is_mptcp is a field from struct tcp_sock used to indicate that the
current tcp_sock is part of the MPTCP protocol.
In this protocol, a first socket (mptcp_sock) is created with
sk_protocol set to IPPROTO_MPTCP (=262) for control purpose but it
isn't directly on the wire. This is the role of the subflow (kernel)
sockets which are classical tcp_sock with sk_protocol set to
IPPROTO_TCP. The only way to differentiate such sockets from plain TCP
sockets is the is_mptcp field from tcp_sock.
Such an exposure in BPF is thus required to be able to differentiate
plain TCP sockets from MPTCP subflow sockets in BPF_PROG_TYPE_SOCK_OPS
programs.
The choice has been made to silently pass the case when CONFIG_MPTCP is
unset by defaulting is_mptcp to 0 in order to make BPF independent of
the MPTCP configuration. Another solution is to make the verifier fail
in 'bpf_tcp_sock_is_valid_ctx_access' but this will add an additional
'#ifdef CONFIG_MPTCP' in the BPF code and a same injected BPF program
will not run if MPTCP is not set.
An example use-case is provided in
https://github.com/multipath-tcp/mptcp_net-next/tree/scripts/bpf/examples
Suggested-by: Matthieu Baerts <[email protected]>
Acked-by: Matthieu Baerts <[email protected]>
Acked-by: Mat Martineau <[email protected]>
Signed-off-by: Nicolas Rybowski <[email protected]>
---
include/uapi/linux/bpf.h | 1 +
net/core/filter.c | 9 ++++++++-
tools/include/uapi/linux/bpf.h | 1 +
3 files changed, 10 insertions(+), 1 deletion(-)
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index 7dd314176df7..7d179eada1c3 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -4060,6 +4060,7 @@ struct bpf_tcp_sock {
__u32 delivered; /* Total data packets delivered incl. rexmits */
__u32 delivered_ce; /* Like the above but only ECE marked packets */
__u32 icsk_retransmits; /* Number of unrecovered [RTO] timeouts */
+ __u32 is_mptcp; /* Is MPTCP subflow? */
};
struct bpf_sock_tuple {
diff --git a/net/core/filter.c b/net/core/filter.c
index d266c6941967..dab48528dceb 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -5837,7 +5837,7 @@ bool bpf_tcp_sock_is_valid_access(int off, int size, enum bpf_access_type type,
struct bpf_insn_access_aux *info)
{
if (off < 0 || off >= offsetofend(struct bpf_tcp_sock,
- icsk_retransmits))
+ is_mptcp))
return false;
if (off % size != 0)
@@ -5971,6 +5971,13 @@ u32 bpf_tcp_sock_convert_ctx_access(enum bpf_access_type type,
case offsetof(struct bpf_tcp_sock, icsk_retransmits):
BPF_INET_SOCK_GET_COMMON(icsk_retransmits);
break;
+ case offsetof(struct bpf_tcp_sock, is_mptcp):
+#ifdef CONFIG_MPTCP
+ BPF_TCP_SOCK_GET_COMMON(is_mptcp);
+#else
+ *insn++ = BPF_MOV32_IMM(si->dst_reg, 0);
+#endif
+ break;
}
return insn - insn_buf;
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index 7dd314176df7..7d179eada1c3 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -4060,6 +4060,7 @@ struct bpf_tcp_sock {
__u32 delivered; /* Total data packets delivered incl. rexmits */
__u32 delivered_ce; /* Like the above but only ECE marked packets */
__u32 icsk_retransmits; /* Number of unrecovered [RTO] timeouts */
+ __u32 is_mptcp; /* Is MPTCP subflow? */
};
struct bpf_sock_tuple {
--
2.28.0
This patch adds a base for MPTCP specific tests.
It is currently limited to the is_mptcp field in case of plain TCP
connection because for the moment there is no easy way to get the subflow
sk from a msk in userspace. This implies that we cannot lookup the
sk_storage attached to the subflow sk in the sockops program.
Acked-by: Matthieu Baerts <[email protected]>
Signed-off-by: Nicolas Rybowski <[email protected]>
---
Notes:
v1 -> v2:
- new patch: mandatory selftests (Alexei)
tools/testing/selftests/bpf/config | 1 +
tools/testing/selftests/bpf/network_helpers.c | 37 +++++-
tools/testing/selftests/bpf/network_helpers.h | 3 +
.../testing/selftests/bpf/prog_tests/mptcp.c | 119 ++++++++++++++++++
tools/testing/selftests/bpf/progs/mptcp.c | 48 +++++++
5 files changed, 203 insertions(+), 5 deletions(-)
create mode 100644 tools/testing/selftests/bpf/prog_tests/mptcp.c
create mode 100644 tools/testing/selftests/bpf/progs/mptcp.c
diff --git a/tools/testing/selftests/bpf/config b/tools/testing/selftests/bpf/config
index 2118e23ac07a..8377836ea976 100644
--- a/tools/testing/selftests/bpf/config
+++ b/tools/testing/selftests/bpf/config
@@ -39,3 +39,4 @@ CONFIG_BPF_JIT=y
CONFIG_BPF_LSM=y
CONFIG_SECURITY=y
CONFIG_LIRC=y
+CONFIG_MPTCP=y
diff --git a/tools/testing/selftests/bpf/network_helpers.c b/tools/testing/selftests/bpf/network_helpers.c
index 12ee40284da0..85cbb683965c 100644
--- a/tools/testing/selftests/bpf/network_helpers.c
+++ b/tools/testing/selftests/bpf/network_helpers.c
@@ -14,6 +14,10 @@
#include "bpf_util.h"
#include "network_helpers.h"
+#ifndef IPPROTO_MPTCP
+#define IPPROTO_MPTCP 262
+#endif
+
#define clean_errno() (errno == 0 ? "None" : strerror(errno))
#define log_err(MSG, ...) ({ \
int __save = errno; \
@@ -66,8 +70,8 @@ static int settimeo(int fd, int timeout_ms)
#define save_errno_close(fd) ({ int __save = errno; close(fd); errno = __save; })
-int start_server(int family, int type, const char *addr_str, __u16 port,
- int timeout_ms)
+static int start_server_proto(int family, int type, int protocol,
+ const char *addr_str, __u16 port, int timeout_ms)
{
struct sockaddr_storage addr = {};
socklen_t len;
@@ -76,7 +80,7 @@ int start_server(int family, int type, const char *addr_str, __u16 port,
if (make_sockaddr(family, addr_str, port, &addr, &len))
return -1;
- fd = socket(family, type, 0);
+ fd = socket(family, type, protocol);
if (fd < 0) {
log_err("Failed to create server socket");
return -1;
@@ -104,6 +108,19 @@ int start_server(int family, int type, const char *addr_str, __u16 port,
return -1;
}
+int start_server(int family, int type, const char *addr_str, __u16 port,
+ int timeout_ms)
+{
+ return start_server_proto(family, type, 0, addr_str, port, timeout_ms);
+}
+
+int start_mptcp_server(int family, const char *addr_str, __u16 port,
+ int timeout_ms)
+{
+ return start_server_proto(family, SOCK_STREAM, IPPROTO_MPTCP, addr_str,
+ port, timeout_ms);
+}
+
int fastopen_connect(int server_fd, const char *data, unsigned int data_len,
int timeout_ms)
{
@@ -153,7 +170,7 @@ static int connect_fd_to_addr(int fd,
return 0;
}
-int connect_to_fd(int server_fd, int timeout_ms)
+static int connect_to_fd_proto(int server_fd, int protocol, int timeout_ms)
{
struct sockaddr_storage addr;
struct sockaddr_in *addr_in;
@@ -173,7 +190,7 @@ int connect_to_fd(int server_fd, int timeout_ms)
}
addr_in = (struct sockaddr_in *)&addr;
- fd = socket(addr_in->sin_family, type, 0);
+ fd = socket(addr_in->sin_family, type, protocol);
if (fd < 0) {
log_err("Failed to create client socket");
return -1;
@@ -192,6 +209,16 @@ int connect_to_fd(int server_fd, int timeout_ms)
return -1;
}
+int connect_to_fd(int server_fd, int timeout_ms)
+{
+ return connect_to_fd_proto(server_fd, 0, timeout_ms);
+}
+
+int connect_to_mptcp_fd(int server_fd, int timeout_ms)
+{
+ return connect_to_fd_proto(server_fd, IPPROTO_MPTCP, timeout_ms);
+}
+
int connect_fd_to_fd(int client_fd, int server_fd, int timeout_ms)
{
struct sockaddr_storage addr;
diff --git a/tools/testing/selftests/bpf/network_helpers.h b/tools/testing/selftests/bpf/network_helpers.h
index 7205f8afdba1..336423a789e9 100644
--- a/tools/testing/selftests/bpf/network_helpers.h
+++ b/tools/testing/selftests/bpf/network_helpers.h
@@ -35,7 +35,10 @@ extern struct ipv6_packet pkt_v6;
int start_server(int family, int type, const char *addr, __u16 port,
int timeout_ms);
+int start_mptcp_server(int family, const char *addr, __u16 port,
+ int timeout_ms);
int connect_to_fd(int server_fd, int timeout_ms);
+int connect_to_mptcp_fd(int server_fd, int timeout_ms);
int connect_fd_to_fd(int client_fd, int server_fd, int timeout_ms);
int fastopen_connect(int server_fd, const char *data, unsigned int data_len,
int timeout_ms);
diff --git a/tools/testing/selftests/bpf/prog_tests/mptcp.c b/tools/testing/selftests/bpf/prog_tests/mptcp.c
new file mode 100644
index 000000000000..0e65d64868e9
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/mptcp.c
@@ -0,0 +1,119 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <test_progs.h>
+#include "cgroup_helpers.h"
+#include "network_helpers.h"
+
+struct mptcp_storage {
+ __u32 invoked;
+ __u32 is_mptcp;
+};
+
+static int verify_sk(int map_fd, int client_fd, const char *msg, __u32 is_mptcp)
+{
+ int err = 0, cfd = client_fd;
+ struct mptcp_storage val;
+
+ /* Currently there is no easy way to get back the subflow sk from the MPTCP
+ * sk, thus we cannot access here the sk_storage associated to the subflow
+ * sk. Also, there is no sk_storage associated with the MPTCP sk since it
+ * does not trigger sockops events.
