Merge branch 'bpf: expose bpf_{s,g}etsockopt helpers to bind{4,6} hooks'

Alexei Starovoitov · Alexei Starovoitov · commit 9e83f54f532b · 2020-12-02T13:25:25.000-08:00
Stanislav Fomichev says:

====================

This might be useful for the listener sockets to pre-populate
some options. Since those helpers require locked sockets,
I'm changing bind hooks to lock/unlock the sockets. This
should not cause any performance overhead because at this
point there shouldn't be any socket lock contention and the
locking/unlocking should be cheap.

Also, as part of the series, I convert test_sock_addr bpf
assembly into C (and preserve the narrow load tests) to
make it easier to extend with th bpf_setsockopt later on.

v2:
* remove version from bpf programs (Andrii Nakryiko)
====================

Signed-off-by: Alexei Starovoitov &lt;ast@kernel.org&gt;
diff --git a/include/linux/bpf-cgroup.h b/include/linux/bpf-cgroup.h
@@ -246,11 +246,11 @@ int bpf_percpu_cgroup_storage_update(struct bpf_map *map, void *key,
 	__ret;								       \
 })
 
-#define BPF_CGROUP_RUN_PROG_INET4_BIND(sk, uaddr)			       \
-	BPF_CGROUP_RUN_SA_PROG(sk, uaddr, BPF_CGROUP_INET4_BIND)
+#define BPF_CGROUP_RUN_PROG_INET4_BIND_LOCK(sk, uaddr)			       \
+	BPF_CGROUP_RUN_SA_PROG_LOCK(sk, uaddr, BPF_CGROUP_INET4_BIND, NULL)
 
-#define BPF_CGROUP_RUN_PROG_INET6_BIND(sk, uaddr)			       \
-	BPF_CGROUP_RUN_SA_PROG(sk, uaddr, BPF_CGROUP_INET6_BIND)
+#define BPF_CGROUP_RUN_PROG_INET6_BIND_LOCK(sk, uaddr)			       \
+	BPF_CGROUP_RUN_SA_PROG_LOCK(sk, uaddr, BPF_CGROUP_INET6_BIND, NULL)
 
 #define BPF_CGROUP_PRE_CONNECT_ENABLED(sk) (cgroup_bpf_enabled && \
 					    sk->sk_prot->pre_connect)
@@ -434,8 +434,8 @@ static inline int bpf_percpu_cgroup_storage_update(struct bpf_map *map,
 #define BPF_CGROUP_RUN_PROG_INET_EGRESS(sk,skb) ({ 0; })
 #define BPF_CGROUP_RUN_PROG_INET_SOCK(sk) ({ 0; })
 #define BPF_CGROUP_RUN_PROG_INET_SOCK_RELEASE(sk) ({ 0; })
-#define BPF_CGROUP_RUN_PROG_INET4_BIND(sk, uaddr) ({ 0; })
-#define BPF_CGROUP_RUN_PROG_INET6_BIND(sk, uaddr) ({ 0; })
+#define BPF_CGROUP_RUN_PROG_INET4_BIND_LOCK(sk, uaddr) ({ 0; })
+#define BPF_CGROUP_RUN_PROG_INET6_BIND_LOCK(sk, uaddr) ({ 0; })
 #define BPF_CGROUP_RUN_PROG_INET4_POST_BIND(sk) ({ 0; })
 #define BPF_CGROUP_RUN_PROG_INET6_POST_BIND(sk) ({ 0; })
 #define BPF_CGROUP_RUN_PROG_INET4_CONNECT(sk, uaddr) ({ 0; })
diff --git a/net/core/filter.c b/net/core/filter.c
@@ -6995,6 +6995,8 @@ sock_addr_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		return &bpf_sk_storage_delete_proto;
 	case BPF_FUNC_setsockopt:
 		switch (prog->expected_attach_type) {
+		case BPF_CGROUP_INET4_BIND:
+		case BPF_CGROUP_INET6_BIND:
 		case BPF_CGROUP_INET4_CONNECT:
 		case BPF_CGROUP_INET6_CONNECT:
 			return &bpf_sock_addr_setsockopt_proto;
@@ -7003,6 +7005,8 @@ sock_addr_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 		}
 	case BPF_FUNC_getsockopt:
 		switch (prog->expected_attach_type) {
+		case BPF_CGROUP_INET4_BIND:
+		case BPF_CGROUP_INET6_BIND:
 		case BPF_CGROUP_INET4_CONNECT:
 		case BPF_CGROUP_INET6_CONNECT:
 			return &bpf_sock_addr_getsockopt_proto;
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
@@ -450,7 +450,7 @@ int inet_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
 	/* BPF prog is run before any checks are done so that if the prog
 	 * changes context in a wrong way it will be caught.
 	 */
-	err = BPF_CGROUP_RUN_PROG_INET4_BIND(sk, uaddr);
+	err = BPF_CGROUP_RUN_PROG_INET4_BIND_LOCK(sk, uaddr);
 	if (err)
 		return err;
 
diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
@@ -451,7 +451,7 @@ int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
 	/* BPF prog is run before any checks are done so that if the prog
 	 * changes context in a wrong way it will be caught.
 	 */
-	err = BPF_CGROUP_RUN_PROG_INET6_BIND(sk, uaddr);
+	err = BPF_CGROUP_RUN_PROG_INET6_BIND_LOCK(sk, uaddr);
 	if (err)
 		return err;
 
diff --git a/tools/testing/selftests/bpf/progs/bind4_prog.c b/tools/testing/selftests/bpf/progs/bind4_prog.c
@@ -0,0 +1,102 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <string.h>
+
+#include <linux/stddef.h>
+#include <linux/bpf.h>
+#include <linux/in.h>
+#include <linux/in6.h>
+#include <sys/socket.h>
+#include <netinet/tcp.h>
+#include <linux/if.h>
+#include <errno.h>
+
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_endian.h>
+
+#define SERV4_IP		0xc0a801feU /* 192.168.1.254 */
+#define SERV4_PORT		4040
+#define SERV4_REWRITE_IP	0x7f000001U /* 127.0.0.1 */
+#define SERV4_REWRITE_PORT	4444
+
+#ifndef IFNAMSIZ
+#define IFNAMSIZ 16
+#endif
+
+static __inline int bind_to_device(struct bpf_sock_addr *ctx)
+{
+	char veth1[IFNAMSIZ] = "test_sock_addr1";
+	char veth2[IFNAMSIZ] = "test_sock_addr2";
+	char missing[IFNAMSIZ] = "nonexistent_dev";
+	char del_bind[IFNAMSIZ] = "";
+
+	if (bpf_setsockopt(ctx, SOL_SOCKET, SO_BINDTODEVICE,
+				&veth1, sizeof(veth1)))
+		return 1;
+	if (bpf_setsockopt(ctx, SOL_SOCKET, SO_BINDTODEVICE,
+				&veth2, sizeof(veth2)))
+		return 1;
+	if (bpf_setsockopt(ctx, SOL_SOCKET, SO_BINDTODEVICE,
+				&missing, sizeof(missing)) != -ENODEV)
+		return 1;
+	if (bpf_setsockopt(ctx, SOL_SOCKET, SO_BINDTODEVICE,
+				&del_bind, sizeof(del_bind)))
+		return 1;
+
+	return 0;
+}
+
+SEC("cgroup/bind4")
+int bind_v4_prog(struct bpf_sock_addr *ctx)
+{
+	struct bpf_sock *sk;
+	__u32 user_ip4;
+	__u16 user_port;
+
+	sk = ctx->sk;
+	if (!sk)
+		return 0;
+
+	if (sk->family != AF_INET)
+		return 0;
+
+	if (ctx->type != SOCK_STREAM && ctx->type != SOCK_DGRAM)
+		return 0;
+
+	if (ctx->user_ip4 != bpf_htonl(SERV4_IP) ||
+	    ctx->user_port != bpf_htons(SERV4_PORT))
+		return 0;
+
+	// u8 narrow loads:
+	user_ip4 = 0;
+	user_ip4 |= ((volatile __u8 *)&ctx->user_ip4)[0] << 0;
+	user_ip4 |= ((volatile __u8 *)&ctx->user_ip4)[1] << 8;
+	user_ip4 |= ((volatile __u8 *)&ctx->user_ip4)[2] << 16;
+	user_ip4 |= ((volatile __u8 *)&ctx->user_ip4)[3] << 24;
+	if (ctx->user_ip4 != user_ip4)
+		return 0;
+
+	user_port = 0;
+	user_port |= ((volatile __u8 *)&ctx->user_port)[0] << 0;
+	user_port |= ((volatile __u8 *)&ctx->user_port)[1] << 8;
+	if (ctx->user_port != user_port)
+		return 0;
+
+	// u16 narrow loads:
+	user_ip4 = 0;
+	user_ip4 |= ((volatile __u16 *)&ctx->user_ip4)[0] << 0;
+	user_ip4 |= ((volatile __u16 *)&ctx->user_ip4)[1] << 16;
+	if (ctx->user_ip4 != user_ip4)
+		return 0;
+
+	/* Bind to device and unbind it. */
+	if (bind_to_device(ctx))
+		return 0;
+
+	ctx->user_ip4 = bpf_htonl(SERV4_REWRITE_IP);
+	ctx->user_port = bpf_htons(SERV4_REWRITE_PORT);
+
+	return 1;
+}
+
+char _license[] SEC("license") = "GPL";
diff --git a/tools/testing/selftests/bpf/progs/bind6_prog.c b/tools/testing/selftests/bpf/progs/bind6_prog.c
@@ -0,0 +1,119 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <string.h>
