1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
|
// Copyright 2019 The gVisor Authors.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
#include <linux/capability.h>
#include <netinet/in.h>
#include <netinet/ip.h>
#include <netinet/ip_icmp.h>
#include <sys/socket.h>
#include <sys/types.h>
#include <unistd.h>
#include <algorithm>
#include <cstdint>
#include "gtest/gtest.h"
#include "test/syscalls/linux/socket_test_util.h"
#include "test/syscalls/linux/unix_domain_socket_test_util.h"
#include "test/util/capability_util.h"
#include "test/util/file_descriptor.h"
#include "test/util/test_util.h"
namespace gvisor {
namespace testing {
namespace {
// The size of an empty ICMP packet and IP header together.
constexpr size_t kEmptyICMPSize = 28;
// ICMP raw sockets get their own special tests because Linux automatically
// responds to ICMP echo requests, and thus a single echo request sent via
// loopback leads to 2 received ICMP packets.
class RawSocketICMPTest : public ::testing::Test {
protected:
// Creates a socket to be used in tests.
void SetUp() override;
// Closes the socket created by SetUp().
void TearDown() override;
// Checks that both an ICMP echo request and reply are received. Calls should
// be wrapped in ASSERT_NO_FATAL_FAILURE.
void ExpectICMPSuccess(const struct icmphdr& icmp);
// Sends icmp via s_.
void SendEmptyICMP(const struct icmphdr& icmp);
// Sends icmp via s_ to the given address.
void SendEmptyICMPTo(int sock, const struct sockaddr_in& addr,
const struct icmphdr& icmp);
// Reads from s_ into recv_buf.
void ReceiveICMP(char* recv_buf, size_t recv_buf_len, size_t expected_size,
struct sockaddr_in* src);
// Reads from sock into recv_buf.
void ReceiveICMPFrom(char* recv_buf, size_t recv_buf_len,
size_t expected_size, struct sockaddr_in* src, int sock);
// The socket used for both reading and writing.
int s_;
// The loopback address.
struct sockaddr_in addr_;
};
void RawSocketICMPTest::SetUp() {
if (!ASSERT_NO_ERRNO_AND_VALUE(HaveCapability(CAP_NET_RAW))) {
ASSERT_THAT(socket(AF_INET, SOCK_RAW, IPPROTO_ICMP),
SyscallFailsWithErrno(EPERM));
GTEST_SKIP();
}
ASSERT_THAT(s_ = socket(AF_INET, SOCK_RAW, IPPROTO_ICMP), SyscallSucceeds());
addr_ = {};
// "On raw sockets sin_port is set to the IP protocol." - ip(7).
addr_.sin_port = IPPROTO_IP;
addr_.sin_addr.s_addr = htonl(INADDR_LOOPBACK);
addr_.sin_family = AF_INET;
}
void RawSocketICMPTest::TearDown() {
// TearDown will be run even if we skip the test.
if (ASSERT_NO_ERRNO_AND_VALUE(HaveCapability(CAP_NET_RAW))) {
EXPECT_THAT(close(s_), SyscallSucceeds());
}
}
// We'll only read an echo in this case, as the kernel won't respond to the
// malformed ICMP checksum.
TEST_F(RawSocketICMPTest, SendAndReceiveBadChecksum) {
SKIP_IF(!ASSERT_NO_ERRNO_AND_VALUE(HaveCapability(CAP_NET_RAW)));
// Prepare and send an ICMP packet. Use arbitrary junk for checksum, sequence,
// and ID. None of that should matter for raw sockets - the kernel should
// still give us the packet.
struct icmphdr icmp;
icmp.type = ICMP_ECHO;
icmp.code = 0;
icmp.checksum = 0;
icmp.un.echo.sequence = 2012;
icmp.un.echo.id = 2014;
ASSERT_NO_FATAL_FAILURE(SendEmptyICMP(icmp));
// Veryify that we get the echo, then that there's nothing else to read.
char recv_buf[kEmptyICMPSize];
struct sockaddr_in src;
ASSERT_NO_FATAL_FAILURE(
ReceiveICMP(recv_buf, sizeof(recv_buf), sizeof(struct icmphdr), &src));
EXPECT_EQ(memcmp(&src, &addr_, sizeof(src)), 0);
// The packet should be identical to what we sent.
EXPECT_EQ(memcmp(recv_buf + sizeof(struct iphdr), &icmp, sizeof(icmp)), 0);
// And there should be nothing left to read.
