summaryrefslogtreecommitdiffstats
path: root/tools
diff options
context:
space:
mode:
authorMartin KaFai Lau <kafai@fb.com>2018-08-08 01:01:31 -0700
committerDaniel Borkmann <daniel@iogearbox.net>2018-08-11 01:58:46 +0200
commit91134d849a0e8fbc70b8607d280e0d325dcaf7bb (patch)
tree4bc949c2a15c2c0d1cdbb390740f508fe7644623 /tools
parent6bc8529c414f931ce0acef3099b015cf2f5c4291 (diff)
downloadlinux-91134d849a0e8fbc70b8607d280e0d325dcaf7bb.tar.bz2
bpf: Test BPF_PROG_TYPE_SK_REUSEPORT
This patch add tests for the new BPF_PROG_TYPE_SK_REUSEPORT. The tests cover: - IPv4/IPv6 + TCP/UDP - TCP syncookie - TCP fastopen - Cases when the bpf_sk_select_reuseport() returning errors - Cases when the bpf prog returns SK_DROP - Values from sk_reuseport_md - outer_map => reuseport_array The test depends on commit 3eee1f75f2b9 ("bpf: fix bpf_skb_load_bytes_relative pkt length check") Signed-off-by: Martin KaFai Lau <kafai@fb.com> Acked-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Diffstat (limited to 'tools')
-rw-r--r--tools/lib/bpf/bpf.c1
-rw-r--r--tools/lib/bpf/bpf.h1
-rw-r--r--tools/testing/selftests/bpf/Makefile4
-rw-r--r--tools/testing/selftests/bpf/bpf_helpers.h4
-rw-r--r--tools/testing/selftests/bpf/test_select_reuseport.c688
-rw-r--r--tools/testing/selftests/bpf/test_select_reuseport_common.h36
-rw-r--r--tools/testing/selftests/bpf/test_select_reuseport_kern.c180
7 files changed, 912 insertions, 2 deletions
diff --git a/tools/lib/bpf/bpf.c b/tools/lib/bpf/bpf.c
index 9ddc89dae962..60aa4ca8b2c5 100644
--- a/tools/lib/bpf/bpf.c
+++ b/tools/lib/bpf/bpf.c
@@ -92,6 +92,7 @@ int bpf_create_map_xattr(const struct bpf_create_map_attr *create_attr)
attr.btf_key_type_id = create_attr->btf_key_type_id;
attr.btf_value_type_id = create_attr->btf_value_type_id;
attr.map_ifindex = create_attr->map_ifindex;
+ attr.inner_map_fd = create_attr->inner_map_fd;
return sys_bpf(BPF_MAP_CREATE, &attr, sizeof(attr));
}
diff --git a/tools/lib/bpf/bpf.h b/tools/lib/bpf/bpf.h
index 0639a30a457d..6f38164b2618 100644
--- a/tools/lib/bpf/bpf.h
+++ b/tools/lib/bpf/bpf.h
@@ -39,6 +39,7 @@ struct bpf_create_map_attr {
__u32 btf_key_type_id;
__u32 btf_value_type_id;
__u32 map_ifindex;
+ __u32 inner_map_fd;
};
int bpf_create_map_xattr(const struct bpf_create_map_attr *create_attr);
diff --git a/tools/testing/selftests/bpf/Makefile b/tools/testing/selftests/bpf/Makefile
index 17a7a5818ee1..daed162043c2 100644
--- a/tools/testing/selftests/bpf/Makefile
+++ b/tools/testing/selftests/bpf/Makefile
@@ -23,7 +23,7 @@ $(TEST_CUSTOM_PROGS): $(OUTPUT)/%: %.c
TEST_GEN_PROGS = test_verifier test_tag test_maps test_lru_map test_lpm_map test_progs \
test_align test_verifier_log test_dev_cgroup test_tcpbpf_user \
test_sock test_btf test_sockmap test_lirc_mode2_user get_cgroup_id_user \
- test_socket_cookie test_cgroup_storage
+ test_socket_cookie test_cgroup_storage test_select_reuseport
TEST_GEN_FILES = test_pkt_access.o test_xdp.o test_l4lb.o test_tcp_estats.o test_obj_id.o \
test_pkt_md_access.o test_xdp_redirect.o test_xdp_meta.o sockmap_parse_prog.o \
@@ -34,7 +34,7 @@ TEST_GEN_FILES = test_pkt_access.o test_xdp.o test_l4lb.o test_tcp_estats.o test
test_btf_haskv.o test_btf_nokv.o test_sockmap_kern.o test_tunnel_kern.o \
test_get_stack_rawtp.o test_sockmap_kern.o test_sockhash_kern.o \
test_lwt_seg6local.o sendmsg4_prog.o sendmsg6_prog.o test_lirc_mode2_kern.o \
- get_cgroup_id_kern.o socket_cookie_prog.o
+ get_cgroup_id_kern.o socket_cookie_prog.o test_select_reuseport_kern.o
# Order correspond to 'make run_tests' order
TEST_PROGS := test_kmod.sh \
diff --git a/tools/testing/selftests/bpf/bpf_helpers.h b/tools/testing/selftests/bpf/bpf_helpers.h
index 9ba1c72d7cf5..5c32266c2c38 100644