+
+#include <linux/stddef.h>
+#include <linux/bpf.h>
+#include <linux/in.h>
+#include <linux/in6.h>
+#include <sys/socket.h>
+#include <netinet/tcp.h>
+#include <linux/if.h>
+#include <errno.h>
+
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_endian.h>
+
+#define SERV6_IP_0		0xfaceb00c /* face:b00c:1234:5678::abcd */
+#define SERV6_IP_1		0x12345678
+#define SERV6_IP_2		0x00000000
+#define SERV6_IP_3		0x0000abcd
+#define SERV6_PORT		6060
+#define SERV6_REWRITE_IP_0	0x00000000
+#define SERV6_REWRITE_IP_1	0x00000000
+#define SERV6_REWRITE_IP_2	0x00000000
+#define SERV6_REWRITE_IP_3	0x00000001
+#define SERV6_REWRITE_PORT	6666
+
+#ifndef IFNAMSIZ
+#define IFNAMSIZ 16
+#endif
+
+static __inline int bind_to_device(struct bpf_sock_addr *ctx)
+{
+	char veth1[IFNAMSIZ] = "test_sock_addr1";
+	char veth2[IFNAMSIZ] = "test_sock_addr2";
+	char missing[IFNAMSIZ] = "nonexistent_dev";
+	char del_bind[IFNAMSIZ] = "";
+
+	if (bpf_setsockopt(ctx, SOL_SOCKET, SO_BINDTODEVICE,
+				&veth1, sizeof(veth1)))
+		return 1;
+	if (bpf_setsockopt(ctx, SOL_SOCKET, SO_BINDTODEVICE,
+				&veth2, sizeof(veth2)))
+		return 1;
+	if (bpf_setsockopt(ctx, SOL_SOCKET, SO_BINDTODEVICE,
+				&missing, sizeof(missing)) != -ENODEV)
+		return 1;
+	if (bpf_setsockopt(ctx, SOL_SOCKET, SO_BINDTODEVICE,
+				&del_bind, sizeof(del_bind)))
+		return 1;
+
+	return 0;
+}
+
+SEC("cgroup/bind6")
+int bind_v6_prog(struct bpf_sock_addr *ctx)
+{
+	struct bpf_sock *sk;
+	__u32 user_ip6;
+	__u16 user_port;
+	int i;
+
+	sk = ctx->sk;
+	if (!sk)
+		return 0;
+
+	if (sk->family != AF_INET6)
+		return 0;
+
+	if (ctx->type != SOCK_STREAM && ctx->type != SOCK_DGRAM)
+		return 0;
+
+	if (ctx->user_ip6[0] != bpf_htonl(SERV6_IP_0) ||
+	    ctx->user_ip6[1] != bpf_htonl(SERV6_IP_1) ||
+	    ctx->user_ip6[2] != bpf_htonl(SERV6_IP_2) ||
+	    ctx->user_ip6[3] != bpf_htonl(SERV6_IP_3) ||
+	    ctx->user_port != bpf_htons(SERV6_PORT))
+		return 0;
+
+	// u8 narrow loads:
+	for (i = 0; i < 4; i++) {
+		user_ip6 = 0;
+		user_ip6 |= ((volatile __u8 *)&ctx->user_ip6[i])[0] << 0;
+		user_ip6 |= ((volatile __u8 *)&ctx->user_ip6[i])[1] << 8;
+		user_ip6 |= ((volatile __u8 *)&ctx->user_ip6[i])[2] << 16;
+		user_ip6 |= ((volatile __u8 *)&ctx->user_ip6[i])[3] << 24;
+		if (ctx->user_ip6[i] != user_ip6)
+			return 0;
+	}
+
+	user_port = 0;
+	user_port |= ((volatile __u8 *)&ctx->user_port)[0] << 0;
+	user_port |= ((volatile __u8 *)&ctx->user_port)[1] << 8;
+	if (ctx->user_port != user_port)
+		return 0;
+
+	// u16 narrow loads:
+	for (i = 0; i < 4; i++) {
+		user_ip6 = 0;
+		user_ip6 |= ((volatile __u16 *)&ctx->user_ip6[i])[0] << 0;
+		user_ip6 |= ((volatile __u16 *)&ctx->user_ip6[i])[1] << 16;
+		if (ctx->user_ip6[i] != user_ip6)
+			return 0;
+	}
+
+	/* Bind to device and unbind it. */
+	if (bind_to_device(ctx))
+		return 0;
+
+	ctx->user_ip6[0] = bpf_htonl(SERV6_REWRITE_IP_0);
+	ctx->user_ip6[1] = bpf_htonl(SERV6_REWRITE_IP_1);
+	ctx->user_ip6[2] = bpf_htonl(SERV6_REWRITE_IP_2);
+	ctx->user_ip6[3] = bpf_htonl(SERV6_REWRITE_IP_3);
+	ctx->user_port = bpf_htons(SERV6_REWRITE_PORT);
+
+	return 1;
+}
+
+char _license[] SEC("license") = "GPL";
diff --git a/tools/testing/selftests/bpf/test_sock_addr.c b/tools/testing/selftests/bpf/test_sock_addr.c