EXPECT_THAT(RetryEINTR(recv)(s_, recv_buf, sizeof(recv_buf), MSG_DONTWAIT),
SyscallFailsWithErrno(EAGAIN));
}
// Send and receive an ICMP packet.
TEST_F(RawSocketICMPTest, SendAndReceive) {
SKIP_IF(!ASSERT_NO_ERRNO_AND_VALUE(HaveCapability(CAP_NET_RAW)));
// Prepare and send an ICMP packet. Use arbitrary junk for sequence and ID.
// None of that should matter for raw sockets - the kernel should still give
// us the packet.
struct icmphdr icmp;
icmp.type = ICMP_ECHO;
icmp.code = 0;
icmp.checksum = 0;
icmp.un.echo.sequence = 2012;
icmp.un.echo.id = 2014;
icmp.checksum = ICMPChecksum(icmp, NULL, 0);
ASSERT_NO_FATAL_FAILURE(SendEmptyICMP(icmp));
ASSERT_NO_FATAL_FAILURE(ExpectICMPSuccess(icmp));
}
// We should be able to create multiple raw sockets for the same protocol and
// receive the same packet on both.
TEST_F(RawSocketICMPTest, MultipleSocketReceive) {
SKIP_IF(!ASSERT_NO_ERRNO_AND_VALUE(HaveCapability(CAP_NET_RAW)));
FileDescriptor s2 =
ASSERT_NO_ERRNO_AND_VALUE(Socket(AF_INET, SOCK_RAW, IPPROTO_ICMP));
// Prepare and send an ICMP packet. Use arbitrary junk for sequence and ID.
// None of that should matter for raw sockets - the kernel should still give
// us the packet.
struct icmphdr icmp;
icmp.type = ICMP_ECHO;
icmp.code = 0;
icmp.checksum = 0;
icmp.un.echo.sequence = 2016;
icmp.un.echo.id = 2018;
icmp.checksum = ICMPChecksum(icmp, NULL, 0);
ASSERT_NO_FATAL_FAILURE(SendEmptyICMP(icmp));
// Both sockets will receive the echo request and reply in indeterminate
// order, so we'll need to read 2 packets from each.
// Receive on socket 1.
constexpr int kBufSize = kEmptyICMPSize;
char recv_buf1[2][kBufSize];
struct sockaddr_in src;
for (int i = 0; i < 2; i++) {
ASSERT_NO_FATAL_FAILURE(ReceiveICMP(recv_buf1[i],
ABSL_ARRAYSIZE(recv_buf1[i]),
sizeof(struct icmphdr), &src));
EXPECT_EQ(memcmp(&src, &addr_, sizeof(src)), 0);
}
// Receive on socket 2.
char recv_buf2[2][kBufSize];
for (int i = 0; i < 2; i++) {
ASSERT_NO_FATAL_FAILURE(
ReceiveICMPFrom(recv_buf2[i], ABSL_ARRAYSIZE(recv_buf2[i]),
sizeof(struct icmphdr), &src, s2.get()));
EXPECT_EQ(memcmp(&src, &addr_, sizeof(src)), 0);
}
// Ensure both sockets receive identical packets.
int types[] = {ICMP_ECHO, ICMP_ECHOREPLY};
for (int type : types) {
auto match_type = [=](char buf[kBufSize]) {
struct icmphdr* icmp =
reinterpret_cast<struct icmphdr*>(buf + sizeof(struct iphdr));
return icmp->type == type;
};
auto icmp1_it =
std::find_if(std::begin(recv_buf1), std::end(recv_buf1), match_type);
auto icmp2_it =
std::find_if(std::begin(recv_buf2), std::end(recv_buf2), match_type);
ASSERT_NE(icmp1_it, std::end(recv_buf1));
ASSERT_NE(icmp2_it, std::end(recv_buf2));
EXPECT_EQ(memcmp(*icmp1_it + sizeof(struct iphdr),
*icmp2_it + sizeof(struct iphdr), sizeof(icmp)),
0);
}
}
// A raw ICMP socket and ping socket should both receive the ICMP packets
// intended for the ping socket.
