commoncap.c 31.6 KB
Newer Older
1
/* Common capabilities, needed by capability.o.
Linus Torvalds's avatar
Linus Torvalds committed
2 3 4 5 6 7 8 9
 *
 *	This program is free software; you can redistribute it and/or modify
 *	it under the terms of the GNU General Public License as published by
 *	the Free Software Foundation; either version 2 of the License, or
 *	(at your option) any later version.
 *
 */

10
#include <linux/capability.h>
11
#include <linux/audit.h>
Linus Torvalds's avatar
Linus Torvalds committed
12 13 14
#include <linux/module.h>
#include <linux/init.h>
#include <linux/kernel.h>
15
#include <linux/lsm_hooks.h>
Linus Torvalds's avatar
Linus Torvalds committed
16 17 18 19 20 21 22 23 24 25
#include <linux/file.h>
#include <linux/mm.h>
#include <linux/mman.h>
#include <linux/pagemap.h>
#include <linux/swap.h>
#include <linux/skbuff.h>
#include <linux/netlink.h>
#include <linux/ptrace.h>
#include <linux/xattr.h>
#include <linux/hugetlb.h>
26
#include <linux/mount.h>
27
#include <linux/sched.h>
28 29
#include <linux/prctl.h>
#include <linux/securebits.h>
30
#include <linux/user_namespace.h>
Al Viro's avatar
Al Viro committed
31
#include <linux/binfmts.h>
32
#include <linux/personality.h>
33

34 35 36 37 38 39 40 41 42 43 44
/*
 * If a non-root user executes a setuid-root binary in
 * !secure(SECURE_NOROOT) mode, then we raise capabilities.
 * However if fE is also set, then the intent is for only
 * the file capabilities to be applied, and the setuid-root
 * bit is left on either to change the uid (plausible) or
 * to get full privilege on a kernel without file capabilities
 * support.  So in that case we do not raise capabilities.
 *
 * Warn if that happens, once per boot.
 */
45
static void warn_setuid_and_fcaps_mixed(const char *fname)
46 47 48 49 50 51 52 53 54 55
{
	static int warned;
	if (!warned) {
		printk(KERN_INFO "warning: `%s' has both setuid-root and"
			" effective capabilities. Therefore not raising all"
			" capabilities.\n", fname);
		warned = 1;
	}
}

56 57
/**
 * cap_capable - Determine whether a task has a particular effective capability
58
 * @cred: The credentials to use
59
 * @ns:  The user namespace in which we need the capability
60 61 62 63 64 65
 * @cap: The capability to check for
 * @audit: Whether to write an audit message or not
 *
 * Determine whether the nominated task has the specified capability amongst
 * its effective set, returning 0 if it does, -ve if it does not.
 *
66 67 68 69
 * NOTE WELL: cap_has_capability() cannot be used like the kernel's capable()
 * and has_capability() functions.  That is, it has the reverse semantics:
 * cap_has_capability() returns 0 when a task has a capability, but the
 * kernel's capable() and has_capability() returns 1 for this case.
70
 */
71 72
int cap_capable(const struct cred *cred, struct user_namespace *targ_ns,
		int cap, int audit)
Linus Torvalds's avatar
Linus Torvalds committed
73
{
74
	struct user_namespace *ns = targ_ns;
75

76 77 78 79 80
	/* See if cred has the capability in the target user namespace
	 * by examining the target user namespace and all of the target
	 * user namespace's parents.
	 */
	for (;;) {
81
		/* Do we have the necessary capabilities? */
82
		if (ns == cred->user_ns)
83 84 85
			return cap_raised(cred->cap_effective, cap) ? 0 : -EPERM;

		/* Have we tried all of the parent namespaces? */
86
		if (ns == &init_user_ns)
87 88
			return -EPERM;

89 90 91 92 93 94 95
		/* 
		 * The owner of the user namespace in the parent of the
		 * user namespace has all caps.
		 */
		if ((ns->parent == cred->user_ns) && uid_eq(ns->owner, cred->euid))
			return 0;

96
		/*
97
		 * If you have a capability in a parent user ns, then you have
98 99
		 * it over all children user namespaces as well.
		 */
100
		ns = ns->parent;
101 102 103
	}

	/* We never get here */
Linus Torvalds's avatar
Linus Torvalds committed
104 105
}

106 107 108 109 110 111 112 113
/**
 * cap_settime - Determine whether the current process may set the system clock
 * @ts: The time to set
 * @tz: The timezone to set
 *
 * Determine whether the current process may set the system clock and timezone
 * information, returning 0 if permission granted, -ve if denied.
 */
114
int cap_settime(const struct timespec64 *ts, const struct timezone *tz)
Linus Torvalds's avatar
Linus Torvalds committed
115 116 117 118 119 120
{
	if (!capable(CAP_SYS_TIME))
		return -EPERM;
	return 0;
}

121
/**
122
 * cap_ptrace_access_check - Determine whether the current process may access
123 124 125 126
 *			   another
 * @child: The process to be accessed
 * @mode: The mode of attachment.
 *
127 128 129 130 131 132
 * If we are in the same or an ancestor user_ns and have all the target
 * task's capabilities, then ptrace access is allowed.
 * If we have the ptrace capability to the target user_ns, then ptrace
 * access is allowed.
 * Else denied.
 *
133 134 135
 * Determine whether a process may access another, returning 0 if permission
 * granted, -ve if denied.
 */
136
int cap_ptrace_access_check(struct task_struct *child, unsigned int mode)
Linus Torvalds's avatar
Linus Torvalds committed
137
{
138
	int ret = 0;
139
	const struct cred *cred, *child_cred;
140
	const kernel_cap_t *caller_caps;
141 142

	rcu_read_lock();
143 144
	cred = current_cred();
	child_cred = __task_cred(child);
145 146 147 148
	if (mode & PTRACE_MODE_FSCREDS)
		caller_caps = &cred->cap_effective;
	else
		caller_caps = &cred->cap_permitted;
149
	if (cred->user_ns == child_cred->user_ns &&
150
	    cap_issubset(child_cred->cap_permitted, *caller_caps))
151
		goto out;
152
	if (ns_capable(child_cred->user_ns, CAP_SYS_PTRACE))
153 154 155
		goto out;
	ret = -EPERM;
out:
156 157
	rcu_read_unlock();
	return ret;
158 159
}