--- a/tools/testing/selftests/bpf/bpf_helpers.h
+++ b/tools/testing/selftests/bpf/bpf_helpers.h
@@ -111,6 +111,8 @@ static int (*bpf_xdp_adjust_tail)(void *ctx, int offset) =
static int (*bpf_skb_get_xfrm_state)(void *ctx, int index, void *state,
int size, int flags) =
(void *) BPF_FUNC_skb_get_xfrm_state;
+static int (*bpf_sk_select_reuseport)(void *ctx, void *map, void *key, __u32 flags) =
+ (void *) BPF_FUNC_sk_select_reuseport;
static int (*bpf_get_stack)(void *ctx, void *buf, int size, int flags) =
(void *) BPF_FUNC_get_stack;
static int (*bpf_fib_lookup)(void *ctx, struct bpf_fib_lookup *params,
@@ -173,6 +175,8 @@ struct bpf_map_def {
static int (*bpf_skb_load_bytes)(void *ctx, int off, void *to, int len) =
(void *) BPF_FUNC_skb_load_bytes;
+static int (*bpf_skb_load_bytes_relative)(void *ctx, int off, void *to, int len, __u32 start_header) =
+ (void *) BPF_FUNC_skb_load_bytes_relative;
static int (*bpf_skb_store_bytes)(void *ctx, int off, void *from, int len, int flags) =
(void *) BPF_FUNC_skb_store_bytes;
static int (*bpf_l3_csum_replace)(void *ctx, int off, int from, int to, int flags) =
diff --git a/tools/testing/selftests/bpf/test_select_reuseport.c b/tools/testing/selftests/bpf/test_select_reuseport.c
new file mode 100644
index 000000000000..75646d9b34aa
--- /dev/null
+++ b/tools/testing/selftests/bpf/test_select_reuseport.c
@@ -0,0 +1,688 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2018 Facebook */
+
+#include <stdlib.h>
+#include <unistd.h>
+#include <stdbool.h>
+#include <string.h>
+#include <errno.h>
+#include <assert.h>
+#include <fcntl.h>
+#include <linux/bpf.h>
+#include <linux/err.h>
+#include <linux/types.h>
+#include <linux/if_ether.h>
+#include <sys/types.h>
+#include <sys/epoll.h>
+#include <sys/socket.h>
+#include <netinet/in.h>
+#include <bpf/bpf.h>
+#include <bpf/libbpf.h>
+#include "bpf_rlimit.h"
+#include "bpf_util.h"
+#include "test_select_reuseport_common.h"
+
+#define MIN_TCPHDR_LEN 20
+#define UDPHDR_LEN 8
+
+#define TCP_SYNCOOKIE_SYSCTL "/proc/sys/net/ipv4/tcp_syncookies"
+#define TCP_FO_SYSCTL "/proc/sys/net/ipv4/tcp_fastopen"
+#define REUSEPORT_ARRAY_SIZE 32
+
+static int result_map, tmp_index_ovr_map, linum_map, data_check_map;
+static enum result expected_results[NR_RESULTS];
+static int sk_fds[REUSEPORT_ARRAY_SIZE];
+static int reuseport_array, outer_map;
+static int select_by_skb_data_prog;
+static int saved_tcp_syncookie;
+static struct bpf_object *obj;
+static int saved_tcp_fo;
+static __u32 index_zero;
+static int epfd;
+
+static union sa46 {
+ struct sockaddr_in6 v6;
+ struct sockaddr_in v4;
+ sa_family_t family;
+} srv_sa;
+
+#define CHECK(condition, tag, format...) ({ \
+ int __ret = !!(condition); \
+ if (__ret) { \
+ printf("%s(%d):FAIL:%s ", __func__, __LINE__, tag); \
+ printf(format); \
+ exit(-1); \
+ } \
+})
+
+static void create_maps(void)
+{
+ struct bpf_create_map_attr attr = {};
+
+ /* Creating reuseport_array */
+ attr.name = "reuseport_array";
+ attr.map_type = BPF_MAP_TYPE_REUSEPORT_SOCKARRAY;
+ attr.key_size = sizeof(__u32);
+ attr.value_size = sizeof(__u32);
+ attr.max_entries = REUSEPORT_ARRAY_SIZE;
+
+ reuseport_array = bpf_create_map_xattr(&attr);
+ CHECK(reuseport_array == -1, "creating reuseport_array",
+ "reuseport_array:%d errno:%d\n", reuseport_array, errno);
+
+ /* Creating outer_map */
+ attr.name = "outer_map";
+ attr.map_type = BPF_MAP_TYPE_ARRAY_OF_MAPS;
+ attr.key_size = sizeof(__u32);
+ attr.value_size = sizeof(__u32);
+ attr.max_entries = 1;
+ attr.inner_map_fd = reuseport_array;
+ outer_map = bpf_create_map_xattr(&attr);
+ CHECK(outer_map == -1, "creating outer_map",
+ "outer_map:%d errno:%d\n", outer_map, errno);
+}
+
+static void prepare_bpf_obj(void)
