aboutgitcodebugslistschat
path: root/test/nstool.c
blob: fc357d8aea9122ca26892c681d77b7b612b1f31c (plain) (blame)
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
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
// SPDX-License-Identifier: GPL-2.0-or-later

/* nstool - maintain a namespace to be entered by other processes
 *
 * Copyright Red Hat
 * Author: David Gibson <david@gibson.dropbear.id.au>
 */

#define _GNU_SOURCE

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <stdbool.h>
#include <stdint.h>
#include <errno.h>
#include <unistd.h>
#include <getopt.h>
#include <stdarg.h>
#include <limits.h>
#include <fcntl.h>
#include <limits.h>
#include <unistd.h>
#include <sys/socket.h>
#include <sys/wait.h>
#include <sys/syscall.h>
#include <sys/prctl.h>
#include <linux/un.h>
#include <sched.h>
#include <linux/capability.h>

#define	ARRAY_SIZE(a)	((int)(sizeof(a) / sizeof((a)[0])))

#define die(...)				\
	do {					\
		fprintf(stderr, __VA_ARGS__);	\
		exit(1);			\
	} while (0)

struct ns_type {
	int flag;
	const char *name;
};

const struct ns_type nstypes[] = {
	{ CLONE_NEWCGROUP, "cgroup" },
	{ CLONE_NEWIPC, "ipc" },
	{ CLONE_NEWNET, "net" },
	{ CLONE_NEWNS, "mnt" },
	{ CLONE_NEWPID, "pid" },
	{ CLONE_NEWTIME, "time" },
	{ CLONE_NEWUSER, "user" },
	{ CLONE_NEWUTS, "uts" },
};

#define for_each_nst(_nst, _flags)				\
	for ((_nst) = &nstypes[0];				\
	     ((_nst) - nstypes) < ARRAY_SIZE(nstypes);		\
	     (_nst)++)						\
		if ((_flags) & (_nst)->flag)

#define for_every_nst(_nst)	for_each_nst(_nst, INT_MAX)

#define NSTOOL_MAGIC	0x7570017575601d75ULL

struct holder_info {
	uint64_t magic;
	pid_t pid;
	uid_t uid;
	gid_t gid;
	char cwd[PATH_MAX];
};

static void usage(void)
{
	die("Usage:\n"
	    "  nstool hold SOCK\n"
	    "    Run within a set of namespaces, open a Unix domain socket\n"
	    "    (the \"control socket\") at SOCK and wait for requests from\n"
	    "    other nstool subcommands.\n"
	    "  nstool info [-pw] pid SOCK\n"
	    "    Print information about the nstool hold process with control\n"
	    "    socket at SOCK\n"
	    "      -p          Print just the holder's PID as seen by the caller\n"
	    "      -w          Retry connecting to SOCK until it is ready\n"
	    "  nstool exec [--keep-caps] SOCK [COMMAND [ARGS...]]\n"
	    "    Execute command or shell in the namespaces of the nstool hold\n"
	    "    with control socket at SOCK\n"
	    "      --keep-caps Give all possible capabilities to COMMAND via\n"
	    "                  the ambient capability mask\n"
	    "  nstool stop SOCK\n"
	    "    Instruct the nstool hold with control socket at SOCK to\n"
	    "    terminate.\n");
}

static void sockaddr_from_path(struct sockaddr_un *addr, const char *sockpath)
{
	if (strlen(sockpath) > UNIX_PATH_MAX)
		die("\"%s\" is too long for Unix socket path (%zu > %d)",
		    sockpath, strlen(sockpath), UNIX_PATH_MAX);

	addr->sun_family = AF_UNIX;
	strncpy(addr->sun_path, sockpath, UNIX_PATH_MAX);
}

static int connect_ctl(const char *sockpath, bool wait,
		       struct holder_info *info,
		       struct ucred *peercred)
{
	int fd = socket(AF_UNIX, SOCK_STREAM | SOCK_CLOEXEC, PF_UNIX);
	struct sockaddr_un addr;
	struct holder_info discard;
	ssize_t len;
	int rc;

	if (fd < 0)
		die("socket(): %s\n", strerror(errno));

	sockaddr_from_path(&addr, sockpath);

	do {
		rc = connect(fd, (struct sockaddr *)&addr, sizeof(addr));
		if (rc < 0 &&
		    (!wait || (errno != ENOENT && errno != ECONNREFUSED)))
			die("connect() to %s: %s\n", sockpath, strerror(errno));
	} while (rc < 0);

	if (!info)
		info = &discard;

