summaryrefslogtreecommitdiff
path: root/src/system.c
blob: 3b43b2a967106fc7903178b5402a22f0211ef672 (plain)
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
/**
 * Seccomp System Interfaces
 *
 * Copyright (c) 2014 Red Hat <pmoore@redhat.com>
 * Author: Paul Moore <paul@paul-moore.com>
 */

/*
 * This library is free software; you can redistribute it and/or modify it
 * under the terms of version 2.1 of the GNU Lesser General Public License as
 * published by the Free Software Foundation.
 *
 * This library is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public License
 * for more details.
 *
 * You should have received a copy of the GNU Lesser General Public License
 * along with this library; if not, see <http://www.gnu.org/licenses>.
 */

#include <stdlib.h>
#include <errno.h>
#include <sys/prctl.h>

#define _GNU_SOURCE
#include <unistd.h>

#include "system.h"

#include <seccomp.h>

#include "arch.h"
#include "db.h"
#include "gen_bpf.h"
#include "helper.h"

/* NOTE: the seccomp syscall allowlist is currently disabled for testing
 *       purposes, but unless we can verify all of the supported ABIs before
 *       our next release we may have to enable the allowlist */
#define SYSCALL_ALLOWLIST_ENABLE	0

/* task global state */
struct task_state {
	/* seccomp(2) syscall */
	int nr_seccomp;

	/* userspace notification fd */
	int notify_fd;

	/* runtime support flags */
	int sup_syscall;
	int sup_flag_tsync;
	int sup_flag_log;
	int sup_action_log;
	int sup_kill_process;
	int sup_flag_spec_allow;
	int sup_flag_new_listener;
	int sup_user_notif;
	int sup_flag_tsync_esrch;
};
static struct task_state state = {
	.nr_seccomp = -1,

	.notify_fd = -1,

	.sup_syscall = -1,
	.sup_flag_tsync = -1,
	.sup_flag_log = -1,
	.sup_action_log = -1,
	.sup_kill_process = -1,
	.sup_flag_spec_allow = -1,
	.sup_flag_new_listener = -1,
	.sup_user_notif = -1,
	.sup_flag_tsync_esrch = -1,
};

/**
 * Reset the task state
 *
 * This function fully resets the library's global "system task state".
 *
 */
void sys_reset_state(void)
{
	state.nr_seccomp = -1;
	state.notify_fd = -1;
	state.sup_syscall = -1;
	state.sup_flag_tsync = -1;
	state.sup_flag_log = -1;
	state.sup_action_log = -1;
	state.sup_kill_process = -1;
	state.sup_flag_spec_allow = -1;
	state.sup_flag_new_listener = -1;
	state.sup_user_notif = -1;
	state.sup_flag_tsync_esrch = -1;
}

/**
 * Check to see if the seccomp() syscall is supported
 *
 * This function attempts to see if the system supports the seccomp() syscall.
 * Unfortunately, there are a few reasons why this check may fail, including
 * a previously loaded seccomp filter, so it is hard to say for certain.
 * Return one if the syscall is supported, zero otherwise.
 *
 */
int sys_chk_seccomp_syscall(void)
{
	int rc;
	int nr_seccomp;

	/* NOTE: it is reasonably safe to assume that we should be able to call
	 *       seccomp() when the caller first starts, but we can't rely on
	 *       it later so we need to cache our findings for use later */
	if (state.sup_syscall >= 0)
		return state.sup_syscall;

#if SYSCALL_ALLOWLIST_ENABLE
	/* architecture allowlist */
	switch (arch_def_native->token) {
	case SCMP_ARCH_X86_64:
	case SCMP_ARCH_ARM:
	case SCMP_ARCH_AARCH64:
	case SCMP_ARCH_PPC64:
	case SCMP_ARCH_PPC64LE:
	case SCMP_ARCH_S390:
	case SCMP_ARCH_S390X:
	case SCMP_ARCH_RISCV64:
		break;
	default:
		goto unsupported;
	}
#endif

	nr_seccomp = arch_syscall_resolve_name(arch_def_native, "seccomp");
	if (nr_seccomp < 0)
		goto unsupported;

