device_cgroup.c 13.2 KB
Newer Older
1
/*
2
 * device_cgroup.c - device cgroup subsystem
3 4 5 6 7 8 9 10 11
 *
 * Copyright 2007 IBM Corp
 */

#include <linux/device_cgroup.h>
#include <linux/cgroup.h>
#include <linux/ctype.h>
#include <linux/list.h>
#include <linux/uaccess.h>
12
#include <linux/seq_file.h>
13
#include <linux/slab.h>
14
#include <linux/rcupdate.h>
15
#include <linux/mutex.h>
16 17 18 19 20 21 22 23 24 25

#define ACC_MKNOD 1
#define ACC_READ  2
#define ACC_WRITE 4
#define ACC_MASK (ACC_MKNOD | ACC_READ | ACC_WRITE)

#define DEV_BLOCK 1
#define DEV_CHAR  2
#define DEV_ALL   4  /* this represents all devices */

26 27
static DEFINE_MUTEX(devcgroup_mutex);

28
/*
29
 * exception list locking rules:
30
 * hold devcgroup_mutex for update/read.
31
 * hold rcu_read_lock() for read.
32 33
 */

34
struct dev_exception_item {
35 36 37 38
	u32 major, minor;
	short type;
	short access;
	struct list_head list;
39
	struct rcu_head rcu;
40 41 42 43
};

struct dev_cgroup {
	struct cgroup_subsys_state css;
44
	struct list_head exceptions;
45
	bool deny_all;
46 47
};

48 49 50 51 52
static inline struct dev_cgroup *css_to_devcgroup(struct cgroup_subsys_state *s)
{
	return container_of(s, struct dev_cgroup, css);
}

53 54
static inline struct dev_cgroup *cgroup_to_devcgroup(struct cgroup *cgroup)
{
55
	return css_to_devcgroup(cgroup_subsys_state(cgroup, devices_subsys_id));
56 57
}

58 59 60 61 62
static inline struct dev_cgroup *task_devcgroup(struct task_struct *task)
{
	return css_to_devcgroup(task_subsys_state(task, devices_subsys_id));
}

63 64
struct cgroup_subsys devices_subsys;

65 66
static int devcgroup_can_attach(struct cgroup *new_cgrp,
				struct cgroup_taskset *set)
67
{
68
	struct task_struct *task = cgroup_taskset_first(set);
69

70 71
	if (current != task && !capable(CAP_SYS_ADMIN))
		return -EPERM;
72 73 74 75
	return 0;
}

/*
76
 * called under devcgroup_mutex
77
 */
78
static int dev_exceptions_copy(struct list_head *dest, struct list_head *orig)
79
{
80
	struct dev_exception_item *ex, *tmp, *new;
81

82 83
	list_for_each_entry(ex, orig, list) {
		new = kmemdup(ex, sizeof(*ex), GFP_KERNEL);
84 85 86 87 88 89 90 91
		if (!new)
			goto free_and_exit;
		list_add_tail(&new->list, dest);
	}

	return 0;

free_and_exit:
92 93 94
	list_for_each_entry_safe(ex, tmp, dest, list) {
		list_del(&ex->list);
		kfree(ex);
95 96 97 98 99
	}
	return -ENOMEM;
}

/*
100
 * called under devcgroup_mutex
101
 */
102 103
static int dev_exception_add(struct dev_cgroup *dev_cgroup,
			     struct dev_exception_item *ex)
104
{
105
	struct dev_exception_item *excopy, *walk;
106

107 108
	excopy = kmemdup(ex, sizeof(*ex), GFP_KERNEL);
	if (!excopy)
109 110
		return -ENOMEM;

111 112
	list_for_each_entry(walk, &dev_cgroup->exceptions, list) {
		if (walk->type != ex->type)
113
			continue;
114
		if (walk->major != ex->major)
115
			continue;
116
		if (walk->minor != ex->minor)
117 118
			continue;

119 120 121
		walk->access |= ex->access;
		kfree(excopy);
		excopy = NULL;
122 123
	}