+{
+ struct bpf_program *prog;
+ struct bpf_map *map;
+ int err;
+ struct bpf_object_open_attr attr = {
+ .file = "test_select_reuseport_kern.o",
+ .prog_type = BPF_PROG_TYPE_SK_REUSEPORT,
+ };
+
+ obj = bpf_object__open_xattr(&attr);
+ CHECK(IS_ERR_OR_NULL(obj), "open test_select_reuseport_kern.o",
+ "obj:%p PTR_ERR(obj):%ld\n", obj, PTR_ERR(obj));
+
+ prog = bpf_program__next(NULL, obj);
+ CHECK(!prog, "get first bpf_program", "!prog\n");
+ bpf_program__set_type(prog, attr.prog_type);
+
+ map = bpf_object__find_map_by_name(obj, "outer_map");
+ CHECK(!map, "find outer_map", "!map\n");
+ err = bpf_map__reuse_fd(map, outer_map);
+ CHECK(err, "reuse outer_map", "err:%d\n", err);
+
+ err = bpf_object__load(obj);
+ CHECK(err, "load bpf_object", "err:%d\n", err);
+
+ select_by_skb_data_prog = bpf_program__fd(prog);
+ CHECK(select_by_skb_data_prog == -1, "get prog fd",
+ "select_by_skb_data_prog:%d\n", select_by_skb_data_prog);
+
+ map = bpf_object__find_map_by_name(obj, "result_map");
+ CHECK(!map, "find result_map", "!map\n");
+ result_map = bpf_map__fd(map);
+ CHECK(result_map == -1, "get result_map fd",
+ "result_map:%d\n", result_map);
+
+ map = bpf_object__find_map_by_name(obj, "tmp_index_ovr_map");
+ CHECK(!map, "find tmp_index_ovr_map", "!map\n");
+ tmp_index_ovr_map = bpf_map__fd(map);
+ CHECK(tmp_index_ovr_map == -1, "get tmp_index_ovr_map fd",
+ "tmp_index_ovr_map:%d\n", tmp_index_ovr_map);
+
+ map = bpf_object__find_map_by_name(obj, "linum_map");
+ CHECK(!map, "find linum_map", "!map\n");
+ linum_map = bpf_map__fd(map);
+ CHECK(linum_map == -1, "get linum_map fd",
+ "linum_map:%d\n", linum_map);
+
+ map = bpf_object__find_map_by_name(obj, "data_check_map");
+ CHECK(!map, "find data_check_map", "!map\n");
+ data_check_map = bpf_map__fd(map);
+ CHECK(data_check_map == -1, "get data_check_map fd",
+ "data_check_map:%d\n", data_check_map);
+}
+
+static void sa46_init_loopback(union sa46 *sa, sa_family_t family)
+{
+ memset(sa, 0, sizeof(*sa));
+ sa->family = family;
+ if (sa->family == AF_INET6)
+ sa->v6.sin6_addr = in6addr_loopback;
+ else
+ sa->v4.sin_addr.s_addr = htonl(INADDR_LOOPBACK);
+}
+
+static void sa46_init_inany(union sa46 *sa, sa_family_t family)
+{
+ memset(sa, 0, sizeof(*sa));
+ sa->family = family;
+ if (sa->family == AF_INET6)
+ sa->v6.sin6_addr = in6addr_any;
+ else
+ sa->v4.sin_addr.s_addr = INADDR_ANY;
+}
+
+static int read_int_sysctl(const char *sysctl)
+{
+ char buf[16];
+ int fd, ret;
+
+ fd = open(sysctl, 0);
+ CHECK(fd == -1, "open(sysctl)", "sysctl:%s fd:%d errno:%d\n",
+ sysctl, fd, errno);
+
+ ret = read(fd, buf, sizeof(buf));
+ CHECK(ret <= 0, "read(sysctl)", "sysctl:%s ret:%d errno:%d\n",
+ sysctl, ret, errno);
+ close(fd);
+
+ return atoi(buf);
+}
+
+static void write_int_sysctl(const char *sysctl, int v)
+{
+ int fd, ret, size;
+ char buf[16];
+
+ fd = open(sysctl, O_RDWR);
+ CHECK(fd == -1, "open(sysctl)", "sysctl:%s fd:%d errno:%d\n",
+ sysctl, fd, errno);
+
+ size = snprintf(buf, sizeof(buf), "%d", v);
+ ret = write(fd, buf, size);
+ CHECK(ret != size, "write(sysctl)",
+ "sysctl:%s ret:%d size:%d errno:%d\n", sysctl, ret, size, errno);
+ close(fd);
+}
+
+static void restore_sysctls(void)
+{
+ write_int_sysctl(TCP_FO_SYSCTL, saved_tcp_fo);
+ write_int_sysctl(TCP_SYNCOOKIE_SYSCTL, saved_tcp_syncookie);
+}
+
+static void enable_fastopen(void)
+{
+ int fo;
+
+ fo = read_int_sysctl(TCP_FO_SYSCTL);
+ write_int_sysctl(TCP_FO_SYSCTL, fo | 7);
+}
+
+static void enable_syncookie(void)
+{
+ write_int_sysctl(TCP_SYNCOOKIE_SYSCTL, 2);
+}
+
+static void disable_syncookie(void)
+{
+ write_int_sysctl(TCP_SYNCOOKIE_SYSCTL, 0);
+}
+
+static __u32 get_linum(void)
+{
+ __u32 linum;
+ int err;
+
