summaryrefslogtreecommitdiffstats
path: root/src/core/bpf-devices.c
blob: 06d2146dcdd35a93da44c33b875bbe1743cb295d (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
/* SPDX-License-Identifier: LGPL-2.1-or-later */

#include <fnmatch.h>
#include <linux/bpf_insn.h>

#include "bpf-devices.h"
#include "bpf-program.h"
#include "devnum-util.h"
#include "fd-util.h"
#include "fileio.h"
#include "nulstr-util.h"
#include "parse-util.h"
#include "path-util.h"
#include "stdio-util.h"
#include "string-util.h"

#define PASS_JUMP_OFF 4096

/* Ensure the high level flags we use and the low-level BPF flags exposed on the kernel are defined the same way */
assert_cc((unsigned) BPF_DEVCG_ACC_MKNOD == (unsigned) CGROUP_DEVICE_MKNOD);
assert_cc((unsigned) BPF_DEVCG_ACC_READ  == (unsigned) CGROUP_DEVICE_READ);
assert_cc((unsigned) BPF_DEVCG_ACC_WRITE == (unsigned) CGROUP_DEVICE_WRITE);

static int bpf_prog_allow_list_device(
                BPFProgram *prog,
                char type,
                int major,
                int minor,
                CGroupDevicePermissions p) {

        int r;

        assert(prog);

        log_trace("%s: %c %d:%d %s", __func__, type, major, minor, cgroup_device_permissions_to_string(p));

        if (p <= 0 || p >= _CGROUP_DEVICE_PERMISSIONS_MAX)
                return -EINVAL;

        assert(IN_SET(type, 'b', 'c'));
        const int bpf_type = type == 'c' ? BPF_DEVCG_DEV_CHAR : BPF_DEVCG_DEV_BLOCK;

        const struct bpf_insn insn[] = {
                BPF_MOV32_REG(BPF_REG_1, BPF_REG_3),
                BPF_ALU32_IMM(BPF_AND, BPF_REG_1, p),
                BPF_JMP_REG(BPF_JNE, BPF_REG_1, BPF_REG_3, 4), /* compare access type */

                BPF_JMP_IMM(BPF_JNE, BPF_REG_2, bpf_type, 3),  /* compare device type */
                BPF_JMP_IMM(BPF_JNE, BPF_REG_4, major, 2),     /* compare major */
                BPF_JMP_IMM(BPF_JNE, BPF_REG_5, minor, 1),     /* compare minor */
                BPF_JMP_A(PASS_JUMP_OFF),                      /* jump to PASS */
        };

        if (p == _CGROUP_DEVICE_PERMISSIONS_ALL)
                r = bpf_program_add_instructions(prog, insn + 3, ELEMENTSOF(insn) - 3);
        else
                r = bpf_program_add_instructions(prog, insn, ELEMENTSOF(insn));
        if (r < 0)
                log_error_errno(r, "Extending device control BPF program failed: %m");

        return r;
}

static int bpf_prog_allow_list_major(
                BPFProgram *prog,
                char type,
                int major,
                CGroupDevicePermissions p) {

        int r;

        assert(prog);

        log_trace("%s: %c %d:* %s", __func__, type, major, cgroup_device_permissions_to_string(p));

        if (p <= 0 || p >= _CGROUP_DEVICE_PERMISSIONS_MAX)
                return -EINVAL;

        assert(IN_SET(type, 'b', 'c'));
        const int bpf_type = type == 'c' ? BPF_DEVCG_DEV_CHAR : BPF_DEVCG_DEV_BLOCK;

        const struct bpf_insn insn[] = {
                BPF_MOV32_REG(BPF_REG_1, BPF_REG_3),
                BPF_ALU32_IMM(BPF_AND, BPF_REG_1, p),
                BPF_JMP_REG(BPF_JNE, BPF_REG_1, BPF_REG_3, 3), /* compare access type */

