bpf: Hooks for sys_bind

== The problem ==

There is a use-case when all processes inside a cgroup should use one
single IP address on a host that has multiple IP configured.  Those
processes should use the IP for both ingress and egress, for TCP and UDP
traffic. So TCP/UDP servers should be bound to that IP to accept
incoming connections on it, and TCP/UDP clients should make outgoing
connections from that IP. It should not require changing application
code since it's often not possible.

Currently it's solved by intercepting glibc wrappers around syscalls
such as `bind(2)` and `connect(2)`. It's done by a shared library that
is preloaded for every process in a cgroup so that whenever TCP/UDP
server calls `bind(2)`, the library replaces IP in sockaddr before
passing arguments to syscall. When application calls `connect(2)` the
library transparently binds the local end of connection to that IP
(`bind(2)` with `IP_BIND_ADDRESS_NO_PORT` to avoid performance penalty).

Shared library approach is fragile though, e.g.:
* some applications clear env vars (incl. `LD_PRELOAD`);
* `/etc/ld.so.preload` doesn't help since some applications are linked
  with option `-z nodefaultlib`;
* other applications don't use glibc and there is nothing to intercept.

== The solution ==

The patch provides much more reliable in-kernel solution for the 1st
part of the problem: binding TCP/UDP servers on desired IP. It does not
depend on application environment and implementation details (whether
glibc is used or not).

It adds new eBPF program type `BPF_PROG_TYPE_CGROUP_SOCK_ADDR` and
attach types `BPF_CGROUP_INET4_BIND` and `BPF_CGROUP_INET6_BIND`
(similar to already existing `BPF_CGROUP_INET_SOCK_CREATE`).

The new program type is intended to be used with sockets (`struct sock`)
in a cgroup and provided by user `struct sockaddr`. Pointers to both of
them are parts of the context passed to programs of newly added types.

The new attach types provides hooks in `bind(2)` system call for both
IPv4 and IPv6 so that one can write a program to override IP addresses
and ports user program tries to bind to and apply such a program for
whole cgroup.

== Implementation notes ==

[1]
Separate attach types for `AF_INET` and `AF_INET6` are added
intentionally to prevent reading/writing to offsets that don't make
sense for corresponding socket family. E.g. if user passes `sockaddr_in`
it doesn't make sense to read from / write to `user_ip6[]` context
fields.

[2]
The write access to `struct bpf_sock_addr_kern` is implemented using
special field as an additional "register".

There are just two registers in `sock_addr_convert_ctx_access`: `src`
with value to write and `dst` with pointer to context that can't be
changed not to break later instructions. But the fields, allowed to
write to, are not available directly and to access them address of
corresponding pointer has to be loaded first. To get additional register
the 1st not used by `src` and `dst` one is taken, its content is saved
to `bpf_sock_addr_kern.tmp_reg`, then the register is used to load
address of pointer field, and finally the register's content is restored
from the temporary field after writing `src` value.

Signed-off-by: Andrey Ignatov <rdna@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

3 files changed, 246 insertions, 0 deletions

diff --git a/net/core/filter.c b/net/core/filter.c
index 7790fd128614..c08e5b121558 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -3699,6 +3699,20 @@ sock_filter_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 }
 
 static const struct bpf_func_proto *
+sock_addr_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
+{
+	switch (func_id) {
+	/* inet and inet6 sockets are created in a process
+	 * context so there is always a valid uid/gid
+	 */
+	case BPF_FUNC_get_current_uid_gid:
+		return &bpf_get_current_uid_gid_proto;
+	default:
+		return bpf_base_func_proto(func_id);
+	}
+}
+
+static const struct bpf_func_proto *
 sk_filter_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
 {
 	switch (func_id) {
@@ -4180,6 +4194,69 @@ void bpf_warn_invalid_xdp_action(u32 act)
 }
 EXPORT_SYMBOL_GPL(bpf_warn_invalid_xdp_action);
 