160 161 162 163
/**
 * cap_ptrace_traceme - Determine whether another process may trace the current
 * @parent: The task proposed to be the tracer
 *
164 165 166 167 168 169
 * If parent is in the same or an ancestor user_ns and has all current's
 * capabilities, then ptrace access is allowed.
 * If parent has the ptrace capability to current's user_ns, then ptrace
 * access is allowed.
 * Else denied.
 *
170 171 172
 * Determine whether the nominated task is permitted to trace the current
 * process, returning 0 if permission is granted, -ve if denied.
 */
173 174
int cap_ptrace_traceme(struct task_struct *parent)
{
175
	int ret = 0;
176
	const struct cred *cred, *child_cred;
177 178

	rcu_read_lock();
179 180
	cred = __task_cred(parent);
	child_cred = current_cred();
181
	if (cred->user_ns == child_cred->user_ns &&
182 183
	    cap_issubset(child_cred->cap_permitted, cred->cap_permitted))
		goto out;
184
	if (has_ns_capability(parent, child_cred->user_ns, CAP_SYS_PTRACE))
185 186 187
		goto out;
	ret = -EPERM;
out:
188 189
	rcu_read_unlock();
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
190 191
}

192 193 194 195 196 197 198 199 200 201 202 203
/**
 * cap_capget - Retrieve a task's capability sets
 * @target: The task from which to retrieve the capability sets
 * @effective: The place to record the effective set
 * @inheritable: The place to record the inheritable set
 * @permitted: The place to record the permitted set
 *
 * This function retrieves the capabilities of the nominated task and returns
 * them to the caller.
 */
int cap_capget(struct task_struct *target, kernel_cap_t *effective,
	       kernel_cap_t *inheritable, kernel_cap_t *permitted)
Linus Torvalds's avatar
Linus Torvalds committed
204
{
205
	const struct cred *cred;
206

Linus Torvalds's avatar
Linus Torvalds committed
207
	/* Derived from kernel/capability.c:sys_capget. */
208 209
	rcu_read_lock();
	cred = __task_cred(target);
210 211 212
	*effective   = cred->cap_effective;
	*inheritable = cred->cap_inheritable;
	*permitted   = cred->cap_permitted;
213
	rcu_read_unlock();
Linus Torvalds's avatar
Linus Torvalds committed
214 215 216
	return 0;
}

217 218 219 220
/*
 * Determine whether the inheritable capabilities are limited to the old
 * permitted set.  Returns 1 if they are limited, 0 if they are not.
 */
221 222 223
static inline int cap_inh_is_capped(void)
{

224 225 226
	/* they are so limited unless the current task has the CAP_SETPCAP
	 * capability
	 */
227
	if (cap_capable(current_cred(), current_cred()->user_ns,
228
			CAP_SETPCAP, SECURITY_CAP_AUDIT) == 0)
229 230
		return 0;
	return 1;
231
}
232

233 234 235 236 237 238 239 240 241 242 243 244
/**
 * cap_capset - Validate and apply proposed changes to current's capabilities
 * @new: The proposed new credentials; alterations should be made here
 * @old: The current task's current credentials
 * @effective: A pointer to the proposed new effective capabilities set
 * @inheritable: A pointer to the proposed new inheritable capabilities set
 * @permitted: A pointer to the proposed new permitted capabilities set
 *
 * This function validates and applies a proposed mass change to the current
 * process's capability sets.  The changes are made to the proposed new
 * credentials, and assuming no error, will be committed by the caller of LSM.
 */
245 246 247 248 249
int cap_capset(struct cred *new,
	       const struct cred *old,
	       const kernel_cap_t *effective,
	       const kernel_cap_t *inheritable,
	       const kernel_cap_t *permitted)
Linus Torvalds's avatar
Linus Torvalds committed
250
{
251 252 253 254
	if (cap_inh_is_capped() &&
	    !cap_issubset(*inheritable,
			  cap_combine(old->cap_inheritable,
				      old->cap_permitted)))
255
		/* incapable of using this inheritable set */
Linus Torvalds's avatar
Linus Torvalds committed
256
		return -EPERM;
257

258
	if (!cap_issubset(*inheritable,
259 260
			  cap_combine(old->cap_inheritable,
				      old->cap_bset)))
261 262
		/* no new pI capabilities outside bounding set */
		return -EPERM;
Linus Torvalds's avatar
Linus Torvalds committed
263 264

	/* verify restrictions on target's new Permitted set */
265
	if (!cap_issubset(*permitted, old->cap_permitted))
Linus Torvalds's avatar
Linus Torvalds committed
266 267 268
		return -EPERM;

	/* verify the _new_Effective_ is a subset of the _new_Permitted_ */
269
	if (!cap_issubset(*effective, *permitted))
Linus Torvalds's avatar
Linus Torvalds committed
270 271
		return -EPERM;

272 273 274
	new->cap_effective   = *effective;
	new->cap_inheritable = *inheritable;
	new->cap_permitted   = *permitted;
275 276 277 278 279 280 281 282 283 284

	/*
	 * Mask off ambient bits that are no longer both permitted and
	 * inheritable.
	 */
	new->cap_ambient = cap_intersect(new->cap_ambient,
					 cap_intersect(*permitted,
						       *inheritable));
	if (WARN_ON(!cap_ambient_invariant_ok(new)))
		return -EINVAL;
Linus Torvalds's avatar
Linus Torvalds committed
285 286 287
	return 0;
}

