samples/bpf: Add test/example of using bpf_probe_write_user bpf helper

This example shows using a kprobe to act as a dnat mechanism to divert
traffic for arbitrary endpoints. It rewrite the arguments to a syscall
while they're still in userspace, and before the syscall has a chance
to copy the argument into kernel space.

Although this is an example, it also acts as a test because the mapped
address is 255.255.255.255:555 -> real address, and that's not a legal
address to connect to. If the helper is broken, the example will fail
on the intermediate steps, as well as the final step to verify the
rewrite of userspace memory succeeded.

Signed-off-by: Sargun Dhillon <sargun@sargun.me>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

3 files changed, 134 insertions, 0 deletions

diff --git a/samples/bpf/Makefile b/samples/bpf/Makefile
index d2d2b35..90ebf7d 100644
--- a/samples/bpf/Makefile
+++ b/samples/bpf/Makefile
@@ -14,6 +14,7 @@ hostprogs-y += tracex3
 hostprogs-y += tracex4
 hostprogs-y += tracex5
 hostprogs-y += tracex6
+hostprogs-y += test_probe_write_user
 hostprogs-y += trace_output
 hostprogs-y += lathist
 hostprogs-y += offwaketime
@@ -37,6 +38,7 @@ tracex3-objs := bpf_load.o libbpf.o tracex3_user.o
 tracex4-objs := bpf_load.o libbpf.o tracex4_user.o
 tracex5-objs := bpf_load.o libbpf.o tracex5_user.o
 tracex6-objs := bpf_load.o libbpf.o tracex6_user.o
+test_probe_write_user-objs := bpf_load.o libbpf.o test_probe_write_user_user.o
 trace_output-objs := bpf_load.o libbpf.o trace_output_user.o
 lathist-objs := bpf_load.o libbpf.o lathist_user.o
 offwaketime-objs := bpf_load.o libbpf.o offwaketime_user.o
@@ -59,6 +61,7 @@ always += tracex3_kern.o
 always += tracex4_kern.o
 always += tracex5_kern.o
 always += tracex6_kern.o
+always += test_probe_write_user_kern.o
 always += trace_output_kern.o
 always += tcbpf1_kern.o
 always += lathist_kern.o
@@ -85,6 +88,7 @@ HOSTLOADLIBES_tracex3 += -lelf
 HOSTLOADLIBES_tracex4 += -lelf -lrt
 HOSTLOADLIBES_tracex5 += -lelf
 HOSTLOADLIBES_tracex6 += -lelf
+HOSTLOADLIBES_test_probe_write_user += -lelf
 HOSTLOADLIBES_trace_output += -lelf -lrt
 HOSTLOADLIBES_lathist += -lelf
 HOSTLOADLIBES_offwaketime += -lelf
diff --git a/samples/bpf/test_probe_write_user_kern.c b/samples/bpf/test_probe_write_user_kern.c
new file mode 100644
index 0000000..3a677c8
--- /dev/null
+++ b/samples/bpf/test_probe_write_user_kern.c
@@ -0,0 +1,52 @@
+/* Copyright (c) 2016 Sargun Dhillon <sargun@sargun.me>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of version 2 of the GNU General Public
+ * License as published by the Free Software Foundation.
+ */
+#include <linux/skbuff.h>
+#include <linux/netdevice.h>
+#include <uapi/linux/bpf.h>
+#include <linux/version.h>
+#include "bpf_helpers.h"
+
+struct bpf_map_def SEC("maps") dnat_map = {
+	.type = BPF_MAP_TYPE_HASH,
+	.key_size = sizeof(struct sockaddr_in),
+	.value_size = sizeof(struct sockaddr_in),
+	.max_entries = 256,
+};
+
+/* kprobe is NOT a stable ABI
+ * kernel functions can be removed, renamed or completely change semantics.
+ * Number of arguments and their positions can change, etc.
+ * In such case this bpf+kprobe example will no longer be meaningful
+ *
+ * This example sits on a syscall, and the syscall ABI is relatively stable
+ * of course, across platforms, and over time, the ABI may change.
+ */
+SEC("kprobe/sys_connect")
+int bpf_prog1(struct pt_regs *ctx)
+{
