOL_GPL(sk_attach_filter);
 
-int sk_attach_bpf(u32 ufd, struct sock *sk)
+int sk_reuseport_attach_filter(struct sock_fprog *fprog, struct sock *sk)
 {
-	struct bpf_prog *prog;
+	struct bpf_prog *prog = __get_filter(fprog, sk);
 	int err;
 
+	if (IS_ERR(prog))
+		return PTR_ERR(prog);
+
+	err = __reuseport_attach_prog(prog, sk);
+	if (err < 0) {
+		__bpf_prog_release(prog);
+		return err;
+	}
+
+	return 0;
+}
+
+static struct bpf_prog *__get_bpf(u32 ufd, struct sock *sk)
+{
+	struct bpf_prog *prog;
+
 	if (sock_flag(sk, SOCK_FILTER_LOCKED))
-		return -EPERM;
+		return ERR_PTR(-EPERM);
 
 	prog = bpf_prog_get(ufd);
 	if (IS_ERR(prog))
-		return PTR_ERR(prog);
+		return prog;
 
 	if (prog->type != BPF_PROG_TYPE_SOCKET_FILTER) {
 		bpf_prog_put(prog);
-		return -EINVAL;
+		return ERR_PTR(-EINVAL);
 	}
 
+	return prog;
+}
+
+int sk_attach_bpf(u32 ufd, struct sock *sk)
+{
+	struct bpf_prog *prog = __get_bpf(ufd, sk);
+	int err;
+
+	if (IS_ERR(prog))
+		return PTR_ERR(prog);
+
 	err = __sk_attach_prog(prog, sk);
 	if (err < 0) {
 		bpf_prog_put(prog);
@@ -1251,6 +1311,23 @@ int sk_attach_bpf(u32 ufd, struct sock *sk)
 	return 0;
 }
 
+int sk_reuseport_attach_bpf(u32 ufd, struct sock *sk)
+{
+	struct bpf_prog *prog = __get_bpf(ufd, sk);
+	int err;
+
+	if (IS_ERR(prog))
+		return PTR_ERR(prog);
+
+	err = __reuseport_attach_prog(prog, sk);
+	if (err < 0) {
+		bpf_prog_put(prog);
+		return err;
+	}
+
+	return 0;
+}
+
 #define BPF_RECOMPUTE_CSUM(flags)	((flags) & 1)
 #define BPF_LDST_LEN			16U
 
diff --git a/net/core/sock.c b/net/core/sock.c
index 565bab7baca9..51270238e269 100644
--- a/net/core/sock.c
+++ b/net/core/sock.c
@@ -134,6 +134,7 @@
 #include <linux/sock_diag.h>
 
 #include <linux/filter.h>
+#include <net/sock_reuseport.h>
 
 #include <trace/events/sock.h>
 
@@ -932,6 +933,32 @@ set_rcvbuf:
 		}
 		break;
 
+	case SO_ATTACH_REUSEPORT_CBPF:
+		ret = -EINVAL;
+		if (optlen == sizeof(struct sock_fprog)) {
+			struct sock_fprog fprog;
+
+			ret = -EFAULT;
+			if (copy_from_user(&fprog, optval, sizeof(fprog)))
+				break;
+
+			ret = sk_reuseport_attach_filter(&fprog, sk);
+		}
+		break;
+
+	case SO_ATTACH_REUSEPORT_EBPF:
+		ret = -EINVAL;
+		if (optlen == sizeof(u32)) {
+			u32 ufd;
+
+			ret = -EFAULT;
+			if (copy_from_user(&ufd, optval, sizeof(ufd)))
+				break;
+
+			ret = sk_reuseport_attach_bpf(ufd, sk);
+		}
+		break;
+
 	case SO_DETACH_FILTER:
 		ret = sk_detach_filter(sk);
 		break;
@@ -1443,6 +1470,8 @@ void sk_destruct(struct sock *sk)
 		sk_filter_uncharge(sk, filter);
 		RCU_INIT_POINTER(sk->sk_filter, NULL);
 	}
+	if (rcu_access_pointer(sk->sk_reuseport_cb))
+		reuseport_detach_sock(sk);
 
 	sock_disable_timestamp(sk, SK_FLAGS_TIMESTAMP);
 
diff --git a/net/core/sock_reuseport.c b/net/core/sock_reuseport.c
index 963c8d5f3027..ae0969c0fc2e 100644
--- a/net/core/sock_reuseport.c
+++ b/net/core/sock_reuseport.c
@@ -1,10 +1,12 @@
 /*
  * To speed up listener socket lookup, create an array to store all sockets
  * listening on the same port.  This allows a decision to be made after finding
- * the first socket.
+ * the first socket.  An optional BPF program can also be configured for
+ * selecting the socket index from the array of available sockets.
  */
 
 #include <net/sock_reuseport.h>
+#include <linux/bpf.h>
 #include <linux/rcupdate.h>
 
 #define INIT_SOCKS 128
@@ -22,6 +24,7 @@ static struct sock_reuseport *__reuseport_alloc(u16 max_socks)
 
 	reuse->max_socks = max_socks;
 