288 289 290
/*
 * Clear proposed capability sets for execve().
 */
291 292
static inline void bprm_clear_caps(struct linux_binprm *bprm)
{
293
	cap_clear(bprm->cred->cap_permitted);
294 295 296
	bprm->cap_effective = false;
}

297 298 299 300 301 302 303 304 305 306 307
/**
 * cap_inode_need_killpriv - Determine if inode change affects privileges
 * @dentry: The inode/dentry in being changed with change marked ATTR_KILL_PRIV
 *
 * Determine if an inode having a change applied that's marked ATTR_KILL_PRIV
 * affects the security markings on that inode, and if it is, should
 * inode_killpriv() be invoked or the change rejected?
 *
 * Returns 0 if granted; +ve if granted, but inode_killpriv() is required; and
 * -ve to deny the change.
 */
308 309
int cap_inode_need_killpriv(struct dentry *dentry)
{
310
	struct inode *inode = d_backing_inode(dentry);
311 312
	int error;

Al Viro's avatar
Al Viro committed
313
	if (!inode->i_op->getxattr)
314 315
	       return 0;

316
	error = inode->i_op->getxattr(dentry, inode, XATTR_NAME_CAPS, NULL, 0);
317 318 319 320 321
	if (error <= 0)
		return 0;
	return 1;
}

322 323 324 325 326 327 328 329
/**
 * cap_inode_killpriv - Erase the security markings on an inode
 * @dentry: The inode/dentry to alter
 *
 * Erase the privilege-enhancing security markings on an inode.
 *
 * Returns 0 if successful, -ve on error.
 */
330 331
int cap_inode_killpriv(struct dentry *dentry)
{
332
	struct inode *inode = d_backing_inode(dentry);
333

Al Viro's avatar
Al Viro committed
334
	if (!inode->i_op->removexattr)
335 336 337 338 339
	       return 0;

	return inode->i_op->removexattr(dentry, XATTR_NAME_CAPS);
}

340 341 342 343
/*
 * Calculate the new process capability sets from the capability sets attached
 * to a file.
 */
344
static inline int bprm_caps_from_vfs_caps(struct cpu_vfs_cap_data *caps,
345
					  struct linux_binprm *bprm,
346 347
					  bool *effective,
					  bool *has_cap)
348
{
349
	struct cred *new = bprm->cred;
350 351 352 353
	unsigned i;
	int ret = 0;

	if (caps->magic_etc & VFS_CAP_FLAGS_EFFECTIVE)
354
		*effective = true;
355

356 357 358
	if (caps->magic_etc & VFS_CAP_REVISION_MASK)
		*has_cap = true;

359 360 361 362 363 364
	CAP_FOR_EACH_U32(i) {
		__u32 permitted = caps->permitted.cap[i];
		__u32 inheritable = caps->inheritable.cap[i];

		/*
		 * pP' = (X & fP) | (pI & fI)
365
		 * The addition of pA' is handled later.
366
		 */
367 368 369
		new->cap_permitted.cap[i] =
			(new->cap_bset.cap[i] & permitted) |
			(new->cap_inheritable.cap[i] & inheritable);
370

371 372
		if (permitted & ~new->cap_permitted.cap[i])
			/* insufficient to execute correctly */
373 374 375 376 377 378 379 380
			ret = -EPERM;
	}

	/*
	 * For legacy apps, with no internal support for recognizing they
	 * do not have enough capabilities, we return an error if they are
	 * missing some "forced" (aka file-permitted) capabilities.
	 */
381
	return *effective ? ret : 0;
382 383
}

384 385 386
/*
 * Extract the on-exec-apply capability sets for an executable file.
 */
387 388
int get_vfs_caps_from_disk(const struct dentry *dentry, struct cpu_vfs_cap_data *cpu_caps)
{
389
	struct inode *inode = d_backing_inode(dentry);
390
	__u32 magic_etc;
391
	unsigned tocopy, i;
392 393 394 395 396
	int size;
	struct vfs_cap_data caps;

	memset(cpu_caps, 0, sizeof(struct cpu_vfs_cap_data));

Al Viro's avatar
Al Viro committed
397
	if (!inode || !inode->i_op->getxattr)
398 399
		return -ENODATA;

400 401
	size = inode->i_op->getxattr((struct dentry *)dentry, inode,
				     XATTR_NAME_CAPS, &caps, XATTR_CAPS_SZ);
402
	if (size == -ENODATA || size == -EOPNOTSUPP)
403 404 405 406
		/* no data, that's ok */
		return -ENODATA;
	if (size < 0)
		return size;
407

408
	if (size < sizeof(magic_etc))
409 410
		return -EINVAL;

411
	cpu_caps->magic_etc = magic_etc = le32_to_cpu(caps.magic_etc);
412

413
	switch (magic_etc & VFS_CAP_REVISION_MASK) {
414 415 416 417 418 419 420 421 422 423
	case VFS_CAP_REVISION_1:
		if (size != XATTR_CAPS_SZ_1)
			return -EINVAL;
		tocopy = VFS_CAP_U32_1;
		break;
	case VFS_CAP_REVISION_2:
		if (size != XATTR_CAPS_SZ_2)
			return -EINVAL;
		tocopy = VFS_CAP_U32_2;
		break;
424 425 426
	default:
		return -EINVAL;
	}
427

428
	CAP_FOR_EACH_U32(i) {
429 430 431 432
		if (i >= tocopy)
			break;
		cpu_caps->permitted.cap[i] = le32_to_cpu(caps.data[i].permitted);
		cpu_caps->inheritable.cap[i] = le32_to_cpu(caps.data[i].inheritable);
433
	}
434