124 125
	if (excopy != NULL)
		list_add_tail_rcu(&excopy->list, &dev_cgroup->exceptions);
126 127 128 129
	return 0;
}

/*
130
 * called under devcgroup_mutex
131
 */
132 133
static void dev_exception_rm(struct dev_cgroup *dev_cgroup,
			     struct dev_exception_item *ex)
134
{
135
	struct dev_exception_item *walk, *tmp;
136

137 138
	list_for_each_entry_safe(walk, tmp, &dev_cgroup->exceptions, list) {
		if (walk->type != ex->type)
139
			continue;
140
		if (walk->major != ex->major)
141
			continue;
142
		if (walk->minor != ex->minor)
143 144
			continue;

145
		walk->access &= ~ex->access;
146
		if (!walk->access) {
147
			list_del_rcu(&walk->list);
148
			kfree_rcu(walk, rcu);
149 150 151 152
		}
	}
}

153
/**
154 155
 * dev_exception_clean - frees all entries of the exception list
 * @dev_cgroup: dev_cgroup with the exception list to be cleaned
156 157 158
 *
 * called under devcgroup_mutex
 */
159
static void dev_exception_clean(struct dev_cgroup *dev_cgroup)
160
{
161
	struct dev_exception_item *ex, *tmp;
162

163 164 165
	list_for_each_entry_safe(ex, tmp, &dev_cgroup->exceptions, list) {
		list_del(&ex->list);
		kfree(ex);
166 167 168
	}
}

169 170 171
/*
 * called from kernel/cgroup.c with cgroup_lock() held.
 */
172
static struct cgroup_subsys_state *devcgroup_create(struct cgroup *cgroup)
173 174 175 176 177 178 179 180
{
	struct dev_cgroup *dev_cgroup, *parent_dev_cgroup;
	struct cgroup *parent_cgroup;
	int ret;

	dev_cgroup = kzalloc(sizeof(*dev_cgroup), GFP_KERNEL);
	if (!dev_cgroup)
		return ERR_PTR(-ENOMEM);
181
	INIT_LIST_HEAD(&dev_cgroup->exceptions);
182 183
	parent_cgroup = cgroup->parent;

184
	if (parent_cgroup == NULL)
185
		dev_cgroup->deny_all = false;
186
	else {
187
		parent_dev_cgroup = cgroup_to_devcgroup(parent_cgroup);
188
		mutex_lock(&devcgroup_mutex);
189 190
		ret = dev_exceptions_copy(&dev_cgroup->exceptions,
					  &parent_dev_cgroup->exceptions);
191
		dev_cgroup->deny_all = parent_dev_cgroup->deny_all;
192
		mutex_unlock(&devcgroup_mutex);
193 194 195 196 197 198 199 200 201
		if (ret) {
			kfree(dev_cgroup);
			return ERR_PTR(ret);
		}
	}

	return &dev_cgroup->css;
}

202
static void devcgroup_destroy(struct cgroup *cgroup)
203 204 205 206
{
	struct dev_cgroup *dev_cgroup;

	dev_cgroup = cgroup_to_devcgroup(cgroup);
207
	dev_exception_clean(dev_cgroup);
208 209 210 211 212
	kfree(dev_cgroup);
}

#define DEVCG_ALLOW 1
#define DEVCG_DENY 2
213 214
#define DEVCG_LIST 3

215
#define MAJMINLEN 13
216
#define ACCLEN 4
217 218 219 220

static void set_access(char *acc, short access)
{
	int idx = 0;
221
	memset(acc, 0, ACCLEN);
222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240
	if (access & ACC_READ)
		acc[idx++] = 'r';
	if (access & ACC_WRITE)
		acc[idx++] = 'w';
	if (access & ACC_MKNOD)
		acc[idx++] = 'm';
}

static char type_to_char(short type)
{
	if (type == DEV_ALL)
		return 'a';
	if (type == DEV_CHAR)
		return 'c';
	if (type == DEV_BLOCK)
		return 'b';
	return 'X';
}