+ * We silently pass this situation at the moment.
+ */
+ if (is_mptcp == 1)
+ return 0;
+
+ if (CHECK_FAIL(bpf_map_lookup_elem(map_fd, &cfd, &val) < 0)) {
+ perror("Failed to read socket storage");
+ return -1;
+ }
+
+ if (val.invoked != 1) {
+ log_err("%s: unexpected invoked count %d != %d",
+ msg, val.invoked, 1);
+ err++;
+ }
+
+ if (val.is_mptcp != is_mptcp) {
+ log_err("%s: unexpected bpf_tcp_sock.is_mptcp %d != %d",
+ msg, val.is_mptcp, is_mptcp);
+ err++;
+ }
+
+ return err;
+}
+
+static int run_test(int cgroup_fd, int server_fd, bool is_mptcp)
+{
+ int client_fd, prog_fd, map_fd, err;
+ struct bpf_object *obj;
+ struct bpf_map *map;
+
+ struct bpf_prog_load_attr attr = {
+ .prog_type = BPF_PROG_TYPE_SOCK_OPS,
+ .file = "./mptcp.o",
+ .expected_attach_type = BPF_CGROUP_SOCK_OPS,
+ };
+
+ err = bpf_prog_load_xattr(&attr, &obj, &prog_fd);
+ if (err) {
+ log_err("Failed to load BPF object");
+ return -1;
+ }
+
+ map = bpf_map__next(NULL, obj);
+ map_fd = bpf_map__fd(map);
+
+ err = bpf_prog_attach(prog_fd, cgroup_fd, BPF_CGROUP_SOCK_OPS, 0);
+ if (err) {
+ log_err("Failed to attach BPF program");
+ goto close_bpf_object;
+ }
+
+ client_fd = is_mptcp ? connect_to_mptcp_fd(server_fd, 0) :
+ connect_to_fd(server_fd, 0);
+ if (client_fd < 0) {
+ err = -1;
+ goto close_client_fd;
+ }
+
+ err += is_mptcp ? verify_sk(map_fd, client_fd, "MPTCP subflow socket", 1) :
+ verify_sk(map_fd, client_fd, "plain TCP socket", 0);
+
+close_client_fd:
+ close(client_fd);
+
+close_bpf_object:
+ bpf_object__close(obj);
+ return err;
+}
+
+void test_mptcp(void)
+{
+ int server_fd, cgroup_fd;
+
+ cgroup_fd = test__join_cgroup("/mptcp");
+ if (CHECK_FAIL(cgroup_fd < 0))
+ return;
+
+ /* without MPTCP */
+ server_fd = start_server(AF_INET, SOCK_STREAM, NULL, 0, 0);
+ if (CHECK_FAIL(server_fd < 0))
+ goto with_mptcp;
+
+ CHECK_FAIL(run_test(cgroup_fd, server_fd, false));
+
+ close(server_fd);
+
+with_mptcp:
+ /* with MPTCP */
+ server_fd = start_mptcp_server(AF_INET, NULL, 0, 0);
+ if (CHECK_FAIL(server_fd < 0))
+ goto close_cgroup_fd;
+
+ CHECK_FAIL(run_test(cgroup_fd, server_fd, true));
+
+ close(server_fd);
+
+close_cgroup_fd:
+ close(cgroup_fd);
+}
diff --git a/tools/testing/selftests/bpf/progs/mptcp.c b/tools/testing/selftests/bpf/progs/mptcp.c
new file mode 100644
index 000000000000..be5ee8dac2b3
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/mptcp.c
@@ -0,0 +1,48 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/bpf.h>
+#include <bpf/bpf_helpers.h>
+
+char _license[] SEC("license") = "GPL";
+__u32 _version SEC("version") = 1;
+
+struct mptcp_storage {
+ __u32 invoked;
+ __u32 is_mptcp;
+};
+
+struct {
+ __uint(type, BPF_MAP_TYPE_SK_STORAGE);
+ __uint(map_flags, BPF_F_NO_PREALLOC);
+ __type(key, int);
+ __type(value, struct mptcp_storage);
+} socket_storage_map SEC(".maps");
+
+SEC("sockops")
+int _sockops(struct bpf_sock_ops *ctx)
+{
+ struct mptcp_storage *storage;
+ struct bpf_tcp_sock *tcp_sk;
+ int op = (int)ctx->op;
+ struct bpf_sock *sk;
+
+ sk = ctx->sk;
+ if (!sk)
+ return 1;
+
+ storage = bpf_sk_storage_get(&socket_storage_map, sk, 0,
+ BPF_SK_STORAGE_GET_F_CREATE);
+ if (!storage)
+ return 1;
+
+ if (op != BPF_SOCK_OPS_TCP_CONNECT_CB)
+ return 1;
+
+ tcp_sk = bpf_tcp_sock(sk);
+ if (!tcp_sk)
+ return 1;
+
+ storage->invoked++;
+ storage->is_mptcp = tcp_sk->is_mptcp;
+
+ return 1;
+}
--
2.28.0
In order to precisely identify the parent MPTCP connection of a subflow,
it is required to access the mptcp_sock's token which uniquely identify a
MPTCP connection.
This patch adds a new structure 'bpf_mptcp_sock' exposing the 'token' field
of the 'mptcp_sock' extracted from a subflow's 'tcp_sock'. It also adds the
declaration of a new BPF helper of the same name to expose the newly
defined structure in the userspace BPF API.
This is the foundation to expose more MPTCP-specific fields through BPF.
Currently, it is limited to the field 'token' of the msk but it is
easily extensible.
