bpf: add 'bpf_mptcp_sock' structure and helper

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

@@ -313,6 +313,7 @@ enum bpf_return_type {
 	RET_PTR_TO_MEM_OR_BTF_ID_OR_NULL, /* returns a pointer to a valid memory or a btf_id or NULL */
 	RET_PTR_TO_MEM_OR_BTF_ID,	/* returns a pointer to a valid memory or a btf_id */
 	RET_PTR_TO_BTF_ID,		/* returns a pointer to a btf_id */
+	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
@@ -412,6 +413,8 @@ enum bpf_reg_type {
 	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_PERCPU_BTF_ID,	 /* reg points to a percpu kernel variable */
+	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
@@ -1859,6 +1862,7 @@ extern const struct bpf_func_proto bpf_snprintf_btf_proto;
 extern const struct bpf_func_proto bpf_per_cpu_ptr_proto;
 extern const struct bpf_func_proto bpf_this_cpu_ptr_proto;
 extern const struct bpf_func_proto bpf_ktime_get_coarse_ns_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);
@@ -1915,6 +1919,7 @@ struct sk_reuseport_kern {
 	u32 reuseport_id;
 	bool bind_inany;
 };
+
 bool bpf_tcp_sock_is_valid_access(int off, int size, enum bpf_access_type type,
 				  struct bpf_insn_access_aux *info);
 
@@ -1965,6 +1970,34 @@ static inline u32 bpf_xdp_sock_convert_ctx_access(enum bpf_access_type type,
 }
 #endif /* CONFIG_INET */
 
+#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 */
+
 enum bpf_text_poke_type {
 	BPF_MOD_CALL,
 	BPF_MOD_JUMP,

include/uapi/linux/bpf.h

@@ -3822,6 +3822,14 @@ union bpf_attr {
  *		The **hash_algo** is returned on success,
  *		**-EOPNOTSUP** if IMA is disabled or **-EINVAL** if
  *		invalid arguments are passed.
+ *
+ * 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),			\
@@ -3986,6 +3994,7 @@ union bpf_attr {
 	FN(bprm_opts_set),		\
 	FN(ktime_get_coarse_ns),	\
 	FN(ima_inode_hash),		\
+	FN(mptcp_sock),			\
 	/* */
 
 /* integer value in 'imm' field of BPF_CALL instruction selects which helper
@@ -4324,6 +4333,10 @@ struct bpf_tcp_sock {
 	__u32 is_mptcp;		/* Is MPTCP subflow? */
 };
 
+struct bpf_mptcp_sock {
+	__u32 token;		/* msk token */
+};
+
 struct bpf_sock_tuple {
 	union {
 		struct {

kernel/bpf/verifier.c

@@ -395,13 +395,15 @@ 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;
 }
 
 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;
 }
@@ -412,6 +414,7 @@ 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 ||
@@ -430,6 +433,8 @@ 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;
 }
@@ -517,6 +522,8 @@ 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_",
@@ -2211,6 +2218,8 @@ 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:
@@ -2838,6 +2847,9 @@ 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;
@@ -2996,6 +3008,9 @@ 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;
@@ -5190,6 +5205,9 @@ static int check_helper_call(struct bpf_verifier_env *env, int func_id, int insn
 	} else if (fn->ret_type == RET_PTR_TO_TCP_SOCK_OR_NULL) {
 		mark_reg_known_zero(env, regs, BPF_REG_0);
 		regs[BPF_REG_0].type = PTR_TO_TCP_SOCK_OR_NULL;
+	} 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;
 	} 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;
@@ -5560,6 +5578,8 @@ 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]);
@@ -7343,6 +7363,8 @@ 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) {
@@ -8825,6 +8847,8 @@ 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
@@ -9352,6 +9376,8 @@ 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:
@@ -10614,6 +10640,9 @@ 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;

net/core/filter.c

@@ -7294,6 +7294,10 @@ 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_sk_base_func_proto(func_id);
 	}

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

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,
+};

scripts/bpf_helpers_doc.py

@@ -430,6 +430,7 @@ class PrinterHelpers(Printer):
             'struct tcp_request_sock',
             'struct udp6_sock',
             'struct task_struct',
+            'struct bpf_mptcp_sock',
 
             'struct __sk_buff',
             'struct sk_msg_md',
@@ -482,6 +483,7 @@ class PrinterHelpers(Printer):
             'struct path',
             'struct btf_ptr',
             'struct inode',
+            'struct bpf_mptcp_sock',
     }
     mapped_types = {
             'u8': '__u8',

tools/include/uapi/linux/bpf.h

@@ -3822,6 +3822,14 @@ union bpf_attr {
  *		The **hash_algo** is returned on success,
  *		**-EOPNOTSUP** if IMA is disabled or **-EINVAL** if
  *		invalid arguments are passed.
+ *
+ * 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),			\
@@ -3986,6 +3994,7 @@ union bpf_attr {
 	FN(bprm_opts_set),		\
 	FN(ktime_get_coarse_ns),	\
 	FN(ima_inode_hash),		\
+	FN(mptcp_sock),			\
 	/* */
 
 /* integer value in 'imm' field of BPF_CALL instruction selects which helper
@@ -4324,6 +4333,10 @@ struct bpf_tcp_sock {
 	__u32 is_mptcp;		/* Is MPTCP subflow? */
 };
 
+struct bpf_mptcp_sock {
+	__u32 token;		/* msk token */
+};
+
 struct bpf_sock_tuple {
 	union {
 		struct {