Merge branch 'cgroup/connect{4,6} programs for unprivileged ICMP ping'

YiFei Zhu says:

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

Usually when a TCP/UDP connection is initiated, we can bind the socket
to a specific IP attached to an interface in a cgroup/connect hook.
But for pings, this is impossible, as the hook is not being called.

This series adds the invocation for cgroup/connect{4,6} programs to
unprivileged ICMP ping (i.e. ping sockets created with SOCK_DGRAM
IPPROTO_ICMP(V6) as opposed to SOCK_RAW). This also adds a test to
verify that the hooks are being called and invoking bpf_bind() from
within the hook actually binds the socket.

Patch 1 adds the invocation of the hook.
Patch 2 deduplicates write_sysctl in BPF test_progs.
Patch 3 adds the tests for this hook.

v1 -> v2:
* Added static to bindaddr_v6 in prog_tests/connect_ping.c
* Deduplicated much of the test logic in prog_tests/connect_ping.c
* Deduplicated write_sysctl() to test_progs.c

v2 -> v3:
* Renamed variable "obj" to "skel" for the BPF skeleton object in
  prog_tests/connect_ping.c

v3 -> v4:
* Fixed error path to destroy skel in prog_tests/connect_ping.c
====================
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>

net/ipv4/ping.c

@@ -33,6 +33,7 @@
 #include <linux/skbuff.h>
 #include <linux/proc_fs.h>
 #include <linux/export.h>
+#include <linux/bpf-cgroup.h>
 #include <net/sock.h>
 #include <net/ping.h>
 #include <net/udp.h>
@@ -295,6 +296,19 @@ void ping_close(struct sock *sk, long timeout)
 }
 EXPORT_SYMBOL_GPL(ping_close);
 
+static int ping_pre_connect(struct sock *sk, struct sockaddr *uaddr,
+			    int addr_len)
+{
+	/* This check is replicated from __ip4_datagram_connect() and
+	 * intended to prevent BPF program called below from accessing bytes
+	 * that are out of the bound specified by user in addr_len.
+	 */
+	if (addr_len < sizeof(struct sockaddr_in))
+		return -EINVAL;
+
+	return BPF_CGROUP_RUN_PROG_INET4_CONNECT_LOCK(sk, uaddr);
+}
+
 /* Checks the bind address and possibly modifies sk->sk_bound_dev_if. */
 static int ping_check_bind_addr(struct sock *sk, struct inet_sock *isk,
 				struct sockaddr *uaddr, int addr_len)
@@ -1009,6 +1023,7 @@ struct proto ping_prot = {
 	.owner =	THIS_MODULE,
 	.init =		ping_init_sock,
 	.close =	ping_close,
+	.pre_connect =	ping_pre_connect,
 	.connect =	ip4_datagram_connect,
 	.disconnect =	__udp_disconnect,
 	.setsockopt =	ip_setsockopt,

net/ipv6/ping.c

@@ -20,6 +20,7 @@
 #include <net/udp.h>
 #include <net/transp_v6.h>
 #include <linux/proc_fs.h>
+#include <linux/bpf-cgroup.h>
 #include <net/ping.h>
 
 static void ping_v6_destroy(struct sock *sk)
@@ -49,6 +50,20 @@ static int dummy_ipv6_chk_addr(struct net *net, const struct in6_addr *addr,
 	return 0;
 }
 
+static int ping_v6_pre_connect(struct sock *sk, struct sockaddr *uaddr,
+			       int addr_len)
+{
+	/* This check is replicated from __ip6_datagram_connect() and
+	 * intended to prevent BPF program called below from accessing
+	 * bytes that are out of the bound specified by user in addr_len.
+	 */
+
+	if (addr_len < SIN6_LEN_RFC2133)
+		return -EINVAL;
+
+	return BPF_CGROUP_RUN_PROG_INET6_CONNECT_LOCK(sk, uaddr);
+}
+
 static int ping_v6_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
 {
 	struct inet_sock *inet = inet_sk(sk);
@@ -191,6 +206,7 @@ struct proto pingv6_prot = {
 	.init =		ping_init_sock,
 	.close =	ping_close,
 	.destroy =	ping_v6_destroy,
+	.pre_connect =	ping_v6_pre_connect,
 	.connect =	ip6_datagram_connect_v6_only,
 	.disconnect =	__udp_disconnect,
 	.setsockopt =	ipv6_setsockopt,

tools/testing/selftests/bpf/prog_tests/btf_skc_cls_ingress.c

@@ -22,26 +22,6 @@ static __u32 duration;
 