435 436 437
	cpu_caps->permitted.cap[CAP_LAST_U32] &= CAP_LAST_U32_VALID_MASK;
	cpu_caps->inheritable.cap[CAP_LAST_U32] &= CAP_LAST_U32_VALID_MASK;

438
	return 0;
439 440
}

441 442 443 444 445
/*
 * Attempt to get the on-exec apply capability sets for an executable file from
 * its xattrs and, if present, apply them to the proposed credentials being
 * constructed by execve().
 */
446
static int get_file_caps(struct linux_binprm *bprm, bool *effective, bool *has_cap)
447 448
{
	int rc = 0;
449
	struct cpu_vfs_cap_data vcaps;
450

451 452
	bprm_clear_caps(bprm);

453 454 455
	if (!file_caps_enabled)
		return 0;

456
	if (!mnt_may_suid(bprm->file->f_path.mnt))
457
		return 0;
458 459 460 461 462 463

	/*
	 * This check is redundant with mnt_may_suid() but is kept to make
	 * explicit that capability bits are limited to s_user_ns and its
	 * descendants.
	 */
464 465
	if (!current_in_userns(bprm->file->f_path.mnt->mnt_sb->s_user_ns))
		return 0;
466

467
	rc = get_vfs_caps_from_disk(bprm->file->f_path.dentry, &vcaps);
468 469 470 471 472 473
	if (rc < 0) {
		if (rc == -EINVAL)
			printk(KERN_NOTICE "%s: get_vfs_caps_from_disk returned %d for %s\n",
				__func__, rc, bprm->filename);
		else if (rc == -ENODATA)
			rc = 0;
474 475 476
		goto out;
	}

477
	rc = bprm_caps_from_vfs_caps(&vcaps, bprm, effective, has_cap);
478 479 480
	if (rc == -EINVAL)
		printk(KERN_NOTICE "%s: cap_from_disk returned %d for %s\n",
		       __func__, rc, bprm->filename);
481 482 483 484 485 486 487 488

out:
	if (rc)
		bprm_clear_caps(bprm);

	return rc;
}

489 490 491 492 493 494 495
/**
 * cap_bprm_set_creds - Set up the proposed credentials for execve().
 * @bprm: The execution parameters, including the proposed creds
 *
 * Set up the proposed credentials for a new execution context being
 * constructed by execve().  The proposed creds in @bprm->cred is altered,
 * which won't take effect immediately.  Returns 0 if successful, -ve on error.
496 497
 */
int cap_bprm_set_creds(struct linux_binprm *bprm)
Linus Torvalds's avatar
Linus Torvalds committed
498
{
499 500
	const struct cred *old = current_cred();
	struct cred *new = bprm->cred;
501
	bool effective, has_cap = false, is_setid;
502
	int ret;
503
	kuid_t root_uid;
Linus Torvalds's avatar
Linus Torvalds committed
504

505 506 507
	if (WARN_ON(!cap_ambient_invariant_ok(old)))
		return -EPERM;

508
	effective = false;
509
	ret = get_file_caps(bprm, &effective, &has_cap);
510 511
	if (ret < 0)
		return ret;
Linus Torvalds's avatar
Linus Torvalds committed
512

513 514
	root_uid = make_kuid(new->user_ns, 0);

515
	if (!issecure(SECURE_NOROOT)) {
516 517 518 519 520
		/*
		 * If the legacy file capability is set, then don't set privs
		 * for a setuid root binary run by a non-root user.  Do set it
		 * for a root user just to cause least surprise to an admin.
		 */
521
		if (has_cap && !uid_eq(new->uid, root_uid) && uid_eq(new->euid, root_uid)) {
522 523 524
			warn_setuid_and_fcaps_mixed(bprm->filename);
			goto skip;
		}
525 526 527 528 529
		/*
		 * To support inheritance of root-permissions and suid-root
		 * executables under compatibility mode, we override the
		 * capability sets for the file.
		 *
530
		 * If only the real uid is 0, we do not set the effective bit.
531
		 */
532
		if (uid_eq(new->euid, root_uid) || uid_eq(new->uid, root_uid)) {
533
			/* pP' = (cap_bset & ~0) | (pI & ~0) */
534 535
			new->cap_permitted = cap_combine(old->cap_bset,
							 old->cap_inheritable);
Linus Torvalds's avatar
Linus Torvalds committed
536
		}
537
		if (uid_eq(new->euid, root_uid))
538
			effective = true;
Linus Torvalds's avatar
Linus Torvalds committed
539
	}
540
skip:
541

542 543 544 545 546
	/* if we have fs caps, clear dangerous personality flags */
	if (!cap_issubset(new->cap_permitted, old->cap_permitted))
		bprm->per_clear |= PER_CLEAR_ON_SETID;


547
	/* Don't let someone trace a set[ug]id/setpcap binary with the revised
548 549 550
	 * credentials unless they have the appropriate permit.
	 *
	 * In addition, if NO_NEW_PRIVS, then ensure we get no new privs.
551
	 */
552 553 554
	is_setid = !uid_eq(new->euid, old->uid) || !gid_eq(new->egid, old->gid);

	if ((is_setid ||
555 556 557
	     !cap_issubset(new->cap_permitted, old->cap_permitted)) &&
	    bprm->unsafe & ~LSM_UNSAFE_PTRACE_CAP) {
		/* downgrade; they get no more than they had, and maybe less */
558 559
		if (!capable(CAP_SETUID) ||
		    (bprm->unsafe & LSM_UNSAFE_NO_NEW_PRIVS)) {
560 561
			new->euid = new->uid;
			new->egid = new->gid;
Linus Torvalds's avatar
Linus Torvalds committed
562
		}
563 564
		new->cap_permitted = cap_intersect(new->cap_permitted,
						   old->cap_permitted);
Linus Torvalds's avatar
Linus Torvalds committed
565 566
	}