                BPF_JMP_IMM(BPF_JNE, BPF_REG_2, bpf_type, 2),  /* compare device type */
                BPF_JMP_IMM(BPF_JNE, BPF_REG_4, major, 1),     /* compare major */
                BPF_JMP_A(PASS_JUMP_OFF),                      /* jump to PASS */
        };

        if (p == _CGROUP_DEVICE_PERMISSIONS_ALL)
                r = bpf_program_add_instructions(prog, insn + 3, ELEMENTSOF(insn) - 3);
        else
                r = bpf_program_add_instructions(prog, insn, ELEMENTSOF(insn));
        if (r < 0)
                log_error_errno(r, "Extending device control BPF program failed: %m");

        return r;
}

static int bpf_prog_allow_list_class(
                BPFProgram *prog,
                char type,
                CGroupDevicePermissions p) {

        int r;

        assert(prog);

        log_trace("%s: %c *:* %s", __func__, type, cgroup_device_permissions_to_string(p));

        if (p <= 0 || p >= _CGROUP_DEVICE_PERMISSIONS_MAX)
                return -EINVAL;

        assert(IN_SET(type, 'b', 'c'));
        const int bpf_type = type == 'c' ? BPF_DEVCG_DEV_CHAR : BPF_DEVCG_DEV_BLOCK;

        const struct bpf_insn insn[] = {
                BPF_MOV32_REG(BPF_REG_1, BPF_REG_3),
                BPF_ALU32_IMM(BPF_AND, BPF_REG_1, p),
                BPF_JMP_REG(BPF_JNE, BPF_REG_1, BPF_REG_3, 2), /* compare access type */

                BPF_JMP_IMM(BPF_JNE, BPF_REG_2, bpf_type, 1), /* compare device type */
                BPF_JMP_A(PASS_JUMP_OFF),                     /* jump to PASS */
        };

        if (p == _CGROUP_DEVICE_PERMISSIONS_ALL)
                r = bpf_program_add_instructions(prog, insn + 3, ELEMENTSOF(insn) - 3);
        else
                r = bpf_program_add_instructions(prog, insn, ELEMENTSOF(insn));
        if (r < 0)
                log_error_errno(r, "Extending device control BPF program failed: %m");

        return r;
}

int bpf_devices_cgroup_init(
                BPFProgram **ret,
                CGroupDevicePolicy policy,
                bool allow_list) {

        const struct bpf_insn pre_insn[] = {
                /* load device type to r2 */
                BPF_LDX_MEM(BPF_W, BPF_REG_2, BPF_REG_1,
                            offsetof(struct bpf_cgroup_dev_ctx, access_type)),
                BPF_ALU32_IMM(BPF_AND, BPF_REG_2, 0xFFFF),

                /* load access type to r3 */
                BPF_LDX_MEM(BPF_W, BPF_REG_3, BPF_REG_1,
                            offsetof(struct bpf_cgroup_dev_ctx, access_type)),
                BPF_ALU32_IMM(BPF_RSH, BPF_REG_3, 16),

                /* load major number to r4 */
                BPF_LDX_MEM(BPF_W, BPF_REG_4, BPF_REG_1,
                            offsetof(struct bpf_cgroup_dev_ctx, major)),

                /* load minor number to r5 */
                BPF_LDX_MEM(BPF_W, BPF_REG_5, BPF_REG_1,
                            offsetof(struct bpf_cgroup_dev_ctx, minor)),
        };

        _cleanup_(bpf_program_freep) BPFProgram *prog = NULL;
        int r;

        assert(ret);

        if (policy == CGROUP_DEVICE_POLICY_AUTO && !allow_list)
                return 0;

        r = bpf_program_new(BPF_PROG_TYPE_CGROUP_DEVICE, "sd_devices", &prog);
        if (r < 0)
                return log_error_errno(r, "Loading device control BPF program failed: %m");

        if (policy == CGROUP_DEVICE_POLICY_CLOSED || allow_list) {
                r = bpf_program_add_instructions(prog, pre_insn, ELEMENTSOF(pre_insn));
                if (r < 0)
                        return log_error_errno(r, "Extending device control BPF program failed: %m");
        }