+ err = bpf_map_lookup_elem(linum_map, &index_zero, &linum);
+ CHECK(err == -1, "lookup_elem(linum_map)", "err:%d errno:%d\n",
+ err, errno);
+
+ return linum;
+}
+
+static void check_data(int type, sa_family_t family, const struct cmd *cmd,
+ int cli_fd)
+{
+ struct data_check expected = {}, result;
+ union sa46 cli_sa;
+ socklen_t addrlen;
+ int err;
+
+ addrlen = sizeof(cli_sa);
+ err = getsockname(cli_fd, (struct sockaddr *)&cli_sa,
+ &addrlen);
+ CHECK(err == -1, "getsockname(cli_fd)", "err:%d errno:%d\n",
+ err, errno);
+
+ err = bpf_map_lookup_elem(data_check_map, &index_zero, &result);
+ CHECK(err == -1, "lookup_elem(data_check_map)", "err:%d errno:%d\n",
+ err, errno);
+
+ if (type == SOCK_STREAM) {
+ expected.len = MIN_TCPHDR_LEN;
+ expected.ip_protocol = IPPROTO_TCP;
+ } else {
+ expected.len = UDPHDR_LEN;
+ expected.ip_protocol = IPPROTO_UDP;
+ }
+
+ if (family == AF_INET6) {
+ expected.eth_protocol = htons(ETH_P_IPV6);
+ expected.bind_inany = !srv_sa.v6.sin6_addr.s6_addr32[3] &&
+ !srv_sa.v6.sin6_addr.s6_addr32[2] &&
+ !srv_sa.v6.sin6_addr.s6_addr32[1] &&
+ !srv_sa.v6.sin6_addr.s6_addr32[0];
+
+ memcpy(&expected.skb_addrs[0], cli_sa.v6.sin6_addr.s6_addr32,
+ sizeof(cli_sa.v6.sin6_addr));
+ memcpy(&expected.skb_addrs[4], &in6addr_loopback,
+ sizeof(in6addr_loopback));
+ expected.skb_ports[0] = cli_sa.v6.sin6_port;
+ expected.skb_ports[1] = srv_sa.v6.sin6_port;
+ } else {
+ expected.eth_protocol = htons(ETH_P_IP);
+ expected.bind_inany = !srv_sa.v4.sin_addr.s_addr;
+
+ expected.skb_addrs[0] = cli_sa.v4.sin_addr.s_addr;
+ expected.skb_addrs[1] = htonl(INADDR_LOOPBACK);
+ expected.skb_ports[0] = cli_sa.v4.sin_port;
+ expected.skb_ports[1] = srv_sa.v4.sin_port;
+ }
+
+ if (memcmp(&result, &expected, offsetof(struct data_check,
+ equal_check_end))) {
+ printf("unexpected data_check\n");
+ printf(" result: (0x%x, %u, %u)\n",
+ result.eth_protocol, result.ip_protocol,
+ result.bind_inany);
+ printf("expected: (0x%x, %u, %u)\n",
+ expected.eth_protocol, expected.ip_protocol,
+ expected.bind_inany);
+ CHECK(1, "data_check result != expected",
+ "bpf_prog_linum:%u\n", get_linum());
+ }
+
+ CHECK(!result.hash, "data_check result.hash empty",
+ "result.hash:%u", result.hash);
+
+ expected.len += cmd ? sizeof(*cmd) : 0;
+ if (type == SOCK_STREAM)
+ CHECK(expected.len > result.len, "expected.len > result.len",
+ "expected.len:%u result.len:%u bpf_prog_linum:%u\n",
+ expected.len, result.len, get_linum());
+ else
+ CHECK(expected.len != result.len, "expected.len != result.len",
+ "expected.len:%u result.len:%u bpf_prog_linum:%u\n",
+ expected.len, result.len, get_linum());
+}
+
+static void check_results(void)
+{
+ __u32 results[NR_RESULTS];
+ __u32 i, broken = 0;
+ int err;
+
+ for (i = 0; i < NR_RESULTS; i++) {
+ err = bpf_map_lookup_elem(result_map, &i, &results[i]);
+ CHECK(err == -1, "lookup_elem(result_map)",
+ "i:%u err:%d errno:%d\n", i, err, errno);
+ }
+
+ for (i = 0; i < NR_RESULTS; i++) {
+ if (results[i] != expected_results[i]) {
+ broken = i;
+ break;
+ }
+ }
+
+ if (i == NR_RESULTS)
+ return;
+
+ printf("unexpected result\n");
+ printf(" result: [");
+ printf("%u", results[0]);
+ for (i = 1; i < NR_RESULTS; i++)
+ printf(", %u", results[i]);
+ printf("]\n");
+
+ printf("expected: [");
+ printf("%u", expected_results[0]);
+ for (i = 1; i < NR_RESULTS; i++)
+ printf(", %u", expected_results[i]);
+ printf("]\n");
+
+ CHECK(expected_results[broken] != results[broken],
+ "unexpected result",
+ "expected_results[%u] != results[%u] bpf_prog_linum:%u\n",
+ broken, broken, get_linum());
+}
+
+static int send_data(int type, sa_family_t family, void *data, size_t len,
+ enum result expected)
+{
+ union sa46 cli_sa;
+ int fd, err;
+