567 568
	new->suid = new->fsuid = new->euid;
	new->sgid = new->fsgid = new->egid;
Linus Torvalds's avatar
Linus Torvalds committed
569

570 571 572 573 574 575 576 577 578 579 580 581 582 583
	/* File caps or setid cancels ambient. */
	if (has_cap || is_setid)
		cap_clear(new->cap_ambient);

	/*
	 * Now that we've computed pA', update pP' to give:
	 *   pP' = (X & fP) | (pI & fI) | pA'
	 */
	new->cap_permitted = cap_combine(new->cap_permitted, new->cap_ambient);

	/*
	 * Set pE' = (fE ? pP' : pA').  Because pA' is zero if fE is set,
	 * this is the same as pE' = (fE ? pP' : 0) | pA'.
	 */
584 585 586
	if (effective)
		new->cap_effective = new->cap_permitted;
	else
587 588 589 590 591
		new->cap_effective = new->cap_ambient;

	if (WARN_ON(!cap_ambient_invariant_ok(new)))
		return -EPERM;

592
	bprm->cap_effective = effective;
Linus Torvalds's avatar
Linus Torvalds committed
593

594 595 596 597 598 599 600 601 602 603 604 605
	/*
	 * Audit candidate if current->cap_effective is set
	 *
	 * We do not bother to audit if 3 things are true:
	 *   1) cap_effective has all caps
	 *   2) we are root
	 *   3) root is supposed to have all caps (SECURE_NOROOT)
	 * Since this is just a normal root execing a process.
	 *
	 * Number 1 above might fail if you don't have a full bset, but I think
	 * that is interesting information to audit.
	 */
606
	if (!cap_issubset(new->cap_effective, new->cap_ambient)) {
607
		if (!cap_issubset(CAP_FULL_SET, new->cap_effective) ||
608
		    !uid_eq(new->euid, root_uid) || !uid_eq(new->uid, root_uid) ||
609 610 611 612 613
		    issecure(SECURE_NOROOT)) {
			ret = audit_log_bprm_fcaps(bprm, new, old);
			if (ret < 0)
				return ret;
		}
614
	}
Linus Torvalds's avatar
Linus Torvalds committed
615

616
	new->securebits &= ~issecure_mask(SECURE_KEEP_CAPS);
617 618 619 620

	if (WARN_ON(!cap_ambient_invariant_ok(new)))
		return -EPERM;

621
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
622 623
}

624 625 626 627 628 629 630 631 632
/**
 * cap_bprm_secureexec - Determine whether a secure execution is required
 * @bprm: The execution parameters
 *
 * Determine whether a secure execution is required, return 1 if it is, and 0
 * if it is not.
 *
 * The credentials have been committed by this point, and so are no longer
 * available through @bprm->cred.
633 634
 */
int cap_bprm_secureexec(struct linux_binprm *bprm)
Linus Torvalds's avatar
Linus Torvalds committed
635
{
636
	const struct cred *cred = current_cred();
637
	kuid_t root_uid = make_kuid(cred->user_ns, 0);
638

639
	if (!uid_eq(cred->uid, root_uid)) {
640 641
		if (bprm->cap_effective)
			return 1;
642
		if (!cap_issubset(cred->cap_permitted, cred->cap_ambient))
643 644 645
			return 1;
	}

646 647
	return (!uid_eq(cred->euid, cred->uid) ||
		!gid_eq(cred->egid, cred->gid));
Linus Torvalds's avatar
Linus Torvalds committed
648 649
}

650 651 652 653 654 655 656 657 658 659 660 661 662 663
/**
 * cap_inode_setxattr - Determine whether an xattr may be altered
 * @dentry: The inode/dentry being altered
 * @name: The name of the xattr to be changed
 * @value: The value that the xattr will be changed to
 * @size: The size of value
 * @flags: The replacement flag
 *
 * Determine whether an xattr may be altered or set on an inode, returning 0 if
 * permission is granted, -ve if denied.
 *
 * This is used to make sure security xattrs don't get updated or set by those
 * who aren't privileged to do so.
 */
664 665
int cap_inode_setxattr(struct dentry *dentry, const char *name,
		       const void *value, size_t size, int flags)
Linus Torvalds's avatar
Linus Torvalds committed
666
{
667 668 669 670
	if (!strcmp(name, XATTR_NAME_CAPS)) {
		if (!capable(CAP_SETFCAP))
			return -EPERM;
		return 0;
671 672 673
	}

	if (!strncmp(name, XATTR_SECURITY_PREFIX,
674
		     sizeof(XATTR_SECURITY_PREFIX) - 1) &&
Linus Torvalds's avatar
Linus Torvalds committed
675 676 677 678 679
	    !capable(CAP_SYS_ADMIN))
		return -EPERM;
	return 0;
}

680 681 682 683 684 685 686 687 688 689 690
/**
 * cap_inode_removexattr - Determine whether an xattr may be removed
 * @dentry: The inode/dentry being altered
 * @name: The name of the xattr to be changed
 *
 * Determine whether an xattr may be removed from an inode, returning 0 if
 * permission is granted, -ve if denied.
 *
 * This is used to make sure security xattrs don't get removed by those who
 * aren't privileged to remove them.
 */
691
int cap_inode_removexattr(struct dentry *dentry, const char *name)
Linus Torvalds's avatar
Linus Torvalds committed
692
{
693 694 695 696
	if (!strcmp(name, XATTR_NAME_CAPS)) {
		if (!capable(CAP_SETFCAP))
			return -EPERM;
		return 0;
697 698 699
	}

	if (!strncmp(name, XATTR_SECURITY_PREFIX,
700
		     sizeof(XATTR_SECURITY_PREFIX) - 1) &&
Linus Torvalds's avatar
Linus Torvalds committed
701 702 703 704 705
	    !capable(CAP_SYS_ADMIN))
		return -EPERM;
	return 0;
}