	/* Always read the info structure, even if we don't need it,
	 * so that the holder doesn't get a broken pipe error
	 */
	len = read(fd, info, sizeof(*info));
	if (len < 0)
		die("read() on control socket %s: %s\n", sockpath, strerror(errno));
	if ((size_t)len < sizeof(*info))
		die("short read() on control socket %s\n", sockpath);

	if (info->magic != NSTOOL_MAGIC)
		die("Control socket %s doesn't appear to belong to nstool\n",
		    sockpath);

	if (peercred) {
		socklen_t optlen = sizeof(*peercred);

		rc = getsockopt(fd, SOL_SOCKET, SO_PEERCRED,
				peercred, &optlen);
		if (rc < 0)
			die("getsockopet(SO_PEERCRED) %s: %s\n",
			    sockpath, strerror(errno));
	}

	return fd;
}

static void cmd_hold(int argc, char *argv[])
{
	int fd = socket(AF_UNIX, SOCK_STREAM | SOCK_CLOEXEC, PF_UNIX);
	struct sockaddr_un addr;
	const char *sockpath = argv[1];
	struct holder_info info;
	int rc;

	if (argc != 2)
		usage();

	if (fd < 0)
		die("socket(): %s\n", strerror(errno));

	sockaddr_from_path(&addr, sockpath);

	rc = bind(fd, (struct sockaddr *)&addr, sizeof(addr));
	if (rc < 0)
		die("bind() to %s: %s\n", sockpath, strerror(errno));

	rc = listen(fd, 0);
	if (rc < 0)
		die("listen() on %s: %s\n", sockpath, strerror(errno));

	info.magic = NSTOOL_MAGIC;
	info.pid = getpid();
	info.uid = getuid();
	info.gid = getgid();
	if (!getcwd(info.cwd, sizeof(info.cwd)))
		die("getcwd(): %s\n", strerror(errno));

	do {
		int afd = accept(fd, NULL, NULL);
		char buf;

		if (afd < 0)
			die("accept(): %s\n", strerror(errno));

		rc = write(afd, &info, sizeof(info));
		if (rc < 0)
			die("write(): %s\n", strerror(errno));
		if ((size_t)rc < sizeof(info))
			die("short write() on control socket\n");

		rc = read(afd, &buf, sizeof(buf));
		if (rc < 0)
			die("read(): %s\n", strerror(errno));

		close(afd);
	} while (rc == 0);

	unlink(sockpath);
}

static ssize_t getlink(char *buf, size_t bufsiz, const char *fmt, ...)
{
	char linkpath[PATH_MAX];
	ssize_t linklen;
	va_list ap;

	va_start(ap, fmt);
	if (vsnprintf(linkpath, sizeof(linkpath), fmt, ap) >= PATH_MAX)
		die("Truncated path \"%s\"\n", linkpath);
	va_end(ap);

	linklen = readlink(linkpath, buf, bufsiz);
	if (linklen < 0)
		die("readlink() on %s: %s\n", linkpath, strerror(errno));
	if ((size_t)linklen >= bufsiz)
		die("Target of symbolic link %s is too long\n", linkpath);

	return linklen;
}

static int detect_namespaces(pid_t pid)
{
	const struct ns_type *nst;
	int flags = 0;

	for_every_nst(nst) {
		char selflink[PATH_MAX], pidlink[PATH_MAX];
		ssize_t selflen, pidlen;

		selflen = getlink(selflink, sizeof(selflink),
				  "/proc/self/ns/%s", nst->name);
		pidlen = getlink(pidlink, sizeof(pidlink),
				 "/proc/%d/ns/%s", pid, nst->name);

		if ((selflen != pidlen) || memcmp(selflink, pidlink, selflen))
			flags |= nst->flag;
	}

	return flags;
}

static void print_nstypes(int flags)
{
	const struct ns_type *nst;
	bool first = true;

	for_each_nst(nst, flags) {
		printf("%s%s", first ? "" : ", " , nst->name);
		first = false;
		flags &= ~nst->flag;
	}

	if (flags)
		printf("%s0x%x", first ? "" : ", ", flags);
}

static void cmd_info(int argc, char *argv[])
{
	const struct option options[] = {
		{"pid",		no_argument, 	NULL,	'p' },
		{"wait",	no_argument,	NULL,	'w' },
		{ 0 },
	};
	bool pidonly = false, waitforsock = false;
	const char *optstring = "pw";
	struct holder_info info;
	struct ucred peercred;
	const char *sockpath;
	int fd, opt;