+ fd = socket(family, type, 0);
+ CHECK(fd == -1, "socket()", "fd:%d errno:%d\n", fd, errno);
+
+ sa46_init_loopback(&cli_sa, family);
+ err = bind(fd, (struct sockaddr *)&cli_sa, sizeof(cli_sa));
+ CHECK(fd == -1, "bind(cli_sa)", "err:%d errno:%d\n", err, errno);
+
+ err = sendto(fd, data, len, MSG_FASTOPEN, (struct sockaddr *)&srv_sa,
+ sizeof(srv_sa));
+ CHECK(err != len && expected >= PASS,
+ "sendto()", "family:%u err:%d errno:%d expected:%d\n",
+ family, err, errno, expected);
+
+ return fd;
+}
+
+static void do_test(int type, sa_family_t family, struct cmd *cmd,
+ enum result expected)
+{
+ int nev, srv_fd, cli_fd;
+ struct epoll_event ev;
+ struct cmd rcv_cmd;
+ ssize_t nread;
+
+ cli_fd = send_data(type, family, cmd, cmd ? sizeof(*cmd) : 0,
+ expected);
+ nev = epoll_wait(epfd, &ev, 1, expected >= PASS ? 5 : 0);
+ CHECK((nev <= 0 && expected >= PASS) ||
+ (nev > 0 && expected < PASS),
+ "nev <> expected",
+ "nev:%d expected:%d type:%d family:%d data:(%d, %d)\n",
+ nev, expected, type, family,
+ cmd ? cmd->reuseport_index : -1,
+ cmd ? cmd->pass_on_failure : -1);
+ check_results();
+ check_data(type, family, cmd, cli_fd);
+
+ if (expected < PASS)
+ return;
+
+ CHECK(expected != PASS_ERR_SK_SELECT_REUSEPORT &&
+ cmd->reuseport_index != ev.data.u32,
+ "check cmd->reuseport_index",
+ "cmd:(%u, %u) ev.data.u32:%u\n",
+ cmd->pass_on_failure, cmd->reuseport_index, ev.data.u32);
+
+ srv_fd = sk_fds[ev.data.u32];
+ if (type == SOCK_STREAM) {
+ int new_fd = accept(srv_fd, NULL, 0);
+
+ CHECK(new_fd == -1, "accept(srv_fd)",
+ "ev.data.u32:%u new_fd:%d errno:%d\n",
+ ev.data.u32, new_fd, errno);
+
+ nread = recv(new_fd, &rcv_cmd, sizeof(rcv_cmd), MSG_DONTWAIT);
+ CHECK(nread != sizeof(rcv_cmd),
+ "recv(new_fd)",
+ "ev.data.u32:%u nread:%zd sizeof(rcv_cmd):%zu errno:%d\n",
+ ev.data.u32, nread, sizeof(rcv_cmd), errno);
+
+ close(new_fd);
+ } else {
+ nread = recv(srv_fd, &rcv_cmd, sizeof(rcv_cmd), MSG_DONTWAIT);
+ CHECK(nread != sizeof(rcv_cmd),
+ "recv(sk_fds)",
+ "ev.data.u32:%u nread:%zd sizeof(rcv_cmd):%zu errno:%d\n",
+ ev.data.u32, nread, sizeof(rcv_cmd), errno);
+ }
+
+ close(cli_fd);
+}
+
+static void test_err_inner_map(int type, sa_family_t family)
+{
+ struct cmd cmd = {
+ .reuseport_index = 0,
+ .pass_on_failure = 0,
+ };
+
+ printf("%s: ", __func__);
+ expected_results[DROP_ERR_INNER_MAP]++;
+ do_test(type, family, &cmd, DROP_ERR_INNER_MAP);
+ printf("OK\n");
+}
+
+static void test_err_skb_data(int type, sa_family_t family)
+{
+ printf("%s: ", __func__);
+ expected_results[DROP_ERR_SKB_DATA]++;
+ do_test(type, family, NULL, DROP_ERR_SKB_DATA);
+ printf("OK\n");
+}
+
+static void test_err_sk_select_port(int type, sa_family_t family)
+{
+ struct cmd cmd = {
+ .reuseport_index = REUSEPORT_ARRAY_SIZE,
+ .pass_on_failure = 0,
+ };
+
+ printf("%s: ", __func__);
+ expected_results[DROP_ERR_SK_SELECT_REUSEPORT]++;
+ do_test(type, family, &cmd, DROP_ERR_SK_SELECT_REUSEPORT);
+ printf("OK\n");
+}
+
+static void test_pass(int type, sa_family_t family)
+{
+ struct cmd cmd;
+ int i;
+
+ printf("%s: ", __func__);
+ cmd.pass_on_failure = 0;
+ for (i = 0; i < REUSEPORT_ARRAY_SIZE; i++) {
+ expected_results[PASS]++;
+ cmd.reuseport_index = i;
+ do_test(type, family, &cmd, PASS);
+ }
+ printf("OK\n");
+}
+
+static void test_syncookie(int type, sa_family_t family)
+{
+ int err, tmp_index = 1;
+ struct cmd cmd = {
+ .reuseport_index = 0,
+ .pass_on_failure = 0,
+ };
+
+ if (type != SOCK_STREAM)
+ return;
+
+ printf("%s: ", __func__);
+ /*
+ * +1 for TCP-SYN and
+ * +1 for the TCP-ACK (ack the syncookie)
+ */
+ expected_results[PASS] += 2;
+ enable_syncookie();
+ /*
+ * Simulate TCP-SYN and TCP-ACK are handled by two different sk:
