util.c 22.8 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9
/*
 * linux/ipc/util.c
 * Copyright (C) 1992 Krishna Balasubramanian
 *
 * Sep 1997 - Call suser() last after "normal" permission checks so we
 *            get BSD style process accounting right.
 *            Occurs in several places in the IPC code.
 *            Chris Evans, <chris@ferret.lmh.ox.ac.uk>
 * Nov 1999 - ipc helper functions, unified SMP locking
10
 *	      Manfred Spraul <manfred@colorfullife.com>
L
Linus Torvalds 已提交
11 12
 * Oct 2002 - One lock per IPC id. RCU ipc_free for lock-free grow_ary().
 *            Mingming Cao <cmm@us.ibm.com>
S
Steve Grubb 已提交
13 14
 * Mar 2006 - support for audit of ipc object properties
 *            Dustin Kirkland <dustin.kirkland@us.ibm.com>
K
Kirill Korotaev 已提交
15 16 17
 * Jun 2006 - namespaces ssupport
 *            OpenVZ, SWsoft Inc.
 *            Pavel Emelianov <xemul@openvz.org>
L
Linus Torvalds 已提交
18 19 20 21 22 23 24 25
 */

#include <linux/mm.h>
#include <linux/shm.h>
#include <linux/init.h>
#include <linux/msg.h>
#include <linux/vmalloc.h>
#include <linux/slab.h>
26
#include <linux/capability.h>
L
Linus Torvalds 已提交
27 28 29 30
#include <linux/highuid.h>
#include <linux/security.h>
#include <linux/rcupdate.h>
#include <linux/workqueue.h>
31 32
#include <linux/seq_file.h>
#include <linux/proc_fs.h>
S
Steve Grubb 已提交
33
#include <linux/audit.h>
K
Kirill Korotaev 已提交
34
#include <linux/nsproxy.h>
N
Nadia Derbey 已提交
35
#include <linux/rwsem.h>
36
#include <linux/memory.h>
37
#include <linux/ipc_namespace.h>
L
Linus Torvalds 已提交
38 39 40 41 42

#include <asm/unistd.h>

#include "util.h"

43 44 45
struct ipc_proc_iface {
	const char *path;
	const char *header;
K
Kirill Korotaev 已提交
46
	int ids;
47 48 49
	int (*show)(struct seq_file *, void *);
};

50 51
#ifdef CONFIG_MEMORY_HOTPLUG

52 53 54 55 56 57 58 59
static void ipc_memory_notifier(struct work_struct *work)
{
	ipcns_notify(IPCNS_MEMCHANGED);
}

static DECLARE_WORK(ipc_memory_wq, ipc_memory_notifier);


60 61 62 63 64 65 66 67 68
static int ipc_memory_callback(struct notifier_block *self,
				unsigned long action, void *arg)
{
	switch (action) {
	case MEM_ONLINE:    /* memory successfully brought online */
	case MEM_OFFLINE:   /* or offline: it's time to recompute msgmni */
		/*
		 * This is done by invoking the ipcns notifier chain with the
		 * IPC_MEMCHANGED event.
69 70 71