 #define PROG_PIN_FILE "/sys/fs/bpf/btf_skc_cls_ingress"
 
-static int write_sysctl(const char *sysctl, const char *value)
-{
-	int fd, err, len;
-
-	fd = open(sysctl, O_WRONLY);
-	if (CHECK(fd == -1, "open sysctl", "open(%s): %s (%d)\n",
-		  sysctl, strerror(errno), errno))
-		return -1;
-
-	len = strlen(value);
-	err = write(fd, value, len);
-	close(fd);
-	if (CHECK(err != len, "write sysctl",
-		  "write(%s, %s, %d): err:%d %s (%d)\n",
-		  sysctl, value, len, err, strerror(errno), errno))
-		return -1;
-
-	return 0;
-}
-
 static int prepare_netns(void)
 {
 	if (CHECK(unshare(CLONE_NEWNET), "create netns",

tools/testing/selftests/bpf/prog_tests/connect_ping.c

+// SPDX-License-Identifier: GPL-2.0-only
+
+/*
+ * Copyright 2022 Google LLC.
+ */
+
+#define _GNU_SOURCE
+#include <sys/mount.h>
+
+#include "test_progs.h"
+#include "cgroup_helpers.h"
+#include "network_helpers.h"
+
+#include "connect_ping.skel.h"
+
+/* 2001:db8::1 */
+#define BINDADDR_V6 { { { 0x20,0x01,0x0d,0xb8,0,0,0,0,0,0,0,0,0,0,0,1 } } }
+static const struct in6_addr bindaddr_v6 = BINDADDR_V6;
+
+static void subtest(int cgroup_fd, struct connect_ping *skel,
+		    int family, int do_bind)
+{
+	struct sockaddr_in sa4 = {
+		.sin_family = AF_INET,
+		.sin_addr.s_addr = htonl(INADDR_LOOPBACK),
+	};
+	struct sockaddr_in6 sa6 = {
+		.sin6_family = AF_INET6,
+		.sin6_addr = IN6ADDR_LOOPBACK_INIT,
+	};
+	struct sockaddr *sa;
+	socklen_t sa_len;
+	int protocol;
+	int sock_fd;
+
+	switch (family) {
+	case AF_INET:
+		sa = (struct sockaddr *)&sa4;
+		sa_len = sizeof(sa4);
+		protocol = IPPROTO_ICMP;
+		break;
+	case AF_INET6:
+		sa = (struct sockaddr *)&sa6;
+		sa_len = sizeof(sa6);
+		protocol = IPPROTO_ICMPV6;
+		break;
+	}
+
+	memset(skel->bss, 0, sizeof(*skel->bss));
+	skel->bss->do_bind = do_bind;
+
+	sock_fd = socket(family, SOCK_DGRAM, protocol);
+	if (!ASSERT_GE(sock_fd, 0, "sock-create"))
+		return;
+
+	if (!ASSERT_OK(connect(sock_fd, sa, sa_len), "connect"))
+		goto close_sock;
+
+	if (!ASSERT_EQ(skel->bss->invocations_v4, family == AF_INET ? 1 : 0,
+		       "invocations_v4"))
+		goto close_sock;
+	if (!ASSERT_EQ(skel->bss->invocations_v6, family == AF_INET6 ? 1 : 0,
+		       "invocations_v6"))
+		goto close_sock;
+	if (!ASSERT_EQ(skel->bss->has_error, 0, "has_error"))
+		goto close_sock;
+
+	if (!ASSERT_OK(getsockname(sock_fd, sa, &sa_len),
+		       "getsockname"))
+		goto close_sock;
+
+	switch (family) {
+	case AF_INET:
+		if (!ASSERT_EQ(sa4.sin_family, family, "sin_family"))
+			goto close_sock;
+		if (!ASSERT_EQ(sa4.sin_addr.s_addr,
+			       htonl(do_bind ? 0x01010101 : INADDR_LOOPBACK),
+			       "sin_addr"))
+			goto close_sock;
+		break;
+	case AF_INET6:
+		if (!ASSERT_EQ(sa6.sin6_family, AF_INET6, "sin6_family"))
+			goto close_sock;
+		if (!ASSERT_EQ(memcmp(&sa6.sin6_addr,
+				      do_bind ? &bindaddr_v6 : &in6addr_loopback,