TEST_F(RawSocketICMPTest, RawAndPingSockets) {
SKIP_IF(!ASSERT_NO_ERRNO_AND_VALUE(HaveCapability(CAP_NET_RAW)));
FileDescriptor ping_sock =
ASSERT_NO_ERRNO_AND_VALUE(Socket(AF_INET, SOCK_DGRAM, IPPROTO_ICMP));
// Ping sockets take care of the ICMP ID and checksum.
struct icmphdr icmp;
icmp.type = ICMP_ECHO;
icmp.code = 0;
icmp.un.echo.sequence = *static_cast<unsigned short*>(&icmp.un.echo.sequence);
ASSERT_THAT(RetryEINTR(sendto)(ping_sock.get(), &icmp, sizeof(icmp), 0,
reinterpret_cast<struct sockaddr*>(&addr_),
sizeof(addr_)),
SyscallSucceedsWithValue(sizeof(icmp)));
// Receive on socket 1, which receives the echo request and reply in
// indeterminate order.
constexpr int kBufSize = kEmptyICMPSize;
char recv_buf1[2][kBufSize];
struct sockaddr_in src;
for (int i = 0; i < 2; i++) {
ASSERT_NO_FATAL_FAILURE(
ReceiveICMP(recv_buf1[i], kBufSize, sizeof(struct icmphdr), &src));
EXPECT_EQ(memcmp(&src, &addr_, sizeof(src)), 0);
}
// Receive on socket 2. Ping sockets only get the echo reply, not the initial
// echo.
char ping_recv_buf[kBufSize];
ASSERT_THAT(RetryEINTR(recv)(ping_sock.get(), ping_recv_buf, kBufSize, 0),
SyscallSucceedsWithValue(sizeof(struct icmphdr)));
// Ensure both sockets receive identical echo reply packets.
auto match_type_raw = [=](char buf[kBufSize]) {
struct icmphdr* icmp =
reinterpret_cast<struct icmphdr*>(buf + sizeof(struct iphdr));
return icmp->type == ICMP_ECHOREPLY;
};
auto raw_reply_it =
std::find_if(std::begin(recv_buf1), std::end(recv_buf1), match_type_raw);
ASSERT_NE(raw_reply_it, std::end(recv_buf1));
EXPECT_EQ(
memcmp(*raw_reply_it + sizeof(struct iphdr), ping_recv_buf, sizeof(icmp)),
0);
}
// A raw ICMP socket should be able to send a malformed short ICMP Echo Request,
// while ping socket should not.
// Neither should be able to receieve a short malformed packet.
TEST_F(RawSocketICMPTest, ShortEchoRawAndPingSockets) {
SKIP_IF(!ASSERT_NO_ERRNO_AND_VALUE(HaveCapability(CAP_NET_RAW)));
FileDescriptor ping_sock =
ASSERT_NO_ERRNO_AND_VALUE(Socket(AF_INET, SOCK_DGRAM, IPPROTO_ICMP));
struct icmphdr icmp;
icmp.type = ICMP_ECHO;
icmp.code = 0;
icmp.un.echo.sequence = 0;
icmp.un.echo.id = 6789;
icmp.checksum = 0;
icmp.checksum = ICMPChecksum(icmp, NULL, 0);
// Omit 2 bytes from ICMP packet.
constexpr int kShortICMPSize = sizeof(icmp) - 2;
// Sending a malformed short ICMP message to a ping socket should fail.
ASSERT_THAT(RetryEINTR(sendto)(ping_sock.get(), &icmp, kShortICMPSize, 0,
reinterpret_cast<struct sockaddr*>(&addr_),
sizeof(addr_)),
SyscallFailsWithErrno(EINVAL));
// Sending a malformed short ICMP message to a raw socket should not fail.
ASSERT_THAT(RetryEINTR(sendto)(s_, &icmp, kShortICMPSize, 0,
reinterpret_cast<struct sockaddr*>(&addr_),
sizeof(addr_)),
SyscallSucceedsWithValue(kShortICMPSize));
// Neither Ping nor Raw socket should have anything to read.
char recv_buf[kEmptyICMPSize];
EXPECT_THAT(RetryEINTR(recv)(ping_sock.get(), recv_buf, sizeof(recv_buf),
MSG_DONTWAIT),
SyscallFailsWithErrno(EAGAIN));
EXPECT_THAT(RetryEINTR(recv)(s_, recv_buf, sizeof(recv_buf), MSG_DONTWAIT),
SyscallFailsWithErrno(EAGAIN));
}
// A raw ICMP socket should be able to send a malformed short ICMP Echo Reply,
// while ping socket should not.
// Neither should be able to receieve a short malformed packet.