	do {
		opt = getopt_long(argc, argv, optstring, options, NULL);

		switch (opt) {
		case 'p':
			pidonly = true;
			break;
		case 'w':
			waitforsock = true;
			break;
		case -1:
			break;
		default:
			usage();
		}
	} while (opt != -1);

	if (optind != argc - 1) {
		usage();
	}

	sockpath = argv[optind];

	fd = connect_ctl(sockpath, waitforsock, &info, &peercred);

	close(fd);

	if (pidonly) {
		printf("%d\n", peercred.pid);
	} else {
		int flags = detect_namespaces(peercred.pid);

		printf("Namespaces: ");
		print_nstypes(flags);
		printf("\n");

		printf("As seen from calling context:\n");
		printf("\tPID:\t%d\n", peercred.pid);
		printf("\tUID:\t%u\n", peercred.uid);
		printf("\tGID:\t%u\n", peercred.gid);

		printf("As seen from holding context:\n");
		printf("\tPID:\t%d\n", info.pid);
		printf("\tUID:\t%u\n", info.uid);
		printf("\tGID:\t%u\n", info.gid);
		printf("\tCWD:\t%s\n", info.cwd);
	}
}

static int openns(const char *fmt, ...)
{
	char nspath[PATH_MAX];
	va_list ap;
	int fd;

	va_start(ap, fmt);
	if (vsnprintf(nspath, sizeof(nspath), fmt, ap) >= PATH_MAX)
		die("Truncated path \"%s\"\n", nspath);
	va_end(ap);

	fd = open(nspath, O_RDONLY | O_CLOEXEC);
	if (fd < 0)
		die("open() %s: %s\n", nspath, strerror(errno));

	return fd;
}

static pid_t sig_pid;
static void sig_handler(int signum)
{
	int err;

	err = kill(sig_pid, signum);
	if (err)
		die("Propagating %s: %s\n", strsignal(signum), strerror(errno));
}

static void wait_for_child(pid_t pid)
{
	struct sigaction sa = {
		.sa_handler = sig_handler,
		.sa_flags = SA_RESETHAND,
	};
	int status, err;

	sig_pid = pid;
	err = sigaction(SIGTERM, &sa, NULL);
	if (err)
		die("sigaction(SIGTERM): %s\n", strerror(errno));

	/* Match the child's exit status, if possible */
	for (;;) {
		pid_t rc;

		rc = waitpid(pid, &status, WUNTRACED);
		if (rc < 0) {
			if (errno == EINTR)
				continue;
			die("waitpid() on %d: %s\n", pid, strerror(errno));
		}
		if (rc != pid)
			die("waitpid() on %d returned %d", pid, rc);
		if (WIFSTOPPED(status)) {
			/* Stop the parent to match */
			kill(getpid(), SIGSTOP);
			/* We must have resumed, resume the child */
			kill(pid, SIGCONT);
			continue;
		}

		break;
	}

	if (WIFEXITED(status))
		exit(WEXITSTATUS(status));
	else if (WIFSIGNALED(status))
		kill(getpid(), WTERMSIG(status));

	die("Unexpected status for child %d\n", pid);
}

static void caps_to_ambient(void)
{
	/* Use raw system calls to avoid the overly complex caps
	 * libraries. */
	struct __user_cap_header_struct header = {
		.version = _LINUX_CAPABILITY_VERSION_3,
		.pid = 0,
	};
	struct __user_cap_data_struct payload[_LINUX_CAPABILITY_U32S_3] =
		{{ 0 }};
	uint64_t effective, cap;

	if (syscall(SYS_capget, &header, payload) < 0)
		die("capget(): %s\n", strerror(errno));

	/* First make caps inheritable */
	payload[0].inheritable = payload[0].permitted;
	payload[1].inheritable = payload[1].permitted;

	if (syscall(SYS_capset, &header, payload) < 0)
		die("capset(): %s\n", strerror(errno));

	effective = ((uint64_t)payload[1].effective << 32) | (uint64_t)payload[0].effective;

	for (cap = 0; cap < (sizeof(effective) * 8); cap++) {
		/* Skip non-existent caps */
		if (prctl(PR_CAPBSET_READ, cap, 0, 0, 0) < 0)
			continue;

		if ((effective & (1 << cap))
		    && prctl(PR_CAP_AMBIENT, PR_CAP_AMBIENT_RAISE, cap, 0, 0) < 0)
			die("prctl(PR_CAP_AMBIENT): %s\n", strerror(errno));
	}
}