+	RCU_INIT_POINTER(reuse->prog, NULL);
 	return reuse;
 }
 
@@ -67,6 +70,7 @@ static struct sock_reuseport *reuseport_grow(struct sock_reuseport *reuse)
 
 	more_reuse->max_socks = more_socks_size;
 	more_reuse->num_socks = reuse->num_socks;
+	more_reuse->prog = reuse->prog;
 
 	memcpy(more_reuse->socks, reuse->socks,
 	       reuse->num_socks * sizeof(struct sock *));
@@ -75,6 +79,10 @@ static struct sock_reuseport *reuseport_grow(struct sock_reuseport *reuse)
 		rcu_assign_pointer(reuse->socks[i]->sk_reuseport_cb,
 				   more_reuse);
 
+	/* Note: we use kfree_rcu here instead of reuseport_free_rcu so
+	 * that reuse and more_reuse can temporarily share a reference
+	 * to prog.
+	 */
 	kfree_rcu(reuse, rcu);
 	return more_reuse;
 }
@@ -116,6 +124,16 @@ int reuseport_add_sock(struct sock *sk, const struct sock *sk2)
 }
 EXPORT_SYMBOL(reuseport_add_sock);
 
+static void reuseport_free_rcu(struct rcu_head *head)
+{
+	struct sock_reuseport *reuse;
+
+	reuse = container_of(head, struct sock_reuseport, rcu);
+	if (reuse->prog)
+		bpf_prog_destroy(reuse->prog);
+	kfree(reuse);
+}
+
 void reuseport_detach_sock(struct sock *sk)
 {
 	struct sock_reuseport *reuse;
@@ -131,7 +149,7 @@ void reuseport_detach_sock(struct sock *sk)
 			reuse->socks[i] = reuse->socks[reuse->num_socks - 1];
 			reuse->num_socks--;
 			if (reuse->num_socks == 0)
-				kfree_rcu(reuse, rcu);
+				call_rcu(&reuse->rcu, reuseport_free_rcu);
 			break;
 		}
 	}
@@ -139,15 +157,53 @@ void reuseport_detach_sock(struct sock *sk)
 }
 EXPORT_SYMBOL(reuseport_detach_sock);
 
+static struct sock *run_bpf(struct sock_reuseport *reuse, u16 socks,
+			    struct bpf_prog *prog, struct sk_buff *skb,
+			    int hdr_len)
+{
+	struct sk_buff *nskb = NULL;
+	u32 index;
+
+	if (skb_shared(skb)) {
+		nskb = skb_clone(skb, GFP_ATOMIC);
+		if (!nskb)
+			return NULL;
+		skb = nskb;
+	}
+
+	/* temporarily advance data past protocol header */
+	if (!pskb_pull(skb, hdr_len)) {
+		consume_skb(nskb);
+		return NULL;
+	}
+	index = bpf_prog_run_save_cb(prog, skb);
+	__skb_push(skb, hdr_len);
+
+	consume_skb(nskb);
+
+	if (index >= socks)
+		return NULL;
+
+	return reuse->socks[index];
+}
+
 /**
  *  reuseport_select_sock - Select a socket from an SO_REUSEPORT group.
  *  @sk: First socket in the group.
- *  @hash: Use this hash to select.
+ *  @hash: When no BPF filter is available, use this hash to select.
+ *  @skb: skb to run through BPF filter.
+ *  @hdr_len: BPF filter expects skb data pointer at payload data.  If
+ *    the skb does not yet point at the payload, this parameter represents
+ *    how far the pointer needs to advance to reach the payload.
  *  Returns a socket that should receive the packet (or NULL on error).
  */
-struct sock *reuseport_select_sock(struct sock *sk, u32 hash)
+struct sock *reuseport_select_sock(struct sock *sk,
+				   u32 hash,
+				   struct sk_buff *skb,
+				   int hdr_len)
 {
 	struct sock_reuseport *reuse;
+	struct bpf_prog *prog;
 	struct sock *sk2 = NULL;
 	u16 socks;
 
@@ -158,12 +214,16 @@ struct sock *reuseport_select_sock(struct sock *sk, u32 hash)
 	if (!reuse)
 		goto out;
 
+	prog = rcu_dereference(reuse->prog);
 	socks = READ_ONCE(reuse->num_socks);
 	if (likely(socks)) {
 		/* paired with smp_wmb() in reuseport_add_sock() */
 		smp_rmb();
 
-		sk2 = reuse->socks[reciprocal_scale(hash, socks)];
+		if (prog && skb)
+			sk2 = run_bpf(reuse, socks, prog, skb, hdr_len);
+		else
+			sk2 = reuse->socks[reciprocal_scale(hash, socks)];
 	}
 
 out:
@@ -171,3 +231,21 @@ out:
 	return sk2;
 }
 EXPORT_SYMBOL(reuseport_select_sock);
+
+struct bpf_prog *
+reuseport_attach_prog(struct sock *sk, struct bpf_prog *prog)
+{
+	struct sock_reuseport *reuse;
+	struct bpf_prog *old_prog;
+
+	spin_lock_bh(&reuseport_lock);
+	reuse = rcu_dereference_protected(sk->sk_reuseport_cb,
+					  lockdep_is_held(&reuseport_lock));
+	old_prog = rcu_dereference_protected(reuse->prog,
+					     lockdep_is_held(&reuseport_lock));
+	rcu_assign_pointer(reuse->prog, prog);
+	spin_unlock_bh(&reuseport_lock);
+
+	return old_prog;
+}
+EXPORT_SYMBOL(reuseport_attach_prog);