Acked-by: Matthieu Baerts <[email protected]>
Acked-by: Mat Martineau <[email protected]>
Signed-off-by: Nicolas Rybowski <[email protected]>
---
include/linux/bpf.h | 33 ++++++++++++++++
include/uapi/linux/bpf.h | 14 +++++++
kernel/bpf/verifier.c | 30 ++++++++++++++
net/core/filter.c | 4 ++
net/mptcp/Makefile | 2 +
net/mptcp/bpf.c | 72 ++++++++++++++++++++++++++++++++++
scripts/bpf_helpers_doc.py | 2 +
tools/include/uapi/linux/bpf.h | 14 +++++++
8 files changed, 171 insertions(+)
create mode 100644 net/mptcp/bpf.c
diff --git a/include/linux/bpf.h b/include/linux/bpf.h
index c6d9f2c444f4..6be74420f6fa 100644
--- a/include/linux/bpf.h
+++ b/include/linux/bpf.h
@@ -305,6 +305,7 @@ enum bpf_return_type {
RET_PTR_TO_SOCK_COMMON_OR_NULL, /* returns a pointer to a sock_common or NULL */
RET_PTR_TO_ALLOC_MEM_OR_NULL, /* returns a pointer to dynamically allocated memory or NULL */
RET_PTR_TO_BTF_ID_OR_NULL, /* returns a pointer to a btf_id or NULL */
+ RET_PTR_TO_MPTCP_SOCK_OR_NULL, /* returns a pointer to mptcp_sock or NULL */
};
/* eBPF function prototype used by verifier to allow BPF_CALLs from eBPF programs
@@ -385,6 +386,8 @@ enum bpf_reg_type {
PTR_TO_RDONLY_BUF_OR_NULL, /* reg points to a readonly buffer or NULL */
PTR_TO_RDWR_BUF, /* reg points to a read/write buffer */
PTR_TO_RDWR_BUF_OR_NULL, /* reg points to a read/write buffer or NULL */
+ PTR_TO_MPTCP_SOCK, /* reg points to struct mptcp_sock */
+ PTR_TO_MPTCP_SOCK_OR_NULL, /* reg points to struct mptcp_sock or NULL */
};
/* The information passed from prog-specific *_is_valid_access
@@ -1785,6 +1788,7 @@ extern const struct bpf_func_proto bpf_skc_to_tcp_timewait_sock_proto;
extern const struct bpf_func_proto bpf_skc_to_tcp_request_sock_proto;
extern const struct bpf_func_proto bpf_skc_to_udp6_sock_proto;
extern const struct bpf_func_proto bpf_copy_from_user_proto;
+extern const struct bpf_func_proto bpf_mptcp_sock_proto;
const struct bpf_func_proto *bpf_tracing_func_proto(
enum bpf_func_id func_id, const struct bpf_prog *prog);
@@ -1841,6 +1845,35 @@ struct sk_reuseport_kern {
u32 reuseport_id;
bool bind_inany;
};
+
+#ifdef CONFIG_MPTCP
+bool bpf_mptcp_sock_is_valid_access(int off, int size,
+ enum bpf_access_type type,
+ struct bpf_insn_access_aux *info);
+
+u32 bpf_mptcp_sock_convert_ctx_access(enum bpf_access_type type,
+ const struct bpf_insn *si,
+ struct bpf_insn *insn_buf,
+ struct bpf_prog *prog,
+ u32 *target_size);
+#else /* CONFIG_MPTCP */
+static inline bool bpf_mptcp_sock_is_valid_access(int off, int size,
+ enum bpf_access_type type,
+ struct bpf_insn_access_aux *info)
+{
+ return false;
+}
+
+static inline u32 bpf_mptcp_sock_convert_ctx_access(enum bpf_access_type type,
+ const struct bpf_insn *si,
+ struct bpf_insn *insn_buf,
+ struct bpf_prog *prog,
+ u32 *target_size)
+{
+ return 0;
+}
+#endif /* CONFIG_MPTCP */
+
bool bpf_tcp_sock_is_valid_access(int off, int size, enum bpf_access_type type,
struct bpf_insn_access_aux *info);
diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
index 7d179eada1c3..ee7f6fd67f47 100644
--- a/include/uapi/linux/bpf.h
+++ b/include/uapi/linux/bpf.h
@@ -3579,6 +3579,15 @@ union bpf_attr {
* the data in *dst*. This is a wrapper of **copy_from_user**\ ().
* Return
* 0 on success, or a negative error in case of failure.
+ *
+ * struct bpf_mptcp_sock *bpf_mptcp_sock(struct bpf_sock *sk)
+ * Description
+ * This helper gets a **struct bpf_mptcp_sock** pointer from a
+ * **struct bpf_sock** pointer.
+ * Return
+ * A **struct bpf_mptcp_sock** pointer on success, or **NULL** in
+ * case of failure.
+ *
*/
#define __BPF_FUNC_MAPPER(FN) \
FN(unspec), \
@@ -3730,6 +3739,7 @@ union bpf_attr {
FN(inode_storage_delete), \
FN(d_path), \
FN(copy_from_user), \
+ FN(mptcp_sock), \
/* */
/* integer value in 'imm' field of BPF_CALL instruction selects which helper
@@ -4063,6 +4073,10 @@ struct bpf_tcp_sock {
__u32 is_mptcp; /* Is MPTCP subflow? */
};
+struct bpf_mptcp_sock {
+ __u32 token; /* msk token */
+};
+
struct bpf_sock_tuple {
union {
struct {
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 814bc6c1ad16..8a9ce121addd 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -392,6 +392,7 @@ static bool type_is_sk_pointer(enum bpf_reg_type type)
return type == PTR_TO_SOCKET ||
type == PTR_TO_SOCK_COMMON ||
type == PTR_TO_TCP_SOCK ||
+ type == PTR_TO_MPTCP_SOCK ||
type == PTR_TO_XDP_SOCK;
}
@@ -399,6 +400,7 @@ static bool reg_type_not_null(enum bpf_reg_type type)
{
return type == PTR_TO_SOCKET ||
type == PTR_TO_TCP_SOCK ||
+ type == PTR_TO_MPTCP_SOCK ||
type == PTR_TO_MAP_VALUE ||
type == PTR_TO_SOCK_COMMON;
}
@@ -409,6 +411,7 @@ static bool reg_type_may_be_null(enum bpf_reg_type type)
type == PTR_TO_SOCKET_OR_NULL ||
type == PTR_TO_SOCK_COMMON_OR_NULL ||
type == PTR_TO_TCP_SOCK_OR_NULL ||
+ type == PTR_TO_MPTCP_SOCK_OR_NULL ||
type == PTR_TO_BTF_ID_OR_NULL ||
type == PTR_TO_MEM_OR_NULL ||
type == PTR_TO_RDONLY_BUF_OR_NULL ||
@@ -427,6 +430,8 @@ static bool reg_type_may_be_refcounted_or_null(enum bpf_reg_type type)