+				      sizeof(sa6.sin6_addr)),
+			       0, "sin6_addr"))
+			goto close_sock;
+		break;
+	}
+
+close_sock:
+	close(sock_fd);
+}
+
+void test_connect_ping(void)
+{
+	struct connect_ping *skel;
+	int cgroup_fd;
+
+	if (!ASSERT_OK(unshare(CLONE_NEWNET | CLONE_NEWNS), "unshare"))
+		return;
+
+	/* overmount sysfs, and making original sysfs private so overmount
+	 * does not propagate to other mntns.
+	 */
+	if (!ASSERT_OK(mount("none", "/sys", NULL, MS_PRIVATE, NULL),
+		       "remount-private-sys"))
+		return;
+	if (!ASSERT_OK(mount("sysfs", "/sys", "sysfs", 0, NULL),
+		       "mount-sys"))
+		return;
+	if (!ASSERT_OK(mount("bpffs", "/sys/fs/bpf", "bpf", 0, NULL),
+		       "mount-bpf"))
+		goto clean_mount;
+
+	if (!ASSERT_OK(system("ip link set dev lo up"), "lo-up"))
+		goto clean_mount;
+	if (!ASSERT_OK(system("ip addr add 1.1.1.1 dev lo"), "lo-addr-v4"))
+		goto clean_mount;
+	if (!ASSERT_OK(system("ip -6 addr add 2001:db8::1 dev lo"), "lo-addr-v6"))
+		goto clean_mount;
+	if (write_sysctl("/proc/sys/net/ipv4/ping_group_range", "0 0"))
+		goto clean_mount;
+
+	cgroup_fd = test__join_cgroup("/connect_ping");
+	if (!ASSERT_GE(cgroup_fd, 0, "cg-create"))
+		goto clean_mount;
+
+	skel = connect_ping__open_and_load();
+	if (!ASSERT_OK_PTR(skel, "skel-load"))
+		goto close_cgroup;
+	skel->links.connect_v4_prog =
+		bpf_program__attach_cgroup(skel->progs.connect_v4_prog, cgroup_fd);
+	if (!ASSERT_OK_PTR(skel->links.connect_v4_prog, "cg-attach-v4"))
+		goto skel_destroy;
+	skel->links.connect_v6_prog =
+		bpf_program__attach_cgroup(skel->progs.connect_v6_prog, cgroup_fd);
+	if (!ASSERT_OK_PTR(skel->links.connect_v6_prog, "cg-attach-v6"))
+		goto skel_destroy;
+
+	/* Connect a v4 ping socket to localhost, assert that only v4 is called,
+	 * and called exactly once, and that the socket's bound address is
+	 * original loopback address.
+	 */
+	if (test__start_subtest("ipv4"))
+		subtest(cgroup_fd, skel, AF_INET, 0);
+
+	/* Connect a v4 ping socket to localhost, assert that only v4 is called,
+	 * and called exactly once, and that the socket's bound address is
+	 * address we explicitly bound.
+	 */
+	if (test__start_subtest("ipv4-bind"))
+		subtest(cgroup_fd, skel, AF_INET, 1);
+
+	/* Connect a v6 ping socket to localhost, assert that only v6 is called,
+	 * and called exactly once, and that the socket's bound address is
+	 * original loopback address.
+	 */
+	if (test__start_subtest("ipv6"))
+		subtest(cgroup_fd, skel, AF_INET6, 0);
+
+	/* Connect a v6 ping socket to localhost, assert that only v6 is called,
+	 * and called exactly once, and that the socket's bound address is
+	 * address we explicitly bound.
+	 */
+	if (test__start_subtest("ipv6-bind"))
+		subtest(cgroup_fd, skel, AF_INET6, 1);
+
+skel_destroy:
+	connect_ping__destroy(skel);
+
+close_cgroup:
+	close(cgroup_fd);
+
+clean_mount:
+	umount2("/sys", MNT_DETACH);
+}