	/* this is an invalid call because the second argument is non-zero, but
	 * depending on the errno value of ENOSYS or EINVAL we can guess if the
	 * seccomp() syscall is supported or not */
	rc = syscall(nr_seccomp, SECCOMP_SET_MODE_STRICT, 1, NULL);
	if (rc < 0 && errno == EINVAL)
		goto supported;

unsupported:
	state.sup_syscall = 0;
	return 0;
supported:
	state.nr_seccomp = nr_seccomp;
	state.sup_syscall = 1;
	return 1;
}

/**
 * Force the seccomp() syscall support setting
 * @param enable the intended support state
 *
 * This function overrides the current seccomp() syscall support setting; this
 * is very much a "use at your own risk" function.
 *
 */
void sys_set_seccomp_syscall(bool enable)
{
	state.sup_syscall = (enable ? 1 : 0);
}

/**
 * Check to see if a seccomp action is supported
 * @param action the seccomp action
 *
 * This function checks to see if a seccomp action is supported by the system.
 * Return one if the action is supported, zero otherwise.
 *
 */
int sys_chk_seccomp_action(uint32_t action)
{
	if (action == SCMP_ACT_KILL_PROCESS) {
		if (state.sup_kill_process < 0) {
			if (sys_chk_seccomp_syscall() == 1 &&
			    syscall(state.nr_seccomp,
				    SECCOMP_GET_ACTION_AVAIL, 0, &action) == 0)
				state.sup_kill_process = 1;
			else
				state.sup_kill_process = 0;
		}

		return state.sup_kill_process;
	} else if (action == SCMP_ACT_KILL_THREAD) {
		return 1;
	} else if (action == SCMP_ACT_TRAP) {
		return 1;
	} else if ((action == SCMP_ACT_ERRNO(action & 0x0000ffff)) &&
		   ((action & 0x0000ffff) < MAX_ERRNO)) {
		return 1;
	} else if (action == SCMP_ACT_TRACE(action & 0x0000ffff)) {
		return 1;
	} else if (action == SCMP_ACT_LOG) {
		if (state.sup_action_log < 0) {
			if (sys_chk_seccomp_syscall() == 1 &&
			    syscall(state.nr_seccomp,
				    SECCOMP_GET_ACTION_AVAIL, 0, &action) == 0)
				state.sup_action_log = 1;
			else
				state.sup_action_log = 0;
		}

		return state.sup_action_log;
	} else if (action == SCMP_ACT_ALLOW) {
		return 1;
	} else if (action == SCMP_ACT_NOTIFY) {
		if (state.sup_user_notif < 0) {
			struct seccomp_notif_sizes sizes;
			if (sys_chk_seccomp_syscall() == 1 &&
			    syscall(state.nr_seccomp,
				    SECCOMP_GET_NOTIF_SIZES, 0, &sizes) == 0)
				state.sup_user_notif = 1;
			else
				state.sup_user_notif = 0;
		}

		return state.sup_user_notif;
	}

	return 0;
}

/**
 * Force a seccomp action support setting
 * @param action the seccomp action
 * @param enable the intended support state
 *
 * This function overrides the current seccomp action support setting; this
 * is very much a "use at your own risk" function.
 */
void sys_set_seccomp_action(uint32_t action, bool enable)
{
	switch (action) {
	case SCMP_ACT_LOG:
		state.sup_action_log = (enable ? 1 : 0);
		break;
	case SCMP_ACT_KILL_PROCESS:
		state.sup_kill_process = (enable ? 1 : 0);
		break;
	case SCMP_ACT_NOTIFY:
		state.sup_user_notif = (enable ? 1 : 0);
		break;
	}
}

/**
 * Check to see if a seccomp() flag is supported by the kernel
 * @param flag the seccomp() flag
 *
 * This function checks to see if a seccomp() flag is supported by the kernel.
 * Return one if the flag is supported, zero otherwise.
 *
 */
static int _sys_chk_flag_kernel(int flag)
{
	/* this is an invalid seccomp(2) call because the last argument
	 * is NULL, but depending on the errno value of EFAULT we can
	 * guess if the filter flag is supported or not */
	if (sys_chk_seccomp_syscall() == 1 &&
	    syscall(state.nr_seccomp,
		    SECCOMP_SET_MODE_FILTER, flag, NULL) == -1 &&
	    errno == EFAULT)
		return 1;