type == PTR_TO_SOCKET_OR_NULL ||
type == PTR_TO_TCP_SOCK ||
type == PTR_TO_TCP_SOCK_OR_NULL ||
+ type == PTR_TO_MPTCP_SOCK ||
+ type == PTR_TO_MPTCP_SOCK_OR_NULL ||
type == PTR_TO_MEM ||
type == PTR_TO_MEM_OR_NULL;
}
@@ -500,6 +505,8 @@ static const char * const reg_type_str[] = {
[PTR_TO_SOCK_COMMON_OR_NULL] = "sock_common_or_null",
[PTR_TO_TCP_SOCK] = "tcp_sock",
[PTR_TO_TCP_SOCK_OR_NULL] = "tcp_sock_or_null",
+ [PTR_TO_MPTCP_SOCK] = "mptcp_sock",
+ [PTR_TO_MPTCP_SOCK_OR_NULL] = "mptcp_sock_or_null",
[PTR_TO_TP_BUFFER] = "tp_buffer",
[PTR_TO_XDP_SOCK] = "xdp_sock",
[PTR_TO_BTF_ID] = "ptr_",
@@ -2177,6 +2184,8 @@ static bool is_spillable_regtype(enum bpf_reg_type type)
case PTR_TO_SOCK_COMMON_OR_NULL:
case PTR_TO_TCP_SOCK:
case PTR_TO_TCP_SOCK_OR_NULL:
+ case PTR_TO_MPTCP_SOCK:
+ case PTR_TO_MPTCP_SOCK_OR_NULL:
case PTR_TO_XDP_SOCK:
case PTR_TO_BTF_ID:
case PTR_TO_BTF_ID_OR_NULL:
@@ -2786,6 +2795,9 @@ static int check_sock_access(struct bpf_verifier_env *env, int insn_idx,
case PTR_TO_TCP_SOCK:
valid = bpf_tcp_sock_is_valid_access(off, size, t, &info);
break;
+ case PTR_TO_MPTCP_SOCK:
+ valid = bpf_mptcp_sock_is_valid_access(off, size, t, &info);
+ break;
case PTR_TO_XDP_SOCK:
valid = bpf_xdp_sock_is_valid_access(off, size, t, &info);
break;
@@ -2944,6 +2956,9 @@ static int check_ptr_alignment(struct bpf_verifier_env *env,
case PTR_TO_TCP_SOCK:
pointer_desc = "tcp_sock ";
break;
+ case PTR_TO_MPTCP_SOCK:
+ pointer_desc = "mptcp_sock ";
+ break;
case PTR_TO_XDP_SOCK:
pointer_desc = "xdp_sock ";
break;
@@ -4997,6 +5012,10 @@ static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn
mark_reg_known_zero(env, regs, BPF_REG_0);
regs[BPF_REG_0].type = PTR_TO_TCP_SOCK_OR_NULL;
regs[BPF_REG_0].id = ++env->id_gen;
+ } else if (fn->ret_type == RET_PTR_TO_MPTCP_SOCK_OR_NULL) {
+ mark_reg_known_zero(env, regs, BPF_REG_0);
+ regs[BPF_REG_0].type = PTR_TO_MPTCP_SOCK_OR_NULL;
+ regs[BPF_REG_0].id = ++env->id_gen;
} else if (fn->ret_type == RET_PTR_TO_ALLOC_MEM_OR_NULL) {
mark_reg_known_zero(env, regs, BPF_REG_0);
regs[BPF_REG_0].type = PTR_TO_MEM_OR_NULL;
@@ -5328,6 +5347,8 @@ static int adjust_ptr_min_max_vals(struct bpf_verifier_env *env,
case PTR_TO_SOCK_COMMON_OR_NULL:
case PTR_TO_TCP_SOCK:
case PTR_TO_TCP_SOCK_OR_NULL:
+ case PTR_TO_MPTCP_SOCK:
+ case PTR_TO_MPTCP_SOCK_OR_NULL:
case PTR_TO_XDP_SOCK:
verbose(env, "R%d pointer arithmetic on %s prohibited\n",
dst, reg_type_str[ptr_reg->type]);
@@ -7048,6 +7069,8 @@ static void mark_ptr_or_null_reg(struct bpf_func_state *state,
reg->type = PTR_TO_SOCK_COMMON;
} else if (reg->type == PTR_TO_TCP_SOCK_OR_NULL) {
reg->type = PTR_TO_TCP_SOCK;
+ } else if (reg->type == PTR_TO_MPTCP_SOCK_OR_NULL) {
+ reg->type = PTR_TO_MPTCP_SOCK;
} else if (reg->type == PTR_TO_BTF_ID_OR_NULL) {
reg->type = PTR_TO_BTF_ID;
} else if (reg->type == PTR_TO_MEM_OR_NULL) {
@@ -8411,6 +8434,8 @@ static bool regsafe(struct bpf_reg_state *rold, struct bpf_reg_state *rcur,
case PTR_TO_SOCK_COMMON_OR_NULL:
case PTR_TO_TCP_SOCK:
case PTR_TO_TCP_SOCK_OR_NULL:
+ case PTR_TO_MPTCP_SOCK:
+ case PTR_TO_MPTCP_SOCK_OR_NULL:
case PTR_TO_XDP_SOCK:
/* Only valid matches are exact, which memcmp() above
* would have accepted
@@ -8938,6 +8963,8 @@ static bool reg_type_mismatch_ok(enum bpf_reg_type type)
case PTR_TO_SOCK_COMMON_OR_NULL:
case PTR_TO_TCP_SOCK:
case PTR_TO_TCP_SOCK_OR_NULL:
+ case PTR_TO_MPTCP_SOCK:
+ case PTR_TO_MPTCP_SOCK_OR_NULL:
case PTR_TO_XDP_SOCK:
case PTR_TO_BTF_ID:
case PTR_TO_BTF_ID_OR_NULL:
@@ -10076,6 +10103,9 @@ static int convert_ctx_accesses(struct bpf_verifier_env *env)
case PTR_TO_TCP_SOCK:
convert_ctx_access = bpf_tcp_sock_convert_ctx_access;
break;
+ case PTR_TO_MPTCP_SOCK:
+ convert_ctx_access = bpf_mptcp_sock_convert_ctx_access;
+ break;
case PTR_TO_XDP_SOCK:
convert_ctx_access = bpf_xdp_sock_convert_ctx_access;
break;
diff --git a/net/core/filter.c b/net/core/filter.c
index dab48528dceb..a9d964e4764e 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -6890,6 +6890,10 @@ sock_ops_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
case BPF_FUNC_tcp_sock:
return &bpf_tcp_sock_proto;
#endif /* CONFIG_INET */
+#ifdef CONFIG_MPTCP
+ case BPF_FUNC_mptcp_sock:
+ return &bpf_mptcp_sock_proto;
+#endif /* CONFIG_MPTCP */
default:
return bpf_base_func_proto(func_id);
}
diff --git a/net/mptcp/Makefile b/net/mptcp/Makefile
index a611968be4d7..aae2ede220ed 100644
--- a/net/mptcp/Makefile
+++ b/net/mptcp/Makefile
@@ -10,3 +10,5 @@ obj-$(CONFIG_INET_MPTCP_DIAG) += mptcp_diag.o
mptcp_crypto_test-objs := crypto_test.o
mptcp_token_test-objs := token_test.o
obj-$(CONFIG_MPTCP_KUNIT_TESTS) += mptcp_crypto_test.o mptcp_token_test.o
+
+obj-$(CONFIG_BPF) += bpf.o
diff --git a/net/mptcp/bpf.c b/net/mptcp/bpf.c
new file mode 100644
index 000000000000..5332469fbb28
--- /dev/null
+++ b/net/mptcp/bpf.c
@@ -0,0 +1,72 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Multipath TCP
+ *
+ * Copyright (c) 2020, Tessares SA.