241
static void set_majmin(char *str, unsigned m)
242 243
{
	if (m == ~0)
Li Zefan's avatar
Li Zefan committed
244
		strcpy(str, "*");
245
	else
Li Zefan's avatar
Li Zefan committed
246
		sprintf(str, "%u", m);
247 248
}

249 250
static int devcgroup_seq_read(struct cgroup *cgroup, struct cftype *cft,
				struct seq_file *m)
251
{
252
	struct dev_cgroup *devcgroup = cgroup_to_devcgroup(cgroup);
253
	struct dev_exception_item *ex;
254
	char maj[MAJMINLEN], min[MAJMINLEN], acc[ACCLEN];
255

256
	rcu_read_lock();
257 258 259 260 261 262 263 264 265 266 267
	/*
	 * To preserve the compatibility:
	 * - Only show the "all devices" when the default policy is to allow
	 * - List the exceptions in case the default policy is to deny
	 * This way, the file remains as a "whitelist of devices"
	 */
	if (devcgroup->deny_all == false) {
		set_access(acc, ACC_MASK);
		set_majmin(maj, ~0);
		set_majmin(min, ~0);
		seq_printf(m, "%c %s:%s %s\n", type_to_char(DEV_ALL),
268
			   maj, min, acc);
269
	} else {
270 271 272 273 274
		list_for_each_entry_rcu(ex, &devcgroup->exceptions, list) {
			set_access(acc, ex->access);
			set_majmin(maj, ex->major);
			set_majmin(min, ex->minor);
			seq_printf(m, "%c %s:%s %s\n", type_to_char(ex->type),
275 276
				   maj, min, acc);
		}
277
	}
278
	rcu_read_unlock();
279

280
	return 0;
281 282
}

283
/**
284 285 286 287 288
 * may_access - verifies if a new exception is part of what is allowed
 *		by a dev cgroup based on the default policy +
 *		exceptions. This is used to make sure a child cgroup
 *		won't have more privileges than its parent or to
 *		verify if a certain access is allowed.
289
 * @dev_cgroup: dev cgroup to be tested against
290
 * @refex: new exception
291
 */
292 293
static int may_access(struct dev_cgroup *dev_cgroup,
		      struct dev_exception_item *refex)
294
{
295
	struct dev_exception_item *ex;
296
	bool match = false;
297

298 299
	list_for_each_entry(ex, &dev_cgroup->exceptions, list) {
		if ((refex->type & DEV_BLOCK) && !(ex->type & DEV_BLOCK))
300
			continue;
301
		if ((refex->type & DEV_CHAR) && !(ex->type & DEV_CHAR))
302
			continue;
303
		if (ex->major != ~0 && ex->major != refex->major)
304
			continue;
305
		if (ex->minor != ~0 && ex->minor != refex->minor)
306
			continue;
307
		if (refex->access & (~ex->access))
308
			continue;
309 310
		match = true;
		break;
311
	}
312 313

	/*
314
	 * In two cases we'll consider this new exception valid:
315
	 * - the dev cgroup has its default policy to allow + exception list:
316
	 *   the new exception should *not* match any of the exceptions
317 318
	 *   (!deny_all, !match)
	 * - the dev cgroup has its default policy to deny + exception list:
319
	 *   the new exception *should* match the exceptions
320 321 322 323
	 *   (deny_all, match)
	 */
	if (dev_cgroup->deny_all == match)
		return 1;
324 325 326 327 328
	return 0;
}

/*
 * parent_has_perm:
329
 * when adding a new allow rule to a device exception list, the rule
330 331
 * must be allowed in the parent device
 */
332
static int parent_has_perm(struct dev_cgroup *childcg,
333
				  struct dev_exception_item *ex)
334
{
335
	struct cgroup *pcg = childcg->css.cgroup->parent;
336 337 338 339 340
	struct dev_cgroup *parent;

	if (!pcg)
		return 1;
	parent = cgroup_to_devcgroup(pcg);
341
	return may_access(parent, ex);
342 343 344
}