        *ret = TAKE_PTR(prog);

        return 0;
}

int bpf_devices_apply_policy(
                BPFProgram **prog,
                CGroupDevicePolicy policy,
                bool allow_list,
                const char *cgroup_path,
                BPFProgram **prog_installed) {

        _cleanup_free_ char *controller_path = NULL;
        int r;

        /* This will assign *prog_installed if everything goes well. */

        assert(prog);
        if (!*prog)
                goto finish;

        const bool deny_everything = policy == CGROUP_DEVICE_POLICY_STRICT && !allow_list;

        const struct bpf_insn post_insn[] = {
                /* return DENY */
                BPF_MOV64_IMM(BPF_REG_0, 0),
                BPF_JMP_A(1),
        };

        const struct bpf_insn exit_insn[] = {
                /* finally return DENY if deny_everything else ALLOW */
                BPF_MOV64_IMM(BPF_REG_0, deny_everything ? 0 : 1),
                BPF_EXIT_INSN()
        };

        if (!deny_everything) {
                r = bpf_program_add_instructions(*prog, post_insn, ELEMENTSOF(post_insn));
                if (r < 0)
                        return log_error_errno(r, "Extending device control BPF program failed: %m");

                /* Fixup PASS_JUMP_OFF jump offsets. */
                for (size_t off = 0; off < (*prog)->n_instructions; off++) {
                        struct bpf_insn *ins = &((*prog)->instructions[off]);

                        if (ins->code == (BPF_JMP | BPF_JA) && ins->off == PASS_JUMP_OFF)
                                ins->off = (*prog)->n_instructions - off - 1;
                }
        }

        r = bpf_program_add_instructions(*prog, exit_insn, ELEMENTSOF(exit_insn));
        if (r < 0)
                return log_error_errno(r, "Extending device control BPF program failed: %m");

        r = cg_get_path(SYSTEMD_CGROUP_CONTROLLER, cgroup_path, NULL, &controller_path);
        if (r < 0)
                return log_error_errno(r, "Failed to determine cgroup path: %m");

        r = bpf_program_cgroup_attach(*prog, BPF_CGROUP_DEVICE, controller_path, BPF_F_ALLOW_MULTI);
        if (r < 0)
                return log_error_errno(r, "Attaching device control BPF program to cgroup %s failed: %m",
                                       empty_to_root(cgroup_path));

 finish:
        /* Unref the old BPF program (which will implicitly detach it) right before attaching the new program. */
        if (prog_installed) {
                bpf_program_free(*prog_installed);
                *prog_installed = TAKE_PTR(*prog);
        }
        return 0;
}

int bpf_devices_supported(void) {
        const struct bpf_insn trivial[] = {
                BPF_MOV64_IMM(BPF_REG_0, 1),
                BPF_EXIT_INSN()
        };

        _cleanup_(bpf_program_freep) BPFProgram *program = NULL;
        static int supported = -1;
        int r;

        /* Checks whether BPF device controller is supported. For this, we check five things:
         *
         * a) whether we are privileged
         * b) whether the unified hierarchy is being used
         * c) the BPF implementation in the kernel supports BPF_PROG_TYPE_CGROUP_DEVICE programs, which we require
         */

        if (supported >= 0)
                return supported;

        if (geteuid() != 0) {
                log_debug("Not enough privileges, BPF device control is not supported.");
                return supported = 0;
        }

        r = cg_unified_controller(SYSTEMD_CGROUP_CONTROLLER);
        if (r < 0)
                return log_error_errno(r, "Can't determine whether the unified hierarchy is used: %m");
        if (r == 0) {
                log_debug("Not running with unified cgroups, BPF device control is not supported.");
                return supported = 0;
        }