+ *
+ * Author: Nicolas Rybowski <[email protected]>
+ *
+ */
+
+#include <linux/bpf.h>
+
+#include "protocol.h"
+
+bool bpf_mptcp_sock_is_valid_access(int off, int size, enum bpf_access_type type,
+ struct bpf_insn_access_aux *info)
+{
+ if (off < 0 || off >= offsetofend(struct bpf_mptcp_sock, token))
+ return false;
+
+ if (off % size != 0)
+ return false;
+
+ switch (off) {
+ default:
+ return size == sizeof(__u32);
+ }
+}
+
+u32 bpf_mptcp_sock_convert_ctx_access(enum bpf_access_type type,
+ const struct bpf_insn *si,
+ struct bpf_insn *insn_buf,
+ struct bpf_prog *prog, u32 *target_size)
+{
+ struct bpf_insn *insn = insn_buf;
+
+#define BPF_MPTCP_SOCK_GET_COMMON(FIELD) \
+ do { \
+ BUILD_BUG_ON(sizeof_field(struct mptcp_sock, FIELD) > \
+ sizeof_field(struct bpf_mptcp_sock, FIELD)); \
+ *insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(struct mptcp_sock, FIELD), \
+ si->dst_reg, si->src_reg, \
+ offsetof(struct mptcp_sock, FIELD)); \
+ } while (0)
+
+ if (insn > insn_buf)
+ return insn - insn_buf;
+
+ switch (si->off) {
+ case offsetof(struct bpf_mptcp_sock, token):
+ BPF_MPTCP_SOCK_GET_COMMON(token);
+ break;
+ }
+
+ return insn - insn_buf;
+}
+
+BPF_CALL_1(bpf_mptcp_sock, struct sock *, sk)
+{
+ if (sk_fullsock(sk) && sk->sk_protocol == IPPROTO_TCP && sk_is_mptcp(sk)) {
+ struct mptcp_subflow_context *mptcp_sfc = mptcp_subflow_ctx(sk);
+
+ return (unsigned long)mptcp_sfc->conn;
+ }
+ return (unsigned long)NULL;
+}
+
+const struct bpf_func_proto bpf_mptcp_sock_proto = {
+ .func = bpf_mptcp_sock,
+ .gpl_only = false,
+ .ret_type = RET_PTR_TO_MPTCP_SOCK_OR_NULL,
+ .arg1_type = ARG_PTR_TO_SOCK_COMMON,
+};
diff --git a/scripts/bpf_helpers_doc.py b/scripts/bpf_helpers_doc.py
index 08388173973f..5c3e7aad8faa 100755
--- a/scripts/bpf_helpers_doc.py
+++ b/scripts/bpf_helpers_doc.py
@@ -428,6 +428,7 @@ class PrinterHelpers(Printer):
'struct tcp_request_sock',
'struct udp6_sock',
'struct task_struct',
+ 'struct bpf_mptcp_sock',
'struct __sk_buff',
'struct sk_msg_md',
@@ -474,6 +475,7 @@ class PrinterHelpers(Printer):
'struct udp6_sock',
'struct task_struct',
'struct path',
+ 'struct bpf_mptcp_sock',
}
mapped_types = {
'u8': '__u8',
diff --git a/tools/include/uapi/linux/bpf.h b/tools/include/uapi/linux/bpf.h
index 7d179eada1c3..ee7f6fd67f47 100644
--- a/tools/include/uapi/linux/bpf.h
+++ b/tools/include/uapi/linux/bpf.h
@@ -3579,6 +3579,15 @@ union bpf_attr {
* the data in *dst*. This is a wrapper of **copy_from_user**\ ().
* Return
* 0 on success, or a negative error in case of failure.
+ *
+ * struct bpf_mptcp_sock *bpf_mptcp_sock(struct bpf_sock *sk)
+ * Description
+ * This helper gets a **struct bpf_mptcp_sock** pointer from a
+ * **struct bpf_sock** pointer.
+ * Return
+ * A **struct bpf_mptcp_sock** pointer on success, or **NULL** in
+ * case of failure.
+ *
*/
#define __BPF_FUNC_MAPPER(FN) \
FN(unspec), \
@@ -3730,6 +3739,7 @@ union bpf_attr {
FN(inode_storage_delete), \
FN(d_path), \
FN(copy_from_user), \
+ FN(mptcp_sock), \
/* */
/* integer value in 'imm' field of BPF_CALL instruction selects which helper
@@ -4063,6 +4073,10 @@ struct bpf_tcp_sock {
__u32 is_mptcp; /* Is MPTCP subflow? */
};
+struct bpf_mptcp_sock {
+ __u32 token; /* msk token */
+};
+
struct bpf_sock_tuple {
union {
struct {
--
2.28.0
Previously it was not possible to make a distinction between plain TCP
sockets and MPTCP subflow sockets on the BPF_PROG_TYPE_SOCK_OPS hook.
This patch series now enables a fine control of subflow sockets. In its
current state, it allows to put different sockopt on each subflow from a
same MPTCP connection (socket mark, TCP congestion algorithm, ...) using
BPF programs.
It should also be the basis of exposing MPTCP-specific fields through BPF.
v1 -> v2:
- add basic mandatory selftests for the new helper and is_mptcp field (Alexei)
- rebase on latest bpf-next
Nicolas Rybowski (5):
bpf: expose is_mptcp flag to bpf_tcp_sock
mptcp: attach subflow socket to parent cgroup
bpf: add 'bpf_mptcp_sock' structure and helper
bpf: selftests: add MPTCP test base
bpf: selftests: add bpf_mptcp_sock() verifier tests
include/linux/bpf.h | 33 +++++
include/uapi/linux/bpf.h | 15 +++
kernel/bpf/verifier.c | 30 +++++
net/core/filter.c | 13 +-
net/mptcp/Makefile | 2 +
net/mptcp/bpf.c | 72 +++++++++++
net/mptcp/subflow.c | 27 ++++
scripts/bpf_helpers_doc.py | 2 +
tools/include/uapi/linux/bpf.h | 15 +++
tools/testing/selftests/bpf/config | 1 +
tools/testing/selftests/bpf/network_helpers.c | 37 +++++-
tools/testing/selftests/bpf/network_helpers.h | 3 +
.../testing/selftests/bpf/prog_tests/mptcp.c | 119 ++++++++++++++++++
tools/testing/selftests/bpf/progs/mptcp.c | 48 +++++++
tools/testing/selftests/bpf/verifier/sock.c | 63 ++++++++++
15 files changed, 474 insertions(+), 6 deletions(-)
create mode 100644 net/mptcp/bpf.c
create mode 100644 tools/testing/selftests/bpf/prog_tests/mptcp.c
create mode 100644 tools/testing/selftests/bpf/progs/mptcp.c
--
2.28.0
It has been observed that the kernel sockets created for the subflows
(except the first one) are not in the same cgroup as their parents.
That's because the additional subflows are created by kernel workers.
This is a problem with eBPF programs attached to the parent's
cgroup won't be executed for the children. But also with any other features
of CGroup linked to a sk.
This patch fixes this behaviour.
As the subflow sockets are created by the kernel, we can't use
'mem_cgroup_sk_alloc' because of the current context being the one of the
kworker. This is why we have to do low level memcg manipulation, if
required.
Suggested-by: Matthieu Baerts <[email protected]>
Suggested-by: Paolo Abeni <[email protected]>
Acked-by: Matthieu Baerts <[email protected]>
Reviewed-by: Mat Martineau <[email protected]>
Signed-off-by: Nicolas Rybowski <[email protected]>
---
net/mptcp/subflow.c | 27 +++++++++++++++++++++++++++
1 file changed, 27 insertions(+)
diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
index e8cac2655c82..535a3f9f8cfc 100644
--- a/net/mptcp/subflow.c
+++ b/net/mptcp/subflow.c
@@ -1130,6 +1130,30 @@ int __mptcp_subflow_connect(struct sock *sk, int ifindex,
return err;
}
+static void mptcp_attach_cgroup(struct sock *parent, struct sock *child)
+{
+#ifdef CONFIG_SOCK_CGROUP_DATA
+ struct sock_cgroup_data *parent_skcd = &parent->sk_cgrp_data,
+ *child_skcd = &child->sk_cgrp_data;
+
+ /* only the additional subflows created by kworkers have to be modified */
+ if (cgroup_id(sock_cgroup_ptr(parent_skcd)) !=
+ cgroup_id(sock_cgroup_ptr(child_skcd))) {
+#ifdef CONFIG_MEMCG
+ struct mem_cgroup *memcg = parent->sk_memcg;
+
+ mem_cgroup_sk_free(child);
+ if (memcg && css_tryget(&memcg->css))
+ child->sk_memcg = memcg;
+#endif /* CONFIG_MEMCG */
+
+ cgroup_sk_free(child_skcd);
+ *child_skcd = *parent_skcd;
+ cgroup_sk_clone(child_skcd);
+ }
+#endif /* CONFIG_SOCK_CGROUP_DATA */
+}
+
int mptcp_subflow_create_socket(struct sock *sk, struct socket **new_sock)
{
struct mptcp_subflow_context *subflow;
@@ -1150,6 +1174,9 @@ int mptcp_subflow_create_socket(struct sock *sk, struct socket **new_sock)
lock_sock(sf->sk);
+ /* the newly created socket has to be in the same cgroup as its parent */
+ mptcp_attach_cgroup(sk, sf->sk);
+
/* kernel sockets do not by default acquire net ref, but TCP timer
* needs it.
*/
--
2.28.0
This patch adds verifier side tests for the new bpf_mptcp_sock() helper.