+ * TCP-SYN: select sk_fds[tmp_index = 1] tmp_index is from the
+ * tmp_index_ovr_map
+ * TCP-ACK: select sk_fds[reuseport_index = 0] reuseport_index
+ * is from the cmd.reuseport_index
+ */
+ err = bpf_map_update_elem(tmp_index_ovr_map, &index_zero,
+ &tmp_index, BPF_ANY);
+ CHECK(err == -1, "update_elem(tmp_index_ovr_map, 0, 1)",
+ "err:%d errno:%d\n", err, errno);
+ do_test(type, family, &cmd, PASS);
+ err = bpf_map_lookup_elem(tmp_index_ovr_map, &index_zero,
+ &tmp_index);
+ CHECK(err == -1 || tmp_index != -1,
+ "lookup_elem(tmp_index_ovr_map)",
+ "err:%d errno:%d tmp_index:%d\n",
+ err, errno, tmp_index);
+ disable_syncookie();
+ printf("OK\n");
+}
+
+static void test_pass_on_err(int type, sa_family_t family)
+{
+ struct cmd cmd = {
+ .reuseport_index = REUSEPORT_ARRAY_SIZE,
+ .pass_on_failure = 1,
+ };
+
+ printf("%s: ", __func__);
+ expected_results[PASS_ERR_SK_SELECT_REUSEPORT] += 1;
+ do_test(type, family, &cmd, PASS_ERR_SK_SELECT_REUSEPORT);
+ printf("OK\n");
+}
+
+static void prepare_sk_fds(int type, sa_family_t family, bool inany)
+{
+ const int first = REUSEPORT_ARRAY_SIZE - 1;
+ int i, err, optval = 1;
+ struct epoll_event ev;
+ socklen_t addrlen;
+
+ if (inany)
+ sa46_init_inany(&srv_sa, family);
+ else
+ sa46_init_loopback(&srv_sa, family);
+ addrlen = sizeof(srv_sa);
+
+ /*
+ * The sk_fds[] is filled from the back such that the order
+ * is exactly opposite to the (struct sock_reuseport *)reuse->socks[].
+ */
+ for (i = first; i >= 0; i--) {
+ sk_fds[i] = socket(family, type, 0);
+ CHECK(sk_fds[i] == -1, "socket()", "sk_fds[%d]:%d errno:%d\n",
+ i, sk_fds[i], errno);
+ err = setsockopt(sk_fds[i], SOL_SOCKET, SO_REUSEPORT,
+ &optval, sizeof(optval));
+ CHECK(err == -1, "setsockopt(SO_REUSEPORT)",
+ "sk_fds[%d] err:%d errno:%d\n",
+ i, err, errno);
+
+ if (i == first) {
+ err = setsockopt(sk_fds[i], SOL_SOCKET,
+ SO_ATTACH_REUSEPORT_EBPF,
+ &select_by_skb_data_prog,
+ sizeof(select_by_skb_data_prog));
+ CHECK(err == -1, "setsockopt(SO_ATTACH_REUEPORT_EBPF)",
+ "err:%d errno:%d\n", err, errno);
+ }
+
+ err = bind(sk_fds[i], (struct sockaddr *)&srv_sa, addrlen);
+ CHECK(err == -1, "bind()", "sk_fds[%d] err:%d errno:%d\n",
+ i, err, errno);
+
+ if (type == SOCK_STREAM) {
+ err = listen(sk_fds[i], 10);
+ CHECK(err == -1, "listen()",
+ "sk_fds[%d] err:%d errno:%d\n",
+ i, err, errno);
+ }
+
+ err = bpf_map_update_elem(reuseport_array, &i, &sk_fds[i],
+ BPF_NOEXIST);
+ CHECK(err == -1, "update_elem(reuseport_array)",
+ "sk_fds[%d] err:%d errno:%d\n", i, err, errno);
+
+ if (i == first) {
+ socklen_t addrlen = sizeof(srv_sa);
+
+ err = getsockname(sk_fds[i], (struct sockaddr *)&srv_sa,
+ &addrlen);
+ CHECK(err == -1, "getsockname()",
+ "sk_fds[%d] err:%d errno:%d\n", i, err, errno);
+ }
+ }
+
+ epfd = epoll_create(1);
+ CHECK(epfd == -1, "epoll_create(1)",
+ "epfd:%d errno:%d\n", epfd, errno);
+
+ ev.events = EPOLLIN;
+ for (i = 0; i < REUSEPORT_ARRAY_SIZE; i++) {
+ ev.data.u32 = i;
+ err = epoll_ctl(epfd, EPOLL_CTL_ADD, sk_fds[i], &ev);
+ CHECK(err, "epoll_ctl(EPOLL_CTL_ADD)", "sk_fds[%d]\n", i);
+ }
+}
+
+static void setup_per_test(int type, unsigned short family, bool inany)
+{
+ int ovr = -1, err;
+
+ prepare_sk_fds(type, family, inany);
+ err = bpf_map_update_elem(tmp_index_ovr_map, &index_zero, &ovr,
+ BPF_ANY);
+ CHECK(err == -1, "update_elem(tmp_index_ovr_map, 0, -1)",
+ "err:%d errno:%d\n", err, errno);
+}
+
+static void cleanup_per_test(void)
+{
+ int i, err;
+
+ for (i = 0; i < REUSEPORT_ARRAY_SIZE; i++)
+ close(sk_fds[i]);
+ close(epfd);
+
+ err = bpf_map_delete_elem(outer_map, &index_zero);
+ CHECK(err == -1, "delete_elem(outer_map)",
+ "err:%d errno:%d\n", err, errno);
+}
+
+static void cleanup(void)
+{
+ close(outer_map);