706
/*
Linus Torvalds's avatar
Linus Torvalds committed
707 708 709 710 711 712 713 714 715 716 717 718 719
 * cap_emulate_setxuid() fixes the effective / permitted capabilities of
 * a process after a call to setuid, setreuid, or setresuid.
 *
 *  1) When set*uiding _from_ one of {r,e,s}uid == 0 _to_ all of
 *  {r,e,s}uid != 0, the permitted and effective capabilities are
 *  cleared.
 *
 *  2) When set*uiding _from_ euid == 0 _to_ euid != 0, the effective
 *  capabilities of the process are cleared.
 *
 *  3) When set*uiding _from_ euid != 0 _to_ euid == 0, the effective
 *  capabilities are set to the permitted capabilities.
 *
720
 *  fsuid is handled elsewhere. fsuid == 0 and {r,e,s}uid!= 0 should
Linus Torvalds's avatar
Linus Torvalds committed
721 722
 *  never happen.
 *
723
 *  -astor
Linus Torvalds's avatar
Linus Torvalds committed
724 725 726 727 728 729 730 731 732 733 734
 *
 * cevans - New behaviour, Oct '99
 * A process may, via prctl(), elect to keep its capabilities when it
 * calls setuid() and switches away from uid==0. Both permitted and
 * effective sets will be retained.
 * Without this change, it was impossible for a daemon to drop only some
 * of its privilege. The call to setuid(!=0) would drop all privileges!
 * Keeping uid 0 is not an option because uid 0 owns too many vital
 * files..
 * Thanks to Olaf Kirch and Peter Benie for spotting this.
 */
735
static inline void cap_emulate_setxuid(struct cred *new, const struct cred *old)
Linus Torvalds's avatar
Linus Torvalds committed
736
{
737 738 739 740 741 742 743
	kuid_t root_uid = make_kuid(old->user_ns, 0);

	if ((uid_eq(old->uid, root_uid) ||
	     uid_eq(old->euid, root_uid) ||
	     uid_eq(old->suid, root_uid)) &&
	    (!uid_eq(new->uid, root_uid) &&
	     !uid_eq(new->euid, root_uid) &&
744 745 746 747 748 749 750 751 752 753 754 755
	     !uid_eq(new->suid, root_uid))) {
		if (!issecure(SECURE_KEEP_CAPS)) {
			cap_clear(new->cap_permitted);
			cap_clear(new->cap_effective);
		}

		/*
		 * Pre-ambient programs expect setresuid to nonroot followed
		 * by exec to drop capabilities.  We should make sure that
		 * this remains the case.
		 */
		cap_clear(new->cap_ambient);
Linus Torvalds's avatar
Linus Torvalds committed
756
	}
757
	if (uid_eq(old->euid, root_uid) && !uid_eq(new->euid, root_uid))
758
		cap_clear(new->cap_effective);
759
	if (!uid_eq(old->euid, root_uid) && uid_eq(new->euid, root_uid))
760
		new->cap_effective = new->cap_permitted;
Linus Torvalds's avatar
Linus Torvalds committed
761 762
}

763 764 765 766 767 768 769 770 771
/**
 * cap_task_fix_setuid - Fix up the results of setuid() call
 * @new: The proposed credentials
 * @old: The current task's current credentials
 * @flags: Indications of what has changed
 *
 * Fix up the results of setuid() call before the credential changes are
 * actually applied, returning 0 to grant the changes, -ve to deny them.
 */
772
int cap_task_fix_setuid(struct cred *new, const struct cred *old, int flags)
Linus Torvalds's avatar
Linus Torvalds committed
773 774 775 776 777
{
	switch (flags) {
	case LSM_SETID_RE:
	case LSM_SETID_ID:
	case LSM_SETID_RES:
778 779
		/* juggle the capabilities to follow [RES]UID changes unless
		 * otherwise suppressed */
780 781
		if (!issecure(SECURE_NO_SETUID_FIXUP))
			cap_emulate_setxuid(new, old);
Linus Torvalds's avatar
Linus Torvalds committed
782 783
		break;

784 785 786 787
	case LSM_SETID_FS:
		/* juggle the capabilties to follow FSUID changes, unless
		 * otherwise suppressed
		 *
788 789 790 791
		 * FIXME - is fsuser used for all CAP_FS_MASK capabilities?
		 *          if not, we might be a bit too harsh here.
		 */
		if (!issecure(SECURE_NO_SETUID_FIXUP)) {
792 793
			kuid_t root_uid = make_kuid(old->user_ns, 0);
			if (uid_eq(old->fsuid, root_uid) && !uid_eq(new->fsuid, root_uid))
794 795
				new->cap_effective =
					cap_drop_fs_set(new->cap_effective);
796

797
			if (!uid_eq(old->fsuid, root_uid) && uid_eq(new->fsuid, root_uid))
798 799 800
				new->cap_effective =
					cap_raise_fs_set(new->cap_effective,
							 new->cap_permitted);
Linus Torvalds's avatar
Linus Torvalds committed
801
		}
802
		break;
803

Linus Torvalds's avatar
Linus Torvalds committed
804 805 806 807 808 809 810
	default:
		return -EINVAL;
	}

	return 0;
}

811 812 813 814 815 816 817 818 819 820
/*
 * Rationale: code calling task_setscheduler, task_setioprio, and
 * task_setnice, assumes that
 *   . if capable(cap_sys_nice), then those actions should be allowed
 *   . if not capable(cap_sys_nice), but acting on your own processes,
 *   	then those actions should be allowed
 * This is insufficient now since you can call code without suid, but
 * yet with increased caps.
 * So we check for increased caps on the target process.
 */
821
static int cap_safe_nice(struct task_struct *p)
822
{
823
	int is_subset, ret = 0;
824 825 826 827

	rcu_read_lock();
	is_subset = cap_issubset(__task_cred(p)->cap_permitted,
				 current_cred()->cap_permitted);
828 829
	if (!is_subset && !ns_capable(__task_cred(p)->user_ns, CAP_SYS_NICE))
		ret = -EPERM;
830 831
	rcu_read_unlock();