Here are the new tests :
- NULL bpf_sock is correctly handled
- We cannot access a field from bpf_mptcp_sock if the latter is NULL
- We can access a field from bpf_mptcp_sock if the latter is not NULL
- We cannot modify a field from bpf_mptcp_sock.
Note that "token" is currently the only field in bpf_mptcp_sock.
Currently, there is no easy way to test the token field since we cannot
get back the mptcp_sock in userspace, this could be a future amelioration.
Acked-by: Matthieu Baerts <[email protected]>
Signed-off-by: Nicolas Rybowski <[email protected]>
---
Notes:
v1 -> v2:
- new patch: mandatory selftests (Alexei)
tools/testing/selftests/bpf/verifier/sock.c | 63 +++++++++++++++++++++
1 file changed, 63 insertions(+)
diff --git a/tools/testing/selftests/bpf/verifier/sock.c b/tools/testing/selftests/bpf/verifier/sock.c
index b1aac2641498..9ce7c7ec3b5e 100644
--- a/tools/testing/selftests/bpf/verifier/sock.c
+++ b/tools/testing/selftests/bpf/verifier/sock.c
@@ -631,3 +631,66 @@
.prog_type = BPF_PROG_TYPE_SK_REUSEPORT,
.result = ACCEPT,
},
+{
+ "bpf_mptcp_sock(skops->sk): no !skops->sk check",
+ .insns = {
+ BPF_LDX_MEM(BPF_DW, BPF_REG_1, BPF_REG_1, offsetof(struct bpf_sock_ops, sk)),
+ BPF_EMIT_CALL(BPF_FUNC_mptcp_sock),
+ BPF_MOV64_IMM(BPF_REG_0, 0),
+ BPF_EXIT_INSN(),
+ },
+ .prog_type = BPF_PROG_TYPE_SOCK_OPS,
+ .result = REJECT,
+ .errstr = "type=sock_or_null expected=sock_common",
+},
+{
+ "bpf_mptcp_sock(skops->sk): no NULL check on ret",
+ .insns = {
+ BPF_LDX_MEM(BPF_DW, BPF_REG_1, BPF_REG_1, offsetof(struct bpf_sock_ops, sk)),
+ BPF_JMP_IMM(BPF_JNE, BPF_REG_1, 0, 2),
+ BPF_MOV64_IMM(BPF_REG_0, 0),
+ BPF_EXIT_INSN(),
+ BPF_EMIT_CALL(BPF_FUNC_mptcp_sock),
+ BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_0, offsetof(struct bpf_mptcp_sock, token)),
+ BPF_MOV64_IMM(BPF_REG_0, 0),
+ BPF_EXIT_INSN(),
+ },
+ .prog_type = BPF_PROG_TYPE_SOCK_OPS,
+ .result = REJECT,
+ .errstr = "invalid mem access 'mptcp_sock_or_null'",
+},
+{
+ "bpf_mptcp_sock(skops->sk): msk->token",
+ .insns = {
+ BPF_LDX_MEM(BPF_DW, BPF_REG_1, BPF_REG_1, offsetof(struct bpf_sock_ops, sk)),
+ BPF_JMP_IMM(BPF_JNE, BPF_REG_1, 0, 2),
+ BPF_MOV64_IMM(BPF_REG_0, 0),
+ BPF_EXIT_INSN(),
+ BPF_EMIT_CALL(BPF_FUNC_mptcp_sock),
+ BPF_JMP_IMM(BPF_JNE, BPF_REG_0, 0, 1),
+ BPF_EXIT_INSN(),
+ BPF_LDX_MEM(BPF_W, BPF_REG_0, BPF_REG_0, offsetof(struct bpf_mptcp_sock, token)),
+ BPF_MOV64_IMM(BPF_REG_0, 0),
+ BPF_EXIT_INSN(),
+ },
+ .prog_type = BPF_PROG_TYPE_SOCK_OPS,
+ .result = ACCEPT,
+},
+{
+ "bpf_mptcp_sock(skops->sk): msk->token cannot be modified",
+ .insns = {
+ BPF_LDX_MEM(BPF_DW, BPF_REG_1, BPF_REG_1, offsetof(struct bpf_sock_ops, sk)),
+ BPF_JMP_IMM(BPF_JNE, BPF_REG_1, 0, 2),
+ BPF_MOV64_IMM(BPF_REG_0, 0),
+ BPF_EXIT_INSN(),
+ BPF_EMIT_CALL(BPF_FUNC_mptcp_sock),
+ BPF_JMP_IMM(BPF_JNE, BPF_REG_0, 0, 1),
+ BPF_EXIT_INSN(),
+ BPF_ST_MEM(BPF_W, BPF_REG_0, offsetof(struct bpf_mptcp_sock, token), 0x2a),
+ BPF_MOV64_IMM(BPF_REG_0, 0),
+ BPF_EXIT_INSN(),
+ },
+ .prog_type = BPF_PROG_TYPE_SOCK_OPS,
+ .result = REJECT,
+ .errstr = "cannot write into mptcp_sock",
+},
--
2.28.0
On Fri, Sep 11, 2020 at 8:02 AM Nicolas Rybowski
<[email protected]> wrote:
>
> This patch adds a base for MPTCP specific tests.
>
> It is currently limited to the is_mptcp field in case of plain TCP
> connection because for the moment there is no easy way to get the subflow
> sk from a msk in userspace. This implies that we cannot lookup the
> sk_storage attached to the subflow sk in the sockops program.
>
> Acked-by: Matthieu Baerts <[email protected]>
> Signed-off-by: Nicolas Rybowski <[email protected]>
Acked-by: Song Liu <[email protected]>
With some nitpicks below.
> ---
>
> Notes:
> v1 -> v2:
> - new patch: mandatory selftests (Alexei)
>
[...]
> int timeout_ms);
> diff --git a/tools/testing/selftests/bpf/prog_tests/mptcp.c b/tools/testing/selftests/bpf/prog_tests/mptcp.c
> new file mode 100644
> index 000000000000..0e65d64868e9
> --- /dev/null
> +++ b/tools/testing/selftests/bpf/prog_tests/mptcp.c
> @@ -0,0 +1,119 @@
> +// SPDX-License-Identifier: GPL-2.0
> +#include <test_progs.h>
> +#include "cgroup_helpers.h"
> +#include "network_helpers.h"
> +
> +struct mptcp_storage {
> + __u32 invoked;
> + __u32 is_mptcp;
> +};
> +
> +static int verify_sk(int map_fd, int client_fd, const char *msg, __u32 is_mptcp)
> +{
> + int err = 0, cfd = client_fd;
> + struct mptcp_storage val;
> +
> + /* Currently there is no easy way to get back the subflow sk from the MPTCP
> + * sk, thus we cannot access here the sk_storage associated to the subflow
> + * sk. Also, there is no sk_storage associated with the MPTCP sk since it
> + * does not trigger sockops events.
> + * We silently pass this situation at the moment.
> + */
> + if (is_mptcp == 1)
> + return 0;
> +
> + if (CHECK_FAIL(bpf_map_lookup_elem(map_fd, &cfd, &val) < 0)) {
> + perror("Failed to read socket storage");
Maybe simplify this with CHECK(), which contains a customized error message?
Same for some other calls.
> + return -1;
> + }
> +
> + if (val.invoked != 1) {
> + log_err("%s: unexpected invoked count %d != %d",
> + msg, val.invoked, 1);
> + err++;
> + }
> +
> + if (val.is_mptcp != is_mptcp) {
> + log_err("%s: unexpected bpf_tcp_sock.is_mptcp %d != %d",
> + msg, val.is_mptcp, is_mptcp);
> + err++;
> + }
> +
> + return err;
> +}
> +
> +static int run_test(int cgroup_fd, int server_fd, bool is_mptcp)
[...]
> +
> + client_fd = is_mptcp ? connect_to_mptcp_fd(server_fd, 0) :
> + connect_to_fd(server_fd, 0);
> + if (client_fd < 0) {
> + err = -1;
> + goto close_client_fd;
This should be "goto close_bpf_object;", and we don't really need the label
close_client_fd.