	return 0;
}

/**
 * Check to see if a seccomp() flag is supported
 * @param flag the seccomp() flag
 *
 * This function checks to see if a seccomp() flag is supported by the system.
 * Return one if the syscall is supported, zero if unsupported, negative values
 * on error.
 *
 */
int sys_chk_seccomp_flag(int flag)
{
	switch (flag) {
	case SECCOMP_FILTER_FLAG_TSYNC:
		if (state.sup_flag_tsync < 0)
			state.sup_flag_tsync = _sys_chk_flag_kernel(flag);
		return state.sup_flag_tsync;
	case SECCOMP_FILTER_FLAG_LOG:
		if (state.sup_flag_log < 0)
			state.sup_flag_log = _sys_chk_flag_kernel(flag);
		return state.sup_flag_log;
	case SECCOMP_FILTER_FLAG_SPEC_ALLOW:
		if (state.sup_flag_spec_allow < 0)
			state.sup_flag_spec_allow = _sys_chk_flag_kernel(flag);
		return state.sup_flag_spec_allow;
	case SECCOMP_FILTER_FLAG_NEW_LISTENER:
		if (state.sup_flag_new_listener < 0)
			state.sup_flag_new_listener = _sys_chk_flag_kernel(flag);
		return state.sup_flag_new_listener;
	case SECCOMP_FILTER_FLAG_TSYNC_ESRCH:
		if (state.sup_flag_tsync_esrch < 0)
			state.sup_flag_tsync_esrch = _sys_chk_flag_kernel(flag);
		return state.sup_flag_tsync_esrch;
	}

	return -EOPNOTSUPP;
}

/**
 * Force a seccomp() syscall flag support setting
 * @param flag the seccomp() flag
 * @param enable the intended support state
 *
 * This function overrides the current seccomp() syscall support setting for a
 * given flag; this is very much a "use at your own risk" function.
 *
 */
void sys_set_seccomp_flag(int flag, bool enable)
{
	switch (flag) {
	case SECCOMP_FILTER_FLAG_TSYNC:
		state.sup_flag_tsync = (enable ? 1 : 0);
		break;
	case SECCOMP_FILTER_FLAG_LOG:
		state.sup_flag_log = (enable ? 1 : 0);
		break;
	case SECCOMP_FILTER_FLAG_SPEC_ALLOW:
		state.sup_flag_spec_allow = (enable ? 1 : 0);
		break;
	case SECCOMP_FILTER_FLAG_NEW_LISTENER:
		state.sup_flag_new_listener = (enable ? 1 : 0);
		break;
	case SECCOMP_FILTER_FLAG_TSYNC_ESRCH:
		state.sup_flag_tsync_esrch = (enable ? 1 : 0);
		break;
	}
}

/**
 * Loads the filter into the kernel
 * @param col the filter collection
 * @param rawrc pass the raw return code if true
 *
 * This function loads the given seccomp filter context into the kernel.  If
 * the filter was loaded correctly, the kernel will be enforcing the filter
 * when this function returns.  Returns zero on success, negative values on
 * error.
 *
 */
int sys_filter_load(struct db_filter_col *col, bool rawrc)
{
	int rc;
	bool tsync_notify;
	struct bpf_program *prgm = NULL;

	rc = gen_bpf_generate(col, &prgm);
	if (rc < 0)
		return rc;

	/* attempt to set NO_NEW_PRIVS */
	if (col->attr.nnp_enable) {
		rc = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
		if (rc < 0)
			goto filter_load_out;
	}

	tsync_notify = state.sup_flag_tsync_esrch > 0 && state.notify_fd == -1;

	/* load the filter into the kernel */
	if (sys_chk_seccomp_syscall() == 1) {
		int flgs = 0;
		if (tsync_notify) {
			if (col->attr.tsync_enable)
				flgs |= SECCOMP_FILTER_FLAG_TSYNC | \
					SECCOMP_FILTER_FLAG_TSYNC_ESRCH;
			if (state.sup_user_notif > 0)
				flgs |= SECCOMP_FILTER_FLAG_NEW_LISTENER;
		} else if (col->attr.tsync_enable)
			flgs |= SECCOMP_FILTER_FLAG_TSYNC;
		else if (state.sup_user_notif > 0 && state.notify_fd == -1)
			flgs |= SECCOMP_FILTER_FLAG_NEW_LISTENER;
		if (col->attr.log_enable)
			flgs |= SECCOMP_FILTER_FLAG_LOG;
		if (col->attr.spec_allow)
			flgs |= SECCOMP_FILTER_FLAG_SPEC_ALLOW;
		rc = syscall(state.nr_seccomp,
			     SECCOMP_SET_MODE_FILTER, flgs, prgm);
		if (tsync_notify && rc > 0) {
			/* return 0 on NEW_LISTENER success, but save the fd */
			state.notify_fd = rc;
			rc = 0;
		} else if (rc > 0 && col->attr.tsync_enable) {
			/* always return -ESRCH if we fail to sync threads */
			errno = ESRCH;
			rc = -errno;
		} else if (rc > 0 && state.sup_user_notif > 0) {
			/* return 0 on NEW_LISTENER success, but save the fd */
			state.notify_fd = rc;
			rc = 0;
		}
	} else
		rc = prctl(PR_SET_SECCOMP, SECCOMP_MODE_FILTER, prgm);