832
	return ret;
833 834
}

835 836 837 838 839 840 841
/**
 * cap_task_setscheduler - Detemine if scheduler policy change is permitted
 * @p: The task to affect
 *
 * Detemine if the requested scheduler policy change is permitted for the
 * specified task, returning 0 if permission is granted, -ve if denied.
 */
842
int cap_task_setscheduler(struct task_struct *p)
843 844 845 846
{
	return cap_safe_nice(p);
}

847 848 849 850 851 852 853 854 855
/**
 * cap_task_ioprio - Detemine if I/O priority change is permitted
 * @p: The task to affect
 * @ioprio: The I/O priority to set
 *
 * Detemine if the requested I/O priority change is permitted for the specified
 * task, returning 0 if permission is granted, -ve if denied.
 */
int cap_task_setioprio(struct task_struct *p, int ioprio)
856 857 858 859
{
	return cap_safe_nice(p);
}

860 861 862 863 864 865 866 867 868
/**
 * cap_task_ioprio - Detemine if task priority change is permitted
 * @p: The task to affect
 * @nice: The nice value to set
 *
 * Detemine if the requested task priority change is permitted for the
 * specified task, returning 0 if permission is granted, -ve if denied.
 */
int cap_task_setnice(struct task_struct *p, int nice)
869 870 871 872
{
	return cap_safe_nice(p);
}

873
/*
874 875
 * Implement PR_CAPBSET_DROP.  Attempt to remove the specified capability from
 * the current task's bounding set.  Returns 0 on success, -ve on error.
876
 */
877
static int cap_prctl_drop(unsigned long cap)
878
{
879 880
	struct cred *new;

881
	if (!ns_capable(current_user_ns(), CAP_SETPCAP))
882 883 884
		return -EPERM;
	if (!cap_valid(cap))
		return -EINVAL;
885

886 887 888
	new = prepare_creds();
	if (!new)
		return -ENOMEM;
889
	cap_lower(new->cap_bset, cap);
890
	return commit_creds(new);
891
}
892

893 894 895 896 897 898 899 900 901 902 903 904
/**
 * cap_task_prctl - Implement process control functions for this security module
 * @option: The process control function requested
 * @arg2, @arg3, @arg4, @arg5: The argument data for this function
 *
 * Allow process control functions (sys_prctl()) to alter capabilities; may
 * also deny access to other functions not otherwise implemented here.
 *
 * Returns 0 or +ve on success, -ENOSYS if this function is not implemented
 * here, other -ve on error.  If -ENOSYS is returned, sys_prctl() and other LSM
 * modules will consider performing the function.
 */
905
int cap_task_prctl(int option, unsigned long arg2, unsigned long arg3,
906
		   unsigned long arg4, unsigned long arg5)
907
{
908
	const struct cred *old = current_cred();
909 910
	struct cred *new;

911 912 913
	switch (option) {
	case PR_CAPBSET_READ:
		if (!cap_valid(arg2))
914 915
			return -EINVAL;
		return !!cap_raised(old->cap_bset, arg2);
916

917
	case PR_CAPBSET_DROP:
918
		return cap_prctl_drop(arg2);
919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939

	/*
	 * The next four prctl's remain to assist with transitioning a
	 * system from legacy UID=0 based privilege (when filesystem
	 * capabilities are not in use) to a system using filesystem
	 * capabilities only - as the POSIX.1e draft intended.
	 *
	 * Note:
	 *
	 *  PR_SET_SECUREBITS =
	 *      issecure_mask(SECURE_KEEP_CAPS_LOCKED)
	 *    | issecure_mask(SECURE_NOROOT)
	 *    | issecure_mask(SECURE_NOROOT_LOCKED)
	 *    | issecure_mask(SECURE_NO_SETUID_FIXUP)
	 *    | issecure_mask(SECURE_NO_SETUID_FIXUP_LOCKED)
	 *
	 * will ensure that the current process and all of its
	 * children will be locked into a pure
	 * capability-based-privilege environment.
	 */
	case PR_SET_SECUREBITS:
940 941 942
		if ((((old->securebits & SECURE_ALL_LOCKS) >> 1)
		     & (old->securebits ^ arg2))			/*[1]*/
		    || ((old->securebits & SECURE_ALL_LOCKS & ~arg2))	/*[2]*/
943
		    || (arg2 & ~(SECURE_ALL_LOCKS | SECURE_ALL_BITS))	/*[3]*/
944
		    || (cap_capable(current_cred(),
945
				    current_cred()->user_ns, CAP_SETPCAP,
946
				    SECURITY_CAP_AUDIT) != 0)		/*[4]*/
947 948 949 950 951 952 953
			/*
			 * [1] no changing of bits that are locked
			 * [2] no unlocking of locks
			 * [3] no setting of unsupported bits
			 * [4] doing anything requires privilege (go read about
			 *     the "sendmail capabilities bug")
			 */
954 955
		    )
			/* cannot change a locked bit */
956 957 958 959 960
			return -EPERM;

		new = prepare_creds();
		if (!new)
			return -ENOMEM;
961
		new->securebits = arg2;
962
		return commit_creds(new);
963

964
	case PR_GET_SECUREBITS:
965
		return old->securebits;
966 967

	case PR_GET_KEEPCAPS:
968
		return !!issecure(SECURE_KEEP_CAPS);
969

970 971
	case PR_SET_KEEPCAPS:
		if (arg2 > 1) /* Note, we rely on arg2 being unsigned here */
972
			return -EINVAL;
973
		if (issecure(SECURE_KEEP_CAPS_LOCKED))
974 975 976 977 978
			return -EPERM;

		new = prepare_creds();
		if (!new)
			return -ENOMEM;
979 980
		if (arg2)
			new->securebits |= issecure_mask(SECURE_KEEP_CAPS);
981
		else
982
			new->securebits &= ~issecure_mask(SECURE_KEEP_CAPS);
983
		return commit_creds(new);
984