> + }
> +
> + err += is_mptcp ? verify_sk(map_fd, client_fd, "MPTCP subflow socket", 1) :
It doesn't really change the logic, but I guess we only need "err = xxx"?
> + verify_sk(map_fd, client_fd, "plain TCP socket", 0);
> +
> +close_client_fd:
> + close(client_fd);
> +
> +close_bpf_object:
> + bpf_object__close(obj);
> + return err;
> +}
> +
On Fri, Sep 11, 2020 at 9:46 AM Nicolas Rybowski
<[email protected]> wrote:
>
> This patch adds verifier side tests for the new bpf_mptcp_sock() helper.
>
> Here are the new tests :
> - NULL bpf_sock is correctly handled
> - We cannot access a field from bpf_mptcp_sock if the latter is NULL
> - We can access a field from bpf_mptcp_sock if the latter is not NULL
> - We cannot modify a field from bpf_mptcp_sock.
>
> Note that "token" is currently the only field in bpf_mptcp_sock.
>
> Currently, there is no easy way to test the token field since we cannot
> get back the mptcp_sock in userspace, this could be a future amelioration.
>
> Acked-by: Matthieu Baerts <[email protected]>
> Signed-off-by: Nicolas Rybowski <[email protected]>
Acked-by: Song Liu <[email protected]>
On Fri, Sep 11, 2020 at 8:07 AM Nicolas Rybowski
<[email protected]> wrote:
>
> is_mptcp is a field from struct tcp_sock used to indicate that the
> current tcp_sock is part of the MPTCP protocol.
>
> In this protocol, a first socket (mptcp_sock) is created with
> sk_protocol set to IPPROTO_MPTCP (=262) for control purpose but it
> isn't directly on the wire. This is the role of the subflow (kernel)
> sockets which are classical tcp_sock with sk_protocol set to
> IPPROTO_TCP. The only way to differentiate such sockets from plain TCP
> sockets is the is_mptcp field from tcp_sock.
>
> Such an exposure in BPF is thus required to be able to differentiate
> plain TCP sockets from MPTCP subflow sockets in BPF_PROG_TYPE_SOCK_OPS
> programs.
>
> The choice has been made to silently pass the case when CONFIG_MPTCP is
> unset by defaulting is_mptcp to 0 in order to make BPF independent of
> the MPTCP configuration. Another solution is to make the verifier fail
> in 'bpf_tcp_sock_is_valid_ctx_access' but this will add an additional
> '#ifdef CONFIG_MPTCP' in the BPF code and a same injected BPF program
> will not run if MPTCP is not set.
>
> An example use-case is provided in
> https://github.com/multipath-tcp/mptcp_net-next/tree/scripts/bpf/examples
>
> Suggested-by: Matthieu Baerts <[email protected]>
> Acked-by: Matthieu Baerts <[email protected]>
> Acked-by: Mat Martineau <[email protected]>
> Signed-off-by: Nicolas Rybowski <[email protected]>
> ---
> include/uapi/linux/bpf.h | 1 +
> net/core/filter.c | 9 ++++++++-
> tools/include/uapi/linux/bpf.h | 1 +
> 3 files changed, 10 insertions(+), 1 deletion(-)
>
> diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> index 7dd314176df7..7d179eada1c3 100644
> --- a/include/uapi/linux/bpf.h
> +++ b/include/uapi/linux/bpf.h
> @@ -4060,6 +4060,7 @@ struct bpf_tcp_sock {
> __u32 delivered; /* Total data packets delivered incl. rexmits */
> __u32 delivered_ce; /* Like the above but only ECE marked packets */
> __u32 icsk_retransmits; /* Number of unrecovered [RTO] timeouts */
> + __u32 is_mptcp; /* Is MPTCP subflow? */
Shall we have an __u32 flags, and make is_mptcp a bit of it?
Thanks,
Song
[...]
On Fri, Sep 11, 2020 at 9:43 AM Nicolas Rybowski
<[email protected]> wrote:
>
> It has been observed that the kernel sockets created for the subflows
> (except the first one) are not in the same cgroup as their parents.
> That's because the additional subflows are created by kernel workers.
>
> This is a problem with eBPF programs attached to the parent's
> cgroup won't be executed for the children. But also with any other features
> of CGroup linked to a sk.
>
> This patch fixes this behaviour.
>
> As the subflow sockets are created by the kernel, we can't use
> 'mem_cgroup_sk_alloc' because of the current context being the one of the
> kworker. This is why we have to do low level memcg manipulation, if
> required.
>
> Suggested-by: Matthieu Baerts <[email protected]>
> Suggested-by: Paolo Abeni <[email protected]>
> Acked-by: Matthieu Baerts <[email protected]>
> Reviewed-by: Mat Martineau <[email protected]>
> Signed-off-by: Nicolas Rybowski <[email protected]>
Acked-by: Song Liu <[email protected]>
[...]
On Fri, Sep 11, 2020 at 8:15 AM Nicolas Rybowski
<[email protected]> wrote:
>
> In order to precisely identify the parent MPTCP connection of a subflow,
> it is required to access the mptcp_sock's token which uniquely identify a
> MPTCP connection.
>
> This patch adds a new structure 'bpf_mptcp_sock' exposing the 'token' field
> of the 'mptcp_sock' extracted from a subflow's 'tcp_sock'. It also adds the
> declaration of a new BPF helper of the same name to expose the newly
> defined structure in the userspace BPF API.
>
> This is the foundation to expose more MPTCP-specific fields through BPF.
>
> Currently, it is limited to the field 'token' of the msk but it is
> easily extensible.
>
> Acked-by: Matthieu Baerts <[email protected]>
> Acked-by: Mat Martineau <[email protected]>
> Signed-off-by: Nicolas Rybowski <[email protected]>
Acked-by: Song Liu <[email protected]>
On Mon, Sep 14, 2020 at 11:21 AM Song Liu <[email protected]> wrote:
>
> On Fri, Sep 11, 2020 at 8:07 AM Nicolas Rybowski
> <[email protected]> wrote:
> >
> > is_mptcp is a field from struct tcp_sock used to indicate that the
> > current tcp_sock is part of the MPTCP protocol.
> >
> > In this protocol, a first socket (mptcp_sock) is created with
> > sk_protocol set to IPPROTO_MPTCP (=262) for control purpose but it
> > isn't directly on the wire. This is the role of the subflow (kernel)
> > sockets which are classical tcp_sock with sk_protocol set to
> > IPPROTO_TCP. The only way to differentiate such sockets from plain TCP
> > sockets is the is_mptcp field from tcp_sock.
> >
> > Such an exposure in BPF is thus required to be able to differentiate
> > plain TCP sockets from MPTCP subflow sockets in BPF_PROG_TYPE_SOCK_OPS
> > programs.
> >
> > The choice has been made to silently pass the case when CONFIG_MPTCP is
> > unset by defaulting is_mptcp to 0 in order to make BPF independent of
> > the MPTCP configuration. Another solution is to make the verifier fail
> > in 'bpf_tcp_sock_is_valid_ctx_access' but this will add an additional
> > '#ifdef CONFIG_MPTCP' in the BPF code and a same injected BPF program
> > will not run if MPTCP is not set.
> >
> > An example use-case is provided in
> > https://github.com/multipath-tcp/mptcp_net-next/tree/scripts/bpf/examples
> >
> > Suggested-by: Matthieu Baerts <[email protected]>
> > Acked-by: Matthieu Baerts <[email protected]>
> > Acked-by: Mat Martineau <[email protected]>
> > Signed-off-by: Nicolas Rybowski <[email protected]>
> > ---
> > include/uapi/linux/bpf.h | 1 +
> > net/core/filter.c | 9 ++++++++-
> > tools/include/uapi/linux/bpf.h | 1 +
> > 3 files changed, 10 insertions(+), 1 deletion(-)
> >
> > diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> > index 7dd314176df7..7d179eada1c3 100644
> > --- a/include/uapi/linux/bpf.h
> > +++ b/include/uapi/linux/bpf.h
> > @@ -4060,6 +4060,7 @@ struct bpf_tcp_sock {
> > __u32 delivered; /* Total data packets delivered incl. rexmits */
> > __u32 delivered_ce; /* Like the above but only ECE marked packets */
> > __u32 icsk_retransmits; /* Number of unrecovered [RTO] timeouts */
> > + __u32 is_mptcp; /* Is MPTCP subflow? */
>
> Shall we have an __u32 flags, and make is_mptcp a bit of it?