TEST_F(RawSocketICMPTest, ShortEchoReplyRawAndPingSockets) {
SKIP_IF(!ASSERT_NO_ERRNO_AND_VALUE(HaveCapability(CAP_NET_RAW)));
FileDescriptor ping_sock =
ASSERT_NO_ERRNO_AND_VALUE(Socket(AF_INET, SOCK_DGRAM, IPPROTO_ICMP));
struct icmphdr icmp;
icmp.type = ICMP_ECHOREPLY;
icmp.code = 0;
icmp.un.echo.sequence = 0;
icmp.un.echo.id = 6789;
icmp.checksum = 0;
icmp.checksum = ICMPChecksum(icmp, NULL, 0);
// Omit 2 bytes from ICMP packet.
constexpr int kShortICMPSize = sizeof(icmp) - 2;
// Sending a malformed short ICMP message to a ping socket should fail.
ASSERT_THAT(RetryEINTR(sendto)(ping_sock.get(), &icmp, kShortICMPSize, 0,
reinterpret_cast<struct sockaddr*>(&addr_),
sizeof(addr_)),
SyscallFailsWithErrno(EINVAL));
// Sending a malformed short ICMP message to a raw socket should not fail.
ASSERT_THAT(RetryEINTR(sendto)(s_, &icmp, kShortICMPSize, 0,
reinterpret_cast<struct sockaddr*>(&addr_),
sizeof(addr_)),
SyscallSucceedsWithValue(kShortICMPSize));
// Neither Ping nor Raw socket should have anything to read.
char recv_buf[kEmptyICMPSize];
EXPECT_THAT(RetryEINTR(recv)(ping_sock.get(), recv_buf, sizeof(recv_buf),
MSG_DONTWAIT),
SyscallFailsWithErrno(EAGAIN));
EXPECT_THAT(RetryEINTR(recv)(s_, recv_buf, sizeof(recv_buf), MSG_DONTWAIT),
SyscallFailsWithErrno(EAGAIN));
}
// Test that connect() sends packets to the right place.
TEST_F(RawSocketICMPTest, SendAndReceiveViaConnect) {
SKIP_IF(!ASSERT_NO_ERRNO_AND_VALUE(HaveCapability(CAP_NET_RAW)));
ASSERT_THAT(
connect(s_, reinterpret_cast<struct sockaddr*>(&addr_), sizeof(addr_)),
SyscallSucceeds());
// Prepare and send an ICMP packet. Use arbitrary junk for sequence and ID.
// None of that should matter for raw sockets - the kernel should still give
// us the packet.
struct icmphdr icmp;
icmp.type = ICMP_ECHO;
icmp.code = 0;
icmp.checksum = 0;
icmp.un.echo.sequence = 2003;
icmp.un.echo.id = 2004;
icmp.checksum = ICMPChecksum(icmp, NULL, 0);
ASSERT_THAT(send(s_, &icmp, sizeof(icmp), 0),
SyscallSucceedsWithValue(sizeof(icmp)));
ASSERT_NO_FATAL_FAILURE(ExpectICMPSuccess(icmp));
}
// Bind to localhost, then send and receive packets.
TEST_F(RawSocketICMPTest, BindSendAndReceive) {
SKIP_IF(!ASSERT_NO_ERRNO_AND_VALUE(HaveCapability(CAP_NET_RAW)));
ASSERT_THAT(
bind(s_, reinterpret_cast<struct sockaddr*>(&addr_), sizeof(addr_)),
SyscallSucceeds());
// Prepare and send an ICMP packet. Use arbitrary junk for checksum, sequence,
// and ID. None of that should matter for raw sockets - the kernel should
// still give us the packet.
struct icmphdr icmp;
icmp.type = ICMP_ECHO;
icmp.code = 0;
icmp.checksum = 0;
icmp.un.echo.sequence = 2004;
icmp.un.echo.id = 2007;
icmp.checksum = ICMPChecksum(icmp, NULL, 0);
ASSERT_NO_FATAL_FAILURE(SendEmptyICMP(icmp));
ASSERT_NO_FATAL_FAILURE(ExpectICMPSuccess(icmp));
}
// Bind and connect to localhost and send/receive packets.
TEST_F(RawSocketICMPTest, BindConnectSendAndReceive) {
SKIP_IF(!ASSERT_NO_ERRNO_AND_VALUE(HaveCapability(CAP_NET_RAW)));
ASSERT_THAT(
bind(s_, reinterpret_cast<struct sockaddr*>(&addr_), sizeof(addr_)),
SyscallSucceeds());
ASSERT_THAT(
connect(s_, reinterpret_cast<struct sockaddr*>(&addr_), sizeof(addr_)),
SyscallSucceeds());
// Prepare and send an ICMP packet. Use arbitrary junk for checksum, sequence,
// and ID. None of that should matter for raw sockets - the kernel should
// still give us the packet.