985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008
	case PR_CAP_AMBIENT:
		if (arg2 == PR_CAP_AMBIENT_CLEAR_ALL) {
			if (arg3 | arg4 | arg5)
				return -EINVAL;

			new = prepare_creds();
			if (!new)
				return -ENOMEM;
			cap_clear(new->cap_ambient);
			return commit_creds(new);
		}

		if (((!cap_valid(arg3)) | arg4 | arg5))
			return -EINVAL;

		if (arg2 == PR_CAP_AMBIENT_IS_SET) {
			return !!cap_raised(current_cred()->cap_ambient, arg3);
		} else if (arg2 != PR_CAP_AMBIENT_RAISE &&
			   arg2 != PR_CAP_AMBIENT_LOWER) {
			return -EINVAL;
		} else {
			if (arg2 == PR_CAP_AMBIENT_RAISE &&
			    (!cap_raised(current_cred()->cap_permitted, arg3) ||
			     !cap_raised(current_cred()->cap_inheritable,
1009 1010
					 arg3) ||
			     issecure(SECURE_NO_CAP_AMBIENT_RAISE)))
1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022
				return -EPERM;

			new = prepare_creds();
			if (!new)
				return -ENOMEM;
			if (arg2 == PR_CAP_AMBIENT_RAISE)
				cap_raise(new->cap_ambient, arg3);
			else
				cap_lower(new->cap_ambient, arg3);
			return commit_creds(new);
		}

1023 1024
	default:
		/* No functionality available - continue with default */
1025
		return -ENOSYS;
1026
	}
Linus Torvalds's avatar
Linus Torvalds committed
1027 1028
}

1029 1030 1031 1032 1033 1034
/**
 * cap_vm_enough_memory - Determine whether a new virtual mapping is permitted
 * @mm: The VM space in which the new mapping is to be made
 * @pages: The size of the mapping
 *
 * Determine whether the allocation of a new virtual mapping by the current
1035
 * task is permitted, returning 1 if permission is granted, 0 if not.
1036
 */
1037
int cap_vm_enough_memory(struct mm_struct *mm, long pages)
Linus Torvalds's avatar
Linus Torvalds committed
1038 1039 1040
{
	int cap_sys_admin = 0;

1041
	if (cap_capable(current_cred(), &init_user_ns, CAP_SYS_ADMIN,
1042
			SECURITY_CAP_NOAUDIT) == 0)
Linus Torvalds's avatar
Linus Torvalds committed
1043
		cap_sys_admin = 1;
1044
	return cap_sys_admin;
Linus Torvalds's avatar
Linus Torvalds committed
1045
}
1046 1047

/*
1048
 * cap_mmap_addr - check if able to map given addr
1049 1050
 * @addr: address attempting to be mapped
 *
1051
 * If the process is attempting to map memory below dac_mmap_min_addr they need
1052 1053 1054 1055
 * CAP_SYS_RAWIO.  The other parameters to this function are unused by the
 * capability security module.  Returns 0 if this mapping should be allowed
 * -EPERM if not.
 */
1056
int cap_mmap_addr(unsigned long addr)
1057 1058 1059
{
	int ret = 0;

1060
	if (addr < dac_mmap_min_addr) {
1061
		ret = cap_capable(current_cred(), &init_user_ns, CAP_SYS_RAWIO,
1062 1063 1064 1065 1066 1067 1068
				  SECURITY_CAP_AUDIT);
		/* set PF_SUPERPRIV if it turns out we allow the low mmap */
		if (ret == 0)
			current->flags |= PF_SUPERPRIV;
	}
	return ret;
}
1069

1070 1071
int cap_mmap_file(struct file *file, unsigned long reqprot,
		  unsigned long prot, unsigned long flags)
1072
{
1073
	return 0;
1074
}
1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104

#ifdef CONFIG_SECURITY

struct security_hook_list capability_hooks[] = {
	LSM_HOOK_INIT(capable, cap_capable),
	LSM_HOOK_INIT(settime, cap_settime),
	LSM_HOOK_INIT(ptrace_access_check, cap_ptrace_access_check),
	LSM_HOOK_INIT(ptrace_traceme, cap_ptrace_traceme),
	LSM_HOOK_INIT(capget, cap_capget),
	LSM_HOOK_INIT(capset, cap_capset),
	LSM_HOOK_INIT(bprm_set_creds, cap_bprm_set_creds),
	LSM_HOOK_INIT(bprm_secureexec, cap_bprm_secureexec),
	LSM_HOOK_INIT(inode_need_killpriv, cap_inode_need_killpriv),
	LSM_HOOK_INIT(inode_killpriv, cap_inode_killpriv),
	LSM_HOOK_INIT(mmap_addr, cap_mmap_addr),
	LSM_HOOK_INIT(mmap_file, cap_mmap_file),
	LSM_HOOK_INIT(task_fix_setuid, cap_task_fix_setuid),
	LSM_HOOK_INIT(task_prctl, cap_task_prctl),
	LSM_HOOK_INIT(task_setscheduler, cap_task_setscheduler),
	LSM_HOOK_INIT(task_setioprio, cap_task_setioprio),
	LSM_HOOK_INIT(task_setnice, cap_task_setnice),
	LSM_HOOK_INIT(vm_enough_memory, cap_vm_enough_memory),
};

void __init capability_add_hooks(void)
{
	security_add_hooks(capability_hooks, ARRAY_SIZE(capability_hooks));
}

#endif /* CONFIG_SECURITY */