>
Bitfields are slow and more annoying to rewrite in verifier, so having
an __u32 field is actually good.
> Thanks,
> Song
> [...]
> On Sep 15, 2020, at 9:35 AM, Nicolas Rybowski <[email protected]> wrote:
>
> Hi Song,
>
> Thanks for the feedback !
>
> On Mon, Sep 14, 2020 at 8:07 PM Song Liu <[email protected]> wrote:
>>
>> On Fri, Sep 11, 2020 at 8:02 AM Nicolas Rybowski
>> <[email protected]> wrote:
>>>
>>> This patch adds a base for MPTCP specific tests.
>>>
>>> It is currently limited to the is_mptcp field in case of plain TCP
>>> connection because for the moment there is no easy way to get the subflow
>>> sk from a msk in userspace. This implies that we cannot lookup the
>>> sk_storage attached to the subflow sk in the sockops program.
>>>
>>> Acked-by: Matthieu Baerts <[email protected]>
>>> Signed-off-by: Nicolas Rybowski <[email protected]>
>>
>> Acked-by: Song Liu <[email protected]>
>>
>> With some nitpicks below.
>>
>>> ---
>>>
>>> Notes:
>>> v1 -> v2:
>>> - new patch: mandatory selftests (Alexei)
>>>
>> [...]
>>> int timeout_ms);
>>> diff --git a/tools/testing/selftests/bpf/prog_tests/mptcp.c b/tools/testing/selftests/bpf/prog_tests/mptcp.c
>>> new file mode 100644
>>> index 000000000000..0e65d64868e9
>>> --- /dev/null
>>> +++ b/tools/testing/selftests/bpf/prog_tests/mptcp.c
>>> @@ -0,0 +1,119 @@
>>> +// SPDX-License-Identifier: GPL-2.0
>>> +#include <test_progs.h>
>>> +#include "cgroup_helpers.h"
>>> +#include "network_helpers.h"
>>> +
>>> +struct mptcp_storage {
>>> + __u32 invoked;
>>> + __u32 is_mptcp;
>>> +};
>>> +
>>> +static int verify_sk(int map_fd, int client_fd, const char *msg, __u32 is_mptcp)
>>> +{
>>> + int err = 0, cfd = client_fd;
>>> + struct mptcp_storage val;
>>> +
>>> + /* Currently there is no easy way to get back the subflow sk from the MPTCP
>>> + * sk, thus we cannot access here the sk_storage associated to the subflow
>>> + * sk. Also, there is no sk_storage associated with the MPTCP sk since it
>>> + * does not trigger sockops events.
>>> + * We silently pass this situation at the moment.
>>> + */
>>> + if (is_mptcp == 1)
>>> + return 0;
>>> +
>>> + if (CHECK_FAIL(bpf_map_lookup_elem(map_fd, &cfd, &val) < 0)) {
>>> + perror("Failed to read socket storage");
>>
>> Maybe simplify this with CHECK(), which contains a customized error message?
>> Same for some other calls.
>>
>
> The whole logic here is strongly inspired from prog_tests/tcp_rtt.c
> where CHECK_FAIL is used.
> Also the CHECK macro will print a PASS message on successful map
> lookup, which is not expected at this point of the tests.
> I think it would be more interesting to leave it as it is to keep a
> cohesion between TCP and MPTCP selftests. What do you think?
I guess CHECK_FAIL makes sense when we don't need the PASS message.
Let's keep this part as-is then.
Thanks,
Song
Hi Song,
Thanks for the feedback !
On Mon, Sep 14, 2020 at 8:07 PM Song Liu <[email protected]> wrote:
>
> On Fri, Sep 11, 2020 at 8:02 AM Nicolas Rybowski
> <[email protected]> wrote:
> >
> > This patch adds a base for MPTCP specific tests.
> >
> > It is currently limited to the is_mptcp field in case of plain TCP
> > connection because for the moment there is no easy way to get the subflow
> > sk from a msk in userspace. This implies that we cannot lookup the
> > sk_storage attached to the subflow sk in the sockops program.
> >
> > Acked-by: Matthieu Baerts <[email protected]>
> > Signed-off-by: Nicolas Rybowski <[email protected]>
>
> Acked-by: Song Liu <[email protected]>
>
> With some nitpicks below.
>
> > ---
> >
> > Notes:
> > v1 -> v2:
> > - new patch: mandatory selftests (Alexei)
> >
> [...]
> > int timeout_ms);
> > diff --git a/tools/testing/selftests/bpf/prog_tests/mptcp.c b/tools/testing/selftests/bpf/prog_tests/mptcp.c
> > new file mode 100644
> > index 000000000000..0e65d64868e9
> > --- /dev/null
> > +++ b/tools/testing/selftests/bpf/prog_tests/mptcp.c
> > @@ -0,0 +1,119 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +#include <test_progs.h>
> > +#include "cgroup_helpers.h"
> > +#include "network_helpers.h"
> > +
> > +struct mptcp_storage {
> > + __u32 invoked;
> > + __u32 is_mptcp;
> > +};
> > +
> > +static int verify_sk(int map_fd, int client_fd, const char *msg, __u32 is_mptcp)
> > +{
> > + int err = 0, cfd = client_fd;
> > + struct mptcp_storage val;
> > +
> > + /* Currently there is no easy way to get back the subflow sk from the MPTCP
> > + * sk, thus we cannot access here the sk_storage associated to the subflow
> > + * sk. Also, there is no sk_storage associated with the MPTCP sk since it
> > + * does not trigger sockops events.
> > + * We silently pass this situation at the moment.
> > + */
> > + if (is_mptcp == 1)
> > + return 0;
> > +
> > + if (CHECK_FAIL(bpf_map_lookup_elem(map_fd, &cfd, &val) < 0)) {
> > + perror("Failed to read socket storage");
>
> Maybe simplify this with CHECK(), which contains a customized error message?
> Same for some other calls.
>
The whole logic here is strongly inspired from prog_tests/tcp_rtt.c
where CHECK_FAIL is used.
Also the CHECK macro will print a PASS message on successful map
lookup, which is not expected at this point of the tests.
I think it would be more interesting to leave it as it is to keep a
cohesion between TCP and MPTCP selftests. What do you think?
If there are no objections, I will send a v3 with the other requested
changes and a rebase on the latest bpf-next.
> > + return -1;
> > + }
> > +
> > + if (val.invoked != 1) {
> > + log_err("%s: unexpected invoked count %d != %d",
> > + msg, val.invoked, 1);
> > + err++;
> > + }
> > +
> > + if (val.is_mptcp != is_mptcp) {
> > + log_err("%s: unexpected bpf_tcp_sock.is_mptcp %d != %d",
> > + msg, val.is_mptcp, is_mptcp);
> > + err++;
> > + }
> > +
> > + return err;
> > +}
> > +
> > +static int run_test(int cgroup_fd, int server_fd, bool is_mptcp)
> [...]
>
> > +
> > + client_fd = is_mptcp ? connect_to_mptcp_fd(server_fd, 0) :
> > + connect_to_fd(server_fd, 0);
> > + if (client_fd < 0) {
> > + err = -1;
> > + goto close_client_fd;
>
> This should be "goto close_bpf_object;", and we don't really need the label
> close_client_fd.
>
> > + }
> > +
> > + err += is_mptcp ? verify_sk(map_fd, client_fd, "MPTCP subflow socket", 1) :
>
> It doesn't really change the logic, but I guess we only need "err = xxx"?
>
> > + verify_sk(map_fd, client_fd, "plain TCP socket", 0);
> > +
> > +close_client_fd:
> > + close(client_fd);
> > +
> > +close_bpf_object:
> > + bpf_object__close(obj);
> > + return err;
> > +}
> > +