+static bool sock_addr_is_valid_access(int off, int size,
+				      enum bpf_access_type type,
+				      const struct bpf_prog *prog,
+				      struct bpf_insn_access_aux *info)
+{
+	const int size_default = sizeof(__u32);
+
+	if (off < 0 || off >= sizeof(struct bpf_sock_addr))
+		return false;
+	if (off % size != 0)
+		return false;
+
+	/* Disallow access to IPv6 fields from IPv4 contex and vise
+	 * versa.
+	 */
+	switch (off) {
+	case bpf_ctx_range(struct bpf_sock_addr, user_ip4):
+		switch (prog->expected_attach_type) {
+		case BPF_CGROUP_INET4_BIND:
+			break;
+		default:
+			return false;
+		}
+		break;
+	case bpf_ctx_range_till(struct bpf_sock_addr, user_ip6[0], user_ip6[3]):
+		switch (prog->expected_attach_type) {
+		case BPF_CGROUP_INET6_BIND:
+			break;
+		default:
+			return false;
+		}
+		break;
+	}
+
+	switch (off) {
+	case bpf_ctx_range(struct bpf_sock_addr, user_ip4):
+	case bpf_ctx_range_till(struct bpf_sock_addr, user_ip6[0], user_ip6[3]):
+		/* Only narrow read access allowed for now. */
+		if (type == BPF_READ) {
+			bpf_ctx_record_field_size(info, size_default);
+			if (!bpf_ctx_narrow_access_ok(off, size, size_default))
+				return false;
+		} else {
+			if (size != size_default)
+				return false;
+		}
+		break;
+	case bpf_ctx_range(struct bpf_sock_addr, user_port):
+		if (size != size_default)
+			return false;
+		break;
+	default:
+		if (type == BPF_READ) {
+			if (size != size_default)
+				return false;
+		} else {
+			return false;
+		}
+	}
+
+	return true;
+}
+
 static bool sock_ops_is_valid_access(int off, int size,
 				     enum bpf_access_type type,
 				     const struct bpf_prog *prog,
@@ -4724,6 +4801,152 @@ static u32 xdp_convert_ctx_access(enum bpf_access_type type,
 	return insn - insn_buf;
 }
 