/*
345
 * Modify the exception list using allow/deny rules.
346 347
 * CAP_SYS_ADMIN is needed for this.  It's at least separate from CAP_MKNOD
 * so we can give a container CAP_MKNOD to let it create devices but not
348
 * modify the exception list.
349 350
 * It seems likely we'll want to add a CAP_CONTAINER capability to allow
 * us to also grant CAP_SYS_ADMIN to containers without giving away the
351
 * device exception list controls, but for now we'll stick with CAP_SYS_ADMIN
352 353 354 355 356
 *
 * Taking rules away is always allowed (given CAP_SYS_ADMIN).  Granting
 * new access is only allowed if you're in the top-level cgroup, or your
 * parent cgroup has the access you're asking for.
 */
357 358
static int devcgroup_update_access(struct dev_cgroup *devcgroup,
				   int filetype, const char *buffer)
359
{
360
	const char *b;
Li Zefan's avatar
Li Zefan committed
361
	char *endp;
Li Zefan's avatar
Li Zefan committed
362
	int count;
363
	struct dev_exception_item ex;
364 365 366 367

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

368
	memset(&ex, 0, sizeof(ex));
369 370 371 372
	b = buffer;

	switch (*b) {
	case 'a':
373 374
		switch (filetype) {
		case DEVCG_ALLOW:
375
			if (!parent_has_perm(devcgroup, &ex))
376
				return -EPERM;
377
			dev_exception_clean(devcgroup);
378 379 380
			devcgroup->deny_all = false;
			break;
		case DEVCG_DENY:
381
			dev_exception_clean(devcgroup);
382 383 384 385 386 387
			devcgroup->deny_all = true;
			break;
		default:
			return -EINVAL;
		}
		return 0;
388
	case 'b':
389
		ex.type = DEV_BLOCK;
390 391
		break;
	case 'c':
392
		ex.type = DEV_CHAR;
393 394
		break;
	default:
395
		return -EINVAL;
396 397
	}
	b++;
398 399
	if (!isspace(*b))
		return -EINVAL;
400 401
	b++;
	if (*b == '*') {
402
		ex.major = ~0;
403 404
		b++;
	} else if (isdigit(*b)) {
405
		ex.major = simple_strtoul(b, &endp, 10);
Li Zefan's avatar
Li Zefan committed
406
		b = endp;
407
	} else {
408
		return -EINVAL;
409
	}
410 411
	if (*b != ':')
		return -EINVAL;
412 413 414 415
	b++;

	/* read minor */
	if (*b == '*') {
416
		ex.minor = ~0;
417 418
		b++;
	} else if (isdigit(*b)) {
419
		ex.minor = simple_strtoul(b, &endp, 10);
Li Zefan's avatar
Li Zefan committed
420
		b = endp;
421
	} else {
422
		return -EINVAL;
423
	}
424 425
	if (!isspace(*b))
		return -EINVAL;
426 427 428
	for (b++, count = 0; count < 3; count++, b++) {
		switch (*b) {
		case 'r':
429
			ex.access |= ACC_READ;
430 431
			break;
		case 'w':
432
			ex.access |= ACC_WRITE;
433 434
			break;
		case 'm':
435
			ex.access |= ACC_MKNOD;
436 437 438 439 440 441
			break;
		case '\n':
		case '\0':
			count = 3;
			break;
		default:
442
			return -EINVAL;
443 444 445 446 447
		}
	}

	switch (filetype) {
	case DEVCG_ALLOW:
448
		if (!parent_has_perm(devcgroup, &ex))
449
			return -EPERM;
450 451 452 453 454 455
		/*
		 * If the default policy is to allow by default, try to remove
		 * an matching exception instead. And be silent about it: we
		 * don't want to break compatibility
		 */
		if (devcgroup->deny_all == false) {
456
			dev_exception_rm(devcgroup, &ex);
457 458
			return 0;
		}
459
		return dev_exception_add(devcgroup, &ex);
460
	case DEVCG_DENY:
461 462 463 464 465 466
		/*
		 * If the default policy is to deny by default, try to remove
		 * an matching exception instead. And be silent about it: we
		 * don't want to break compatibility
		 */
		if (devcgroup->deny_all == true) {
467
			dev_exception_rm(devcgroup, &ex);
468 469
			return 0;
		}
470
		return dev_exception_add(devcgroup, &ex);
471
	default:
472
		return -EINVAL;
473
	}
474 475
	return 0;
}
476