+ close(reuseport_array);
+ bpf_object__close(obj);
+}
+
+static void test_all(void)
+{
+ /* Extra SOCK_STREAM to test bind_inany==true */
+ const int types[] = { SOCK_STREAM, SOCK_DGRAM, SOCK_STREAM };
+ const char * const type_strings[] = { "TCP", "UDP", "TCP" };
+ const char * const family_strings[] = { "IPv6", "IPv4" };
+ const unsigned short families[] = { AF_INET6, AF_INET };
+ const bool bind_inany[] = { false, false, true };
+ int t, f, err;
+
+ for (f = 0; f < ARRAY_SIZE(families); f++) {
+ unsigned short family = families[f];
+
+ for (t = 0; t < ARRAY_SIZE(types); t++) {
+ bool inany = bind_inany[t];
+ int type = types[t];
+
+ printf("######## %s/%s %s ########\n",
+ family_strings[f], type_strings[t],
+ inany ? " INANY " : "LOOPBACK");
+
+ setup_per_test(type, family, inany);
+
+ test_err_inner_map(type, family);
+
+ /* Install reuseport_array to the outer_map */
+ err = bpf_map_update_elem(outer_map, &index_zero,
+ &reuseport_array, BPF_ANY);
+ CHECK(err == -1, "update_elem(outer_map)",
+ "err:%d errno:%d\n", err, errno);
+
+ test_err_skb_data(type, family);
+ test_err_sk_select_port(type, family);
+ test_pass(type, family);
+ test_syncookie(type, family);
+ test_pass_on_err(type, family);
+
+ cleanup_per_test();
+ printf("\n");
+ }
+ }
+}
+
+int main(int argc, const char **argv)
+{
+ create_maps();
+ prepare_bpf_obj();
+ saved_tcp_fo = read_int_sysctl(TCP_FO_SYSCTL);
+ saved_tcp_syncookie = read_int_sysctl(TCP_SYNCOOKIE_SYSCTL);
+ enable_fastopen();
+ disable_syncookie();
+ atexit(restore_sysctls);
+
+ test_all();
+
+ cleanup();
+ return 0;
+}
diff --git a/tools/testing/selftests/bpf/test_select_reuseport_common.h b/tools/testing/selftests/bpf/test_select_reuseport_common.h
new file mode 100644
index 000000000000..08eb2a9f145f
--- /dev/null
+++ b/tools/testing/selftests/bpf/test_select_reuseport_common.h
@@ -0,0 +1,36 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Copyright (c) 2018 Facebook */
+
+#ifndef __TEST_SELECT_REUSEPORT_COMMON_H
+#define __TEST_SELECT_REUSEPORT_COMMON_H
+
+#include <linux/types.h>
+
+enum result {
+ DROP_ERR_INNER_MAP,
+ DROP_ERR_SKB_DATA,
+ DROP_ERR_SK_SELECT_REUSEPORT,
+ DROP_MISC,
+ PASS,
+ PASS_ERR_SK_SELECT_REUSEPORT,
+ NR_RESULTS,
+};
+
+struct cmd {
+ __u32 reuseport_index;
+ __u32 pass_on_failure;
+};
+
+struct data_check {
+ __u32 ip_protocol;
+ __u32 skb_addrs[8];
+ __u16 skb_ports[2];
+ __u16 eth_protocol;
+ __u8 bind_inany;
+ __u8 equal_check_end[0];
+
+ __u32 len;
+ __u32 hash;
+};
+
+#endif
diff --git a/tools/testing/selftests/bpf/test_select_reuseport_kern.c b/tools/testing/selftests/bpf/test_select_reuseport_kern.c
new file mode 100644
index 000000000000..5b54ec637ada
--- /dev/null
+++ b/tools/testing/selftests/bpf/test_select_reuseport_kern.c
@@ -0,0 +1,180 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2018 Facebook */
+
+#include <stdlib.h>
+#include <linux/in.h>
+#include <linux/ip.h>
+#include <linux/ipv6.h>
+#include <linux/tcp.h>
+#include <linux/udp.h>
+#include <linux/bpf.h>
+#include <linux/types.h>
+#include <linux/if_ether.h>
+
+#include "bpf_endian.h"
+#include "bpf_helpers.h"
+#include "test_select_reuseport_common.h"
+
+int _version SEC("version") = 1;
+
+#ifndef offsetof
+#define offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER)
+#endif
+
+struct bpf_map_def SEC("maps") outer_map = {
+ .type = BPF_MAP_TYPE_ARRAY_OF_MAPS,
+ .key_size = sizeof(__u32),
+ .value_size = sizeof(__u32),
+ .max_entries = 1,
+};
+
+struct bpf_map_def SEC("maps") result_map = {
+ .type = BPF_MAP_TYPE_ARRAY,
+ .key_size = sizeof(__u32),
+ .value_size = sizeof(__u32),
+ .max_entries = NR_RESULTS,
+};
+
+struct bpf_map_def SEC("maps") tmp_index_ovr_map = {
+ .type = BPF_MAP_TYPE_ARRAY,