+/* SOCK_ADDR_LOAD_NESTED_FIELD() loads Nested Field S.F.NF where S is type of
+ * context Structure, F is Field in context structure that contains a pointer
+ * to Nested Structure of type NS that has the field NF.
+ *
+ * SIZE encodes the load size (BPF_B, BPF_H, etc). It's up to caller to make
+ * sure that SIZE is not greater than actual size of S.F.NF.
+ *
+ * If offset OFF is provided, the load happens from that offset relative to
+ * offset of NF.
+ */
+#define SOCK_ADDR_LOAD_NESTED_FIELD_SIZE_OFF(S, NS, F, NF, SIZE, OFF)	       \
+	do {								       \
+		*insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(S, F), si->dst_reg,     \
+				      si->src_reg, offsetof(S, F));	       \
+		*insn++ = BPF_LDX_MEM(					       \
+			SIZE, si->dst_reg, si->dst_reg,			       \
+			bpf_target_off(NS, NF, FIELD_SIZEOF(NS, NF),	       \
+				       target_size)			       \
+				+ OFF);					       \
+	} while (0)
+
+#define SOCK_ADDR_LOAD_NESTED_FIELD(S, NS, F, NF)			       \
+	SOCK_ADDR_LOAD_NESTED_FIELD_SIZE_OFF(S, NS, F, NF,		       \
+					     BPF_FIELD_SIZEOF(NS, NF), 0)
+
+/* SOCK_ADDR_STORE_NESTED_FIELD_OFF() has semantic similar to
+ * SOCK_ADDR_LOAD_NESTED_FIELD_SIZE_OFF() but for store operation.
+ *
+ * It doesn't support SIZE argument though since narrow stores are not
+ * supported for now.
+ *
+ * In addition it uses Temporary Field TF (member of struct S) as the 3rd
+ * "register" since two registers available in convert_ctx_access are not
+ * enough: we can't override neither SRC, since it contains value to store, nor
+ * DST since it contains pointer to context that may be used by later
+ * instructions. But we need a temporary place to save pointer to nested
+ * structure whose field we want to store to.
+ */
+#define SOCK_ADDR_STORE_NESTED_FIELD_OFF(S, NS, F, NF, OFF, TF)		       \
+	do {								       \
+		int tmp_reg = BPF_REG_9;				       \
+		if (si->src_reg == tmp_reg || si->dst_reg == tmp_reg)	       \
+			--tmp_reg;					       \
+		if (si->src_reg == tmp_reg || si->dst_reg == tmp_reg)	       \
+			--tmp_reg;					       \
+		*insn++ = BPF_STX_MEM(BPF_DW, si->dst_reg, tmp_reg,	       \
+				      offsetof(S, TF));			       \
+		*insn++ = BPF_LDX_MEM(BPF_FIELD_SIZEOF(S, F), tmp_reg,	       \
+				      si->dst_reg, offsetof(S, F));	       \
+		*insn++ = BPF_STX_MEM(					       \
+			BPF_FIELD_SIZEOF(NS, NF), tmp_reg, si->src_reg,	       \
+			bpf_target_off(NS, NF, FIELD_SIZEOF(NS, NF),	       \
+				       target_size)			       \
+				+ OFF);					       \
+		*insn++ = BPF_LDX_MEM(BPF_DW, tmp_reg, si->dst_reg,	       \
+				      offsetof(S, TF));			       \
+	} while (0)
+
+#define SOCK_ADDR_LOAD_OR_STORE_NESTED_FIELD_SIZE_OFF(S, NS, F, NF, SIZE, OFF, \
+						      TF)		       \
+	do {								       \
+		if (type == BPF_WRITE) {				       \
+			SOCK_ADDR_STORE_NESTED_FIELD_OFF(S, NS, F, NF, OFF,    \
+							 TF);		       \
+		} else {						       \
+			SOCK_ADDR_LOAD_NESTED_FIELD_SIZE_OFF(		       \
+				S, NS, F, NF, SIZE, OFF);  \
+		}							       \
+	} while (0)
+
+#define SOCK_ADDR_LOAD_OR_STORE_NESTED_FIELD(S, NS, F, NF, TF)		       \
+	SOCK_ADDR_LOAD_OR_STORE_NESTED_FIELD_SIZE_OFF(			       \
+		S, NS, F, NF, BPF_FIELD_SIZEOF(NS, NF), 0, TF)
+
+static u32 sock_addr_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;
+	int off;
+
+	switch (si->off) {
+	case offsetof(struct bpf_sock_addr, user_family):
+		SOCK_ADDR_LOAD_NESTED_FIELD(struct bpf_sock_addr_kern,
+					    struct sockaddr, uaddr, sa_family);
+		break;
+
+	case offsetof(struct bpf_sock_addr, user_ip4):
+		SOCK_ADDR_LOAD_OR_STORE_NESTED_FIELD_SIZE_OFF(
+			struct bpf_sock_addr_kern, struct sockaddr_in, uaddr,
+			sin_addr, BPF_SIZE(si->code), 0, tmp_reg);
+		break;
+
+	case bpf_ctx_range_till(struct bpf_sock_addr, user_ip6[0], user_ip6[3]):
+		off = si->off;
+		off -= offsetof(struct bpf_sock_addr, user_ip6[0]);
+		SOCK_ADDR_LOAD_OR_STORE_NESTED_FIELD_SIZE_OFF(
+			struct bpf_sock_addr_kern, struct sockaddr_in6, uaddr,
+			sin6_addr.s6_addr32[0], BPF_SIZE(si->code), off,
+			tmp_reg);
+		break;
+
+	case offsetof(struct bpf_sock_addr, user_port):
+		/* To get port we need to know sa_family first and then treat
+		 * sockaddr as either sockaddr_in or sockaddr_in6.
+		 * Though we can simplify since port field has same offset and
+		 * size in both structures.
+		 * Here we check this invariant and use just one of the
+		 * structures if it's true.
+		 */
+		BUILD_BUG_ON(offsetof(struct sockaddr_in, sin_port) !=
+			     offsetof(struct sockaddr_in6, sin6_port));
+		BUILD_BUG_ON(FIELD_SIZEOF(struct sockaddr_in, sin_port) !=
+			     FIELD_SIZEOF(struct sockaddr_in6, sin6_port));
+		SOCK_ADDR_LOAD_OR_STORE_NESTED_FIELD(struct bpf_sock_addr_kern,
+						     struct sockaddr_in6, uaddr,
+						     sin6_port, tmp_reg);
+		break;
+
+	case offsetof(struct bpf_sock_addr, family):
+		SOCK_ADDR_LOAD_NESTED_FIELD(struct bpf_sock_addr_kern,
+					    struct sock, sk, sk_family);
+		break;
+
+	case offsetof(struct bpf_sock_addr, type):
+		SOCK_ADDR_LOAD_NESTED_FIELD_SIZE_OFF(
+			struct bpf_sock_addr_kern, struct sock, sk,
+			__sk_flags_offset, BPF_W, 0);
+		*insn++ = BPF_ALU32_IMM(BPF_AND, si->dst_reg, SK_FL_TYPE_MASK);
+		*insn++ = BPF_ALU32_IMM(BPF_RSH, si->dst_reg, SK_FL_TYPE_SHIFT);
+		break;
+
+	case offsetof(struct bpf_sock_addr, protocol):
+		SOCK_ADDR_LOAD_NESTED_FIELD_SIZE_OFF(
+			struct bpf_sock_addr_kern, struct sock, sk,
+			__sk_flags_offset, BPF_W, 0);
+		*insn++ = BPF_ALU32_IMM(BPF_AND, si->dst_reg, SK_FL_PROTO_MASK);
+		*insn++ = BPF_ALU32_IMM(BPF_RSH, si->dst_reg,
+					SK_FL_PROTO_SHIFT);
+		break;
+	}
+
+	return insn - insn_buf;
+}
+
 static u32 sock_ops_convert_ctx_access(enum bpf_access_type type,
 				       const struct bpf_insn *si,
 				       struct bpf_insn *insn_buf,
@@ -5181,6 +5404,15 @@ const struct bpf_verifier_ops cg_sock_verifier_ops = {
 const struct bpf_prog_ops cg_sock_prog_ops = {
 };
 