static void cmd_exec(int argc, char *argv[])
{
	enum {
		OPT_EXEC_KEEPCAPS = CHAR_MAX + 1,
	};
	const struct option options[] = {
		{"keep-caps",	no_argument, 	NULL,	OPT_EXEC_KEEPCAPS },
		{ 0 },
	};
	const char *shargs[] = { NULL, NULL };
	const char *sockpath = argv[1];
	int nfd[ARRAY_SIZE(nstypes)];
	const char *optstring = "";
	const struct ns_type *nst;
	int ctlfd, flags, opt, rc;
	const char *const *xargs;
	struct holder_info info;
	bool keepcaps = false;
	struct ucred peercred;
	const char *exe;
	pid_t xpid;

	do {
		opt = getopt_long(argc, argv, optstring, options, NULL);

		switch (opt) {
		case OPT_EXEC_KEEPCAPS:
			keepcaps = true;
			break;
		case -1:
			break;
		default:
			usage();
		}
	} while (opt != -1);

	if (argc < optind + 1)
		usage();

	sockpath = argv[optind];

	ctlfd = connect_ctl(sockpath, false, &info, &peercred);

	flags = detect_namespaces(peercred.pid);

	for_each_nst(nst, flags) {
		int *fd = &nfd[nst - nstypes];
		*fd = openns("/proc/%d/ns/%s", peercred.pid, nst->name);
	}

	/* First pass, will get things where we need the privileges of
	 * the initial userns */
	for_each_nst(nst, flags) {
		int fd = nfd[nst - nstypes];

		rc = setns(fd, nst->flag);
		if (rc == 0) {
			flags &= ~nst->flag;
		}
	}

	/* Second pass, will get things where we need the privileges
	 * of the target userns */
	for_each_nst(nst, flags) {
		int fd = nfd[nst - nstypes];

		rc = setns(fd, nst->flag);
		if (rc < 0)
			die("setns() type %s: %s\n",
			    nst->name, strerror(errno));
	}

	/* If we've entered a mount ns, our cwd has changed to /.
	 * Switch to the cwd of the holder, which is probably less
	 * surprising. */
	if (flags & CLONE_NEWNS) {
		rc = chdir(info.cwd);
		if (rc < 0)
			die("chdir(\"%s\"): %s\n", info.cwd, strerror(errno));
	}

	/* Fork to properly enter PID namespace */
	xpid = fork();
	if (xpid < 0)
		die("fork(): %s\n", strerror(errno));

	if (xpid > 0) {
		/* Close the control socket so the waiting parent
		 * doesn't block the holder */
		close(ctlfd);
		wait_for_child(xpid);
	}

	/* CHILD */
	if (argc > optind + 1) {
		exe = argv[optind + 1];
		xargs = (const char *const *)(argv + optind + 1);
	} else {
		exe = getenv("SHELL");
		if (!exe)
			exe = "/bin/sh";

		shargs[0] = exe;

		xargs = shargs;
	}

	if (keepcaps)
		caps_to_ambient();

	rc = execvp(exe, (char *const *)xargs);
	if (rc < 0)
		die("execv() %s: %s\n", exe, strerror(errno));
	die("Returned from exec()\n");
}

static void cmd_stop(int argc, char *argv[])
{
	const char *sockpath = argv[1];
	int fd, rc;
	char buf = 'Q';

	if (argc != 2)
		usage();

	fd = connect_ctl(sockpath, false, NULL, NULL);

	rc = write(fd, &buf, sizeof(buf));
	if (rc < 0)
		die("write() to %s: %s\n", sockpath, strerror(errno));

	close(fd);
}

int main(int argc, char *argv[])
{
	const char *subcmd = argv[1];
	int fd;

	if (argc < 2)
		usage();

	fd = socket(AF_UNIX, SOCK_STREAM, PF_UNIX);
	if (fd < 0)
		die("socket(): %s\n", strerror(errno));

	if (strcmp(subcmd, "hold") == 0)
		cmd_hold(argc - 1, argv + 1);
	else if (strcmp(subcmd, "info") == 0)
		cmd_info(argc - 1, argv + 1);
	else if (strcmp(subcmd, "exec") == 0)
		cmd_exec(argc - 1, argv + 1);
	else if (strcmp(subcmd, "stop") == 0)
		cmd_stop(argc - 1, argv + 1);
	else
		usage();

	exit(0);
}