filter_load_out:
	/* cleanup and return */
	gen_bpf_release(prgm);
	if (rc == -ESRCH)
		return -ESRCH;
	if (rc < 0)
		return (rawrc ? -errno : -ECANCELED);
	return rc;
}

/**
 * Return the userspace notification fd
 *
 * This function returns the userspace notification fd from
 * SECCOMP_FILTER_FLAG_NEW_LISTENER.  If the notification fd has not yet been
 * set, or an error has occurred, -1 is returned.
 *
 */
int sys_notify_fd(void)
{
	return state.notify_fd;
}

/**
 * Allocate a pair of notification request/response structures
 * @param req the request location
 * @param resp the response location
 *
 * This function allocates a pair of request/response structure by computing
 * the correct sized based on the currently running kernel. It returns zero on
 * success, and negative values on failure.
 *
 */
int sys_notify_alloc(struct seccomp_notif **req,
		     struct seccomp_notif_resp **resp)
{
	int rc;
	static struct seccomp_notif_sizes sizes = { 0, 0, 0 };

	if (state.sup_syscall <= 0)
		return -EOPNOTSUPP;

	if (sizes.seccomp_notif == 0 && sizes.seccomp_notif_resp == 0) {
		rc = syscall(__NR_seccomp, SECCOMP_GET_NOTIF_SIZES, 0, &sizes);
		if (rc < 0)
			return -ECANCELED;
	}
	if (sizes.seccomp_notif == 0 || sizes.seccomp_notif_resp == 0)
		return -EFAULT;

	if (req) {
		*req = zmalloc(sizes.seccomp_notif);
		if (!*req)
			return -ENOMEM;
	}

	if (resp) {
		*resp = zmalloc(sizes.seccomp_notif_resp);
		if (!*resp) {
			if (req)
				free(*req);
			return -ENOMEM;
		}
	}

	return 0;
}

/**
 * Receive a notification from a seccomp notification fd
 * @param fd the notification fd
 * @param req the request buffer to save into
 *
 * Blocks waiting for a notification on this fd. This function is thread safe
 * (synchronization is performed in the kernel). Returns zero on success,
 * negative values on error.
 *
 */
int sys_notify_receive(int fd, struct seccomp_notif *req)
{
	if (state.sup_user_notif <= 0)
		return -EOPNOTSUPP;

	if (ioctl(fd, SECCOMP_IOCTL_NOTIF_RECV, req) < 0)
		return -ECANCELED;

	return 0;
}

/**
 * Send a notification response to a seccomp notification fd
 * @param fd the notification fd
 * @param resp the response buffer to use
 *
 * Sends a notification response on this fd. This function is thread safe
 * (synchronization is performed in the kernel). Returns zero on success,
 * negative values on error.
 *
 */
int sys_notify_respond(int fd, struct seccomp_notif_resp *resp)
{
	if (state.sup_user_notif <= 0)
		return -EOPNOTSUPP;

	if (ioctl(fd, SECCOMP_IOCTL_NOTIF_SEND, resp) < 0)
		return -ECANCELED;
	return 0;
}

/**
 * Check if a notification id is still valid
 * @param fd the notification fd
 * @param id the id to test
 *
 * Checks to see if a notification id is still valid. Returns 0 on success, and
 * negative values on failure.
 *
 */
int sys_notify_id_valid(int fd, uint64_t id)
{
	if (state.sup_user_notif <= 0)
		return -EOPNOTSUPP;

	if (ioctl(fd, SECCOMP_IOCTL_NOTIF_ID_VALID, &id) < 0)
		return -ENOENT;
	return 0;
}