+const struct bpf_verifier_ops cg_sock_addr_verifier_ops = {
+	.get_func_proto		= sock_addr_func_proto,
+	.is_valid_access	= sock_addr_is_valid_access,
+	.convert_ctx_access	= sock_addr_convert_ctx_access,
+};
+
+const struct bpf_prog_ops cg_sock_addr_prog_ops = {
+};
+
 const struct bpf_verifier_ops sock_ops_verifier_ops = {
 	.get_func_proto		= sock_ops_func_proto,
 	.is_valid_access	= sock_ops_is_valid_access,
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index e8c7fad8c329..2dec266507dc 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -450,6 +450,13 @@ int inet_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
 	if (addr_len < sizeof(struct sockaddr_in))
 		goto out;
 
+	/* BPF prog is run before any checks are done so that if the prog
+	 * changes context in a wrong way it will be caught.
+	 */
+	err = BPF_CGROUP_RUN_PROG_INET4_BIND(sk, uaddr);
+	if (err)
+		goto out;
+
 	if (addr->sin_family != AF_INET) {
 		/* Compatibility games : accept AF_UNSPEC (mapped to AF_INET)
 		 * only if s_addr is INADDR_ANY.
diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
index dbbe04018813..fa24e3f06ac6 100644
--- a/net/ipv6/af_inet6.c
+++ b/net/ipv6/af_inet6.c
@@ -295,6 +295,13 @@ int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
 	if (addr_len < SIN6_LEN_RFC2133)
 		return -EINVAL;
 
+	/* BPF prog is run before any checks are done so that if the prog
+	 * changes context in a wrong way it will be caught.
+	 */
+	err = BPF_CGROUP_RUN_PROG_INET6_BIND(sk, uaddr);
+	if (err)
+		return err;
+
 	if (addr->sin6_family != AF_INET6)
 		return -EAFNOSUPPORT;