        r = bpf_program_new(BPF_PROG_TYPE_CGROUP_DEVICE, "sd_devices", &program);
        if (r < 0) {
                log_debug_errno(r, "Can't allocate CGROUP DEVICE BPF program, BPF device control is not supported: %m");
                return supported = 0;
        }

        r = bpf_program_add_instructions(program, trivial, ELEMENTSOF(trivial));
        if (r < 0) {
                log_debug_errno(r, "Can't add trivial instructions to CGROUP DEVICE BPF program, BPF device control is not supported: %m");
                return supported = 0;
        }

        r = bpf_program_load_kernel(program, NULL, 0);
        if (r < 0) {
                log_debug_errno(r, "Can't load kernel CGROUP DEVICE BPF program, BPF device control is not supported: %m");
                return supported = 0;
        }

        return supported = 1;
}

static int allow_list_device_pattern(
                BPFProgram *prog,
                const char *path,
                char type,
                const unsigned *maj,
                const unsigned *min,
                CGroupDevicePermissions p) {

        assert(IN_SET(type, 'b', 'c'));

        if (cg_all_unified() > 0) {
                if (!prog)
                        return 0;

                if (maj && min)
                        return bpf_prog_allow_list_device(prog, type, *maj, *min, p);
                else if (maj)
                        return bpf_prog_allow_list_major(prog, type, *maj, p);
                else
                        return bpf_prog_allow_list_class(prog, type, p);

        } else {
                char buf[2+DECIMAL_STR_MAX(unsigned)*2+2+4];
                int r;

                if (maj && min)
                        xsprintf(buf, "%c %u:%u %s", type, *maj, *min, cgroup_device_permissions_to_string(p));
                else if (maj)
                        xsprintf(buf, "%c %u:* %s", type, *maj, cgroup_device_permissions_to_string(p));
                else
                        xsprintf(buf, "%c *:* %s", type, cgroup_device_permissions_to_string(p));

                /* Changing the devices list of a populated cgroup might result in EINVAL, hence ignore
                 * EINVAL here. */

                r = cg_set_attribute("devices", path, "devices.allow", buf);
                if (r < 0)
                        log_full_errno(IN_SET(r, -ENOENT, -EROFS, -EINVAL, -EACCES, -EPERM) ? LOG_DEBUG : LOG_WARNING,
                                       r, "Failed to set devices.allow on %s: %m", path);

                return r;
        }
}

int bpf_devices_allow_list_device(
                BPFProgram *prog,
                const char *path,
                const char *node,
                CGroupDevicePermissions p) {

        mode_t mode;
        dev_t rdev;
        int r;

        assert(path);
        assert(p >= 0 && p < _CGROUP_DEVICE_PERMISSIONS_MAX);

        log_trace("%s: %s %s", __func__, node, cgroup_device_permissions_to_string(p));

        /* Some special handling for /dev/block/%u:%u, /dev/char/%u:%u, /run/systemd/inaccessible/chr and
         * /run/systemd/inaccessible/blk paths. Instead of stat()ing these we parse out the major/minor directly. This
         * means clients can use these path without the device node actually around */
        r = device_path_parse_major_minor(node, &mode, &rdev);
        if (r < 0) {
                if (r != -ENODEV)
                        return log_warning_errno(r, "Couldn't parse major/minor from device path '%s': %m", node);

                struct stat st;
                if (stat(node, &st) < 0)
                        return log_warning_errno(errno, "Couldn't stat device %s: %m", node);

                if (!S_ISCHR(st.st_mode) && !S_ISBLK(st.st_mode))
                        return log_warning_errno(SYNTHETIC_ERRNO(ENODEV), "%s is not a device.", node);