tools/testing/selftests/bpf/prog_tests/tcp_hdr_options.c

@@ -54,26 +54,6 @@ static int create_netns(void)
 	return 0;
 }
 
-static int write_sysctl(const char *sysctl, const char *value)
-{
-	int fd, err, len;
-
-	fd = open(sysctl, O_WRONLY);
-	if (CHECK(fd == -1, "open sysctl", "open(%s): %s (%d)\n",
-		  sysctl, strerror(errno), errno))
-		return -1;
-
-	len = strlen(value);
-	err = write(fd, value, len);
-	close(fd);
-	if (CHECK(err != len, "write sysctl",
-		  "write(%s, %s): err:%d %s (%d)\n",
-		  sysctl, value, err, strerror(errno), errno))
-		return -1;
-
-	return 0;
-}
-
 static void print_hdr_stg(const struct hdr_stg *hdr_stg, const char *prefix)
 {
 	fprintf(stderr, "%s{active:%u, resend_syn:%u, syncookie:%u, fastopen:%u}\n",

tools/testing/selftests/bpf/progs/connect_ping.c

+// SPDX-License-Identifier: GPL-2.0-only
+
+/*
+ * Copyright 2022 Google LLC.
+ */
+
+#include <linux/bpf.h>
+#include <bpf/bpf_helpers.h>
+#include <bpf/bpf_endian.h>
+#include <netinet/in.h>
+#include <sys/socket.h>
+
+/* 2001:db8::1 */
+#define BINDADDR_V6 { { { 0x20,0x01,0x0d,0xb8,0,0,0,0,0,0,0,0,0,0,0,1 } } }
+
+__u32 do_bind = 0;
+__u32 has_error = 0;
+__u32 invocations_v4 = 0;
+__u32 invocations_v6 = 0;
+
+SEC("cgroup/connect4")
+int connect_v4_prog(struct bpf_sock_addr *ctx)
+{
+	struct sockaddr_in sa = {
+		.sin_family = AF_INET,
+		.sin_addr.s_addr = bpf_htonl(0x01010101),
+	};
+
+	__sync_fetch_and_add(&invocations_v4, 1);
+
+	if (do_bind && bpf_bind(ctx, (struct sockaddr *)&sa, sizeof(sa)))
+		has_error = 1;
+
+	return 1;
+}
+
+SEC("cgroup/connect6")
+int connect_v6_prog(struct bpf_sock_addr *ctx)
+{
+	struct sockaddr_in6 sa = {
+		.sin6_family = AF_INET6,
+		.sin6_addr = BINDADDR_V6,
+	};
+
+	__sync_fetch_and_add(&invocations_v6, 1);
+
+	if (do_bind && bpf_bind(ctx, (struct sockaddr *)&sa, sizeof(sa)))
+		has_error = 1;
+
+	return 1;
+}
+
+char _license[] SEC("license") = "GPL";

tools/testing/selftests/bpf/test_progs.c

@@ -943,6 +943,23 @@ int trigger_module_test_write(int write_sz)
 	return 0;
 }
 
+int write_sysctl(const char *sysctl, const char *value)
+{
+	int fd, err, len;
+
+	fd = open(sysctl, O_WRONLY);
+	if (!ASSERT_NEQ(fd, -1, "open sysctl"))
+		return -1;
+
+	len = strlen(value);
+	err = write(fd, value, len);
+	close(fd);
+	if (!ASSERT_EQ(err, len, "write sysctl"))
+		return -1;
+
+	return 0;
+}
+
 #define MAX_BACKTRACE_SZ 128
 void crash_handler(int signum)
 {

tools/testing/selftests/bpf/test_progs.h

@@ -384,6 +384,7 @@ int extract_build_id(char *build_id, size_t size);
 int kern_sync_rcu(void);
 int trigger_module_test_read(int read_sz);
 int trigger_module_test_write(int write_sz);
+int write_sysctl(const char *sysctl, const char *value);
 
 #ifdef __x86_64__
 #define SYS_NANOSLEEP_KPROBE_NAME "__x64_sys_nanosleep"