477 478 479 480
static int devcgroup_access_write(struct cgroup *cgrp, struct cftype *cft,
				  const char *buffer)
{
	int retval;
481 482

	mutex_lock(&devcgroup_mutex);
483 484
	retval = devcgroup_update_access(cgroup_to_devcgroup(cgrp),
					 cft->private, buffer);
485
	mutex_unlock(&devcgroup_mutex);
486 487 488 489 490 491
	return retval;
}

static struct cftype dev_cgroup_files[] = {
	{
		.name = "allow",
492
		.write_string  = devcgroup_access_write,
493 494 495 496
		.private = DEVCG_ALLOW,
	},
	{
		.name = "deny",
497
		.write_string = devcgroup_access_write,
498 499
		.private = DEVCG_DENY,
	},
500 501 502 503 504
	{
		.name = "list",
		.read_seq_string = devcgroup_seq_read,
		.private = DEVCG_LIST,
	},
505
	{ }	/* terminate */
506 507 508 509 510 511
};

struct cgroup_subsys devices_subsys = {
	.name = "devices",
	.can_attach = devcgroup_can_attach,
	.create = devcgroup_create,
512
	.destroy = devcgroup_destroy,
513
	.subsys_id = devices_subsys_id,
514
	.base_cftypes = dev_cgroup_files,
515 516 517 518 519 520 521 522 523

	/*
	 * While devices cgroup has the rudimentary hierarchy support which
	 * checks the parent's restriction, it doesn't properly propagates
	 * config changes in ancestors to their descendents.  A child
	 * should only be allowed to add more restrictions to the parent's
	 * configuration.  Fix it and remove the following.
	 */
	.broken_hierarchy = true,
524 525
};

526 527 528 529 530 531 532 533 534 535
/**
 * __devcgroup_check_permission - checks if an inode operation is permitted
 * @dev_cgroup: the dev cgroup to be tested against
 * @type: device type
 * @major: device major number
 * @minor: device minor number
 * @access: combination of ACC_WRITE, ACC_READ and ACC_MKNOD
 *
 * returns 0 on success, -EPERM case the operation is not permitted
 */
536
static int __devcgroup_check_permission(short type, u32 major, u32 minor,
537
				        short access)
538
{
539
	struct dev_cgroup *dev_cgroup;
540
	struct dev_exception_item ex;
541
	int rc;
542

543 544 545 546 547
	memset(&ex, 0, sizeof(ex));
	ex.type = type;
	ex.major = major;
	ex.minor = minor;
	ex.access = access;
548

549
	rcu_read_lock();
550
	dev_cgroup = task_devcgroup(current);
551
	rc = may_access(dev_cgroup, &ex);
552
	rcu_read_unlock();
553

554 555
	if (!rc)
		return -EPERM;
556

557 558
	return 0;
}
559

560 561 562 563 564 565 566 567 568 569 570 571 572
int __devcgroup_inode_permission(struct inode *inode, int mask)
{
	short type, access = 0;

	if (S_ISBLK(inode->i_mode))
		type = DEV_BLOCK;
	if (S_ISCHR(inode->i_mode))
		type = DEV_CHAR;
	if (mask & MAY_WRITE)
		access |= ACC_WRITE;
	if (mask & MAY_READ)
		access |= ACC_READ;

573 574
	return __devcgroup_check_permission(type, imajor(inode), iminor(inode),
			access);
575 576 577 578
}

int devcgroup_inode_mknod(int mode, dev_t dev)
{
579
	short type;
580

581 582 583
	if (!S_ISBLK(mode) && !S_ISCHR(mode))
		return 0;

584 585 586 587
	if (S_ISBLK(mode))
		type = DEV_BLOCK;
	else
		type = DEV_CHAR;
588

589 590
	return __devcgroup_check_permission(type, MAJOR(dev), MINOR(dev),
			ACC_MKNOD);
591

592
}