                mode = st.st_mode;
                rdev = (dev_t) st.st_rdev;
        }

        unsigned maj = major(rdev), min = minor(rdev);
        return allow_list_device_pattern(prog, path, S_ISCHR(mode) ? 'c' : 'b', &maj, &min, p);
}

int bpf_devices_allow_list_major(
                BPFProgram *prog,
                const char *path,
                const char *name,
                char type,
                CGroupDevicePermissions permissions) {

        unsigned maj;
        int r;

        assert(path);
        assert(IN_SET(type, 'b', 'c'));
        assert(permissions >= 0 && permissions < _CGROUP_DEVICE_PERMISSIONS_MAX);

        if (streq(name, "*"))
                /* If the name is a wildcard, then apply this list to all devices of this type */
                return allow_list_device_pattern(prog, path, type, NULL, NULL, permissions);

        if (safe_atou(name, &maj) >= 0 && DEVICE_MAJOR_VALID(maj))
                /* The name is numeric and suitable as major. In that case, let's take its major, and create
                 * the entry directly. */
                return allow_list_device_pattern(prog, path, type, &maj, NULL, permissions);

        _cleanup_fclose_ FILE *f = NULL;
        bool good = false, any = false;

        f = fopen("/proc/devices", "re");
        if (!f)
                return log_warning_errno(errno, "Cannot open /proc/devices to resolve %s: %m", name);

        for (;;) {
                _cleanup_free_ char *line = NULL;
                char *w, *p;

                r = read_line(f, LONG_LINE_MAX, &line);
                if (r < 0)
                        return log_warning_errno(r, "Failed to read /proc/devices: %m");
                if (r == 0)
                        break;

                if (type == 'c' && streq(line, "Character devices:")) {
                        good = true;
                        continue;
                }

                if (type == 'b' && streq(line, "Block devices:")) {
                        good = true;
                        continue;
                }

                if (isempty(line)) {
                        good = false;
                        continue;
                }

                if (!good)
                        continue;

                p = strstrip(line);

                w = strpbrk(p, WHITESPACE);
                if (!w)
                        continue;
                *w = 0;

                r = safe_atou(p, &maj);
                if (r < 0)
                        continue;
                if (maj <= 0)
                        continue;

                w++;
                w += strspn(w, WHITESPACE);

                if (fnmatch(name, w, 0) != 0)
                        continue;

                any = true;
                (void) allow_list_device_pattern(prog, path, type, &maj, NULL, permissions);
        }

        if (!any)
                return log_debug_errno(SYNTHETIC_ERRNO(ENOENT),
                                       "Device allow list pattern \"%s\" did not match anything.", name);

        return 0;
}

int bpf_devices_allow_list_static(
                BPFProgram *prog,
                const char *path) {

        static const char auto_devices[] =
                "/dev/null\0" "rwm\0"
                "/dev/zero\0" "rwm\0"
                "/dev/full\0" "rwm\0"
                "/dev/random\0" "rwm\0"
                "/dev/urandom\0" "rwm\0"
                "/dev/tty\0" "rwm\0"
                "/dev/ptmx\0" "rwm\0"
                /* Allow /run/systemd/inaccessible/{chr,blk} devices for mapping InaccessiblePaths */
                "/run/systemd/inaccessible/chr\0" "rwm\0"
                "/run/systemd/inaccessible/blk\0" "rwm\0";
        int r = 0, k;

        NULSTR_FOREACH_PAIR(node, acc, auto_devices) {
                k = bpf_devices_allow_list_device(prog, path, node, cgroup_device_permissions_from_string(acc));
                if (r >= 0 && k < 0)
                        r = k;
        }

        /* PTS (/dev/pts) devices may not be duplicated, but accessed */
        k = bpf_devices_allow_list_major(prog, path, "pts", 'c', CGROUP_DEVICE_READ|CGROUP_DEVICE_WRITE);
        if (r >= 0 && k < 0)
                r = k;

        return r;
}