+ .key_size = sizeof(__u32),
+ .value_size = sizeof(int),
+ .max_entries = 1,
+};
+
+struct bpf_map_def SEC("maps") linum_map = {
+ .type = BPF_MAP_TYPE_ARRAY,
+ .key_size = sizeof(__u32),
+ .value_size = sizeof(__u32),
+ .max_entries = 1,
+};
+
+struct bpf_map_def SEC("maps") data_check_map = {
+ .type = BPF_MAP_TYPE_ARRAY,
+ .key_size = sizeof(__u32),
+ .value_size = sizeof(struct data_check),
+ .max_entries = 1,
+};
+
+#define GOTO_DONE(_result) ({ \
+ result = (_result); \
+ linum = __LINE__; \
+ goto done; \
+})
+
+SEC("select_by_skb_data")
+int _select_by_skb_data(struct sk_reuseport_md *reuse_md)
+{
+ __u32 linum, index = 0, flags = 0, index_zero = 0;
+ __u32 *result_cnt, *linum_value;
+ struct data_check data_check = {};
+ struct cmd *cmd, cmd_copy;
+ void *data, *data_end;
+ void *reuseport_array;
+ enum result result;
+ int *index_ovr;
+ int err;
+
+ data = reuse_md->data;
+ data_end = reuse_md->data_end;
+ data_check.len = reuse_md->len;
+ data_check.eth_protocol = reuse_md->eth_protocol;
+ data_check.ip_protocol = reuse_md->ip_protocol;
+ data_check.hash = reuse_md->hash;
+ data_check.bind_inany = reuse_md->bind_inany;
+ if (data_check.eth_protocol == bpf_htons(ETH_P_IP)) {
+ if (bpf_skb_load_bytes_relative(reuse_md,
+ offsetof(struct iphdr, saddr),
+ data_check.skb_addrs, 8,
+ BPF_HDR_START_NET))
+ GOTO_DONE(DROP_MISC);
+ } else {
+ if (bpf_skb_load_bytes_relative(reuse_md,
+ offsetof(struct ipv6hdr, saddr),
+ data_check.skb_addrs, 32,
+ BPF_HDR_START_NET))
+ GOTO_DONE(DROP_MISC);
+ }
+
+ /*
+ * The ip_protocol could be a compile time decision
+ * if the bpf_prog.o is dedicated to either TCP or
+ * UDP.
+ *
+ * Otherwise, reuse_md->ip_protocol or
+ * the protocol field in the iphdr can be used.
+ */
+ if (data_check.ip_protocol == IPPROTO_TCP) {
+ struct tcphdr *th = data;
+
+ if (th + 1 > data_end)
+ GOTO_DONE(DROP_MISC);
+
+ data_check.skb_ports[0] = th->source;
+ data_check.skb_ports[1] = th->dest;
+
+ if ((th->doff << 2) + sizeof(*cmd) > data_check.len)
+ GOTO_DONE(DROP_ERR_SKB_DATA);
+ if (bpf_skb_load_bytes(reuse_md, th->doff << 2, &cmd_copy,
+ sizeof(cmd_copy)))
+ GOTO_DONE(DROP_MISC);
+ cmd = &cmd_copy;
+ } else if (data_check.ip_protocol == IPPROTO_UDP) {
+ struct udphdr *uh = data;
+
+ if (uh + 1 > data_end)
+ GOTO_DONE(DROP_MISC);
+
+ data_check.skb_ports[0] = uh->source;
+ data_check.skb_ports[1] = uh->dest;
+
+ if (sizeof(struct udphdr) + sizeof(*cmd) > data_check.len)
+ GOTO_DONE(DROP_ERR_SKB_DATA);
+ if (data + sizeof(struct udphdr) + sizeof(*cmd) > data_end) {
+ if (bpf_skb_load_bytes(reuse_md, sizeof(struct udphdr),
+ &cmd_copy, sizeof(cmd_copy)))
+ GOTO_DONE(DROP_MISC);
+ cmd = &cmd_copy;
+ } else {
+ cmd = data + sizeof(struct udphdr);
+ }
+ } else {
+ GOTO_DONE(DROP_MISC);
+ }
+
+ reuseport_array = bpf_map_lookup_elem(&outer_map, &index_zero);
+ if (!reuseport_array)
+ GOTO_DONE(DROP_ERR_INNER_MAP);
+
+ index = cmd->reuseport_index;
+ index_ovr = bpf_map_lookup_elem(&tmp_index_ovr_map, &index_zero);
+ if (!index_ovr)
+ GOTO_DONE(DROP_MISC);
+
+ if (*index_ovr != -1) {
+ index = *index_ovr;
+ *index_ovr = -1;
+ }
+ err = bpf_sk_select_reuseport(reuse_md, reuseport_array, &index,
+ flags);
+ if (!err)
+ GOTO_DONE(PASS);
+
+ if (cmd->pass_on_failure)
+ GOTO_DONE(PASS_ERR_SK_SELECT_REUSEPORT);
+ else
+ GOTO_DONE(DROP_ERR_SK_SELECT_REUSEPORT);
+
+done:
+ result_cnt = bpf_map_lookup_elem(&result_map, &result);
+ if (!result_cnt)
+ return SK_DROP;
+
+ bpf_map_update_elem(&linum_map, &index_zero, &linum, BPF_ANY);
+ bpf_map_update_elem(&data_check_map, &index_zero, &data_check, BPF_ANY);
+
+ (*result_cnt)++;
+ return result < PASS ? SK_DROP : SK_PASS;
+}
+
+char _license[] SEC("license") = "GPL";