struct icmphdr icmp;
icmp.type = ICMP_ECHO;
icmp.code = 0;
icmp.checksum = 0;
icmp.un.echo.sequence = 2010;
icmp.un.echo.id = 7;
icmp.checksum = ICMPChecksum(icmp, NULL, 0);
ASSERT_NO_FATAL_FAILURE(SendEmptyICMP(icmp));
ASSERT_NO_FATAL_FAILURE(ExpectICMPSuccess(icmp));
}
void RawSocketICMPTest::ExpectICMPSuccess(const struct icmphdr& icmp) {
// We're going to receive both the echo request and reply, but the order is
// indeterminate.
char recv_buf[kEmptyICMPSize];
struct sockaddr_in src;
bool received_request = false;
bool received_reply = false;
for (int i = 0; i < 2; i++) {
// Receive the packet.
ASSERT_NO_FATAL_FAILURE(ReceiveICMP(recv_buf, ABSL_ARRAYSIZE(recv_buf),
sizeof(struct icmphdr), &src));
EXPECT_EQ(memcmp(&src, &addr_, sizeof(src)), 0);
struct icmphdr* recvd_icmp =
reinterpret_cast<struct icmphdr*>(recv_buf + sizeof(struct iphdr));
switch (recvd_icmp->type) {
case ICMP_ECHO:
EXPECT_FALSE(received_request);
received_request = true;
// The packet should be identical to what we sent.
EXPECT_EQ(memcmp(recv_buf + sizeof(struct iphdr), &icmp, sizeof(icmp)),
0);
break;
case ICMP_ECHOREPLY:
EXPECT_FALSE(received_reply);
received_reply = true;
// Most fields should be the same.
EXPECT_EQ(recvd_icmp->code, icmp.code);
EXPECT_EQ(recvd_icmp->un.echo.sequence, icmp.un.echo.sequence);
EXPECT_EQ(recvd_icmp->un.echo.id, icmp.un.echo.id);
// A couple are different.
EXPECT_EQ(recvd_icmp->type, ICMP_ECHOREPLY);
// The checksum computed over the reply should still be valid.
EXPECT_EQ(ICMPChecksum(*recvd_icmp, NULL, 0), 0);
break;
}
}
ASSERT_TRUE(received_request);
ASSERT_TRUE(received_reply);
}
void RawSocketICMPTest::SendEmptyICMP(const struct icmphdr& icmp) {
ASSERT_NO_FATAL_FAILURE(SendEmptyICMPTo(s_, addr_, icmp));
}
void RawSocketICMPTest::SendEmptyICMPTo(int sock,
const struct sockaddr_in& addr,
const struct icmphdr& icmp) {
// It's safe to use const_cast here because sendmsg won't modify the iovec or
// address.
struct iovec iov = {};
iov.iov_base = static_cast<void*>(const_cast<struct icmphdr*>(&icmp));
iov.iov_len = sizeof(icmp);
struct msghdr msg = {};
msg.msg_name = static_cast<void*>(const_cast<struct sockaddr_in*>(&addr));
msg.msg_namelen = sizeof(addr);
msg.msg_iov = &iov;
msg.msg_iovlen = 1;
msg.msg_control = NULL;
msg.msg_controllen = 0;
msg.msg_flags = 0;
ASSERT_THAT(sendmsg(sock, &msg, 0), SyscallSucceedsWithValue(sizeof(icmp)));
}
void RawSocketICMPTest::ReceiveICMP(char* recv_buf, size_t recv_buf_len,
size_t expected_size,
struct sockaddr_in* src) {
ASSERT_NO_FATAL_FAILURE(
ReceiveICMPFrom(recv_buf, recv_buf_len, expected_size, src, s_));
}
void RawSocketICMPTest::ReceiveICMPFrom(char* recv_buf, size_t recv_buf_len,
size_t expected_size,
struct sockaddr_in* src, int sock) {
struct iovec iov = {};
iov.iov_base = recv_buf;
iov.iov_len = recv_buf_len;
struct msghdr msg = {};
msg.msg_name = src;
msg.msg_namelen = sizeof(*src);
msg.msg_iov = &iov;
msg.msg_iovlen = 1;
msg.msg_control = NULL;
msg.msg_controllen = 0;
msg.msg_flags = 0;
// We should receive the ICMP packet plus 20 bytes of IP header.
ASSERT_THAT(recvmsg(sock, &msg, 0),
SyscallSucceedsWithValue(expected_size + sizeof(struct iphdr)));
}
} // namespace
} // namespace testing
} // namespace gvisor
|