+	struct sockaddr_in new_addr, orig_addr = {};
+	struct sockaddr_in *mapped_addr;
+	void *sockaddr_arg = (void *)PT_REGS_PARM2(ctx);
+	int sockaddr_len = (int)PT_REGS_PARM3(ctx);
+
+	if (sockaddr_len > sizeof(orig_addr))
+		return 0;
+
+	if (bpf_probe_read(&orig_addr, sizeof(orig_addr), sockaddr_arg) != 0)
+		return 0;
+
+	mapped_addr = bpf_map_lookup_elem(&dnat_map, &orig_addr);
+	if (mapped_addr != NULL) {
+		memcpy(&new_addr, mapped_addr, sizeof(new_addr));
+		bpf_probe_write_user(sockaddr_arg, &new_addr,
+				     sizeof(new_addr));
+	}
+	return 0;
+}
+
+char _license[] SEC("license") = "GPL";
+u32 _version SEC("version") = LINUX_VERSION_CODE;
diff --git a/samples/bpf/test_probe_write_user_user.c b/samples/bpf/test_probe_write_user_user.c
new file mode 100644
index 0000000..a44bf34
--- /dev/null
+++ b/samples/bpf/test_probe_write_user_user.c
@@ -0,0 +1,78 @@
+#include <stdio.h>
+#include <assert.h>
+#include <linux/bpf.h>
+#include <unistd.h>
+#include "libbpf.h"
+#include "bpf_load.h"
+#include <sys/socket.h>
+#include <string.h>
+#include <netinet/in.h>
+#include <arpa/inet.h>
+
+int main(int ac, char **argv)
+{
+	int serverfd, serverconnfd, clientfd;
+	socklen_t sockaddr_len;
+	struct sockaddr serv_addr, mapped_addr, tmp_addr;
+	struct sockaddr_in *serv_addr_in, *mapped_addr_in, *tmp_addr_in;
+	char filename[256];
+	char *ip;
+
+	serv_addr_in = (struct sockaddr_in *)&serv_addr;
+	mapped_addr_in = (struct sockaddr_in *)&mapped_addr;
+	tmp_addr_in = (struct sockaddr_in *)&tmp_addr;
+
+	snprintf(filename, sizeof(filename), "%s_kern.o", argv[0]);
+
+	if (load_bpf_file(filename)) {
+		printf("%s", bpf_log_buf);
+		return 1;
+	}
+
+	assert((serverfd = socket(AF_INET, SOCK_STREAM, 0)) > 0);
+	assert((clientfd = socket(AF_INET, SOCK_STREAM, 0)) > 0);
+
+	/* Bind server to ephemeral port on lo */
+	memset(&serv_addr, 0, sizeof(serv_addr));
+	serv_addr_in->sin_family = AF_INET;
+	serv_addr_in->sin_port = 0;
+	serv_addr_in->sin_addr.s_addr = htonl(INADDR_LOOPBACK);
+
+	assert(bind(serverfd, &serv_addr, sizeof(serv_addr)) == 0);
+
+	sockaddr_len = sizeof(serv_addr);
+	assert(getsockname(serverfd, &serv_addr, &sockaddr_len) == 0);
+	ip = inet_ntoa(serv_addr_in->sin_addr);
+	printf("Server bound to: %s:%d\n", ip, ntohs(serv_addr_in->sin_port));
+
+	memset(&mapped_addr, 0, sizeof(mapped_addr));
+	mapped_addr_in->sin_family = AF_INET;
+	mapped_addr_in->sin_port = htons(5555);
+	mapped_addr_in->sin_addr.s_addr = inet_addr("255.255.255.255");
+
+	assert(!bpf_update_elem(map_fd[0], &mapped_addr, &serv_addr, BPF_ANY));
+
+	assert(listen(serverfd, 5) == 0);
+
+	ip = inet_ntoa(mapped_addr_in->sin_addr);
+	printf("Client connecting to: %s:%d\n",
+	       ip, ntohs(mapped_addr_in->sin_port));
+	assert(connect(clientfd, &mapped_addr, sizeof(mapped_addr)) == 0);
+
+	sockaddr_len = sizeof(tmp_addr);
+	ip = inet_ntoa(tmp_addr_in->sin_addr);
+	assert((serverconnfd = accept(serverfd, &tmp_addr, &sockaddr_len)) > 0);
+	printf("Server received connection from: %s:%d\n",
+	       ip, ntohs(tmp_addr_in->sin_port));
+
+	sockaddr_len = sizeof(tmp_addr);
+	assert(getpeername(clientfd, &tmp_addr, &sockaddr_len) == 0);
+	ip = inet_ntoa(tmp_addr_in->sin_addr);
+	printf("Client's peer address: %s:%d\n",
+	       ip, ntohs(tmp_addr_in->sin_port));
+
+	/* Is the server's getsockname = the socket getpeername */
+	assert(memcmp(&serv_addr, &tmp_addr, sizeof(struct sockaddr_in)) == 0);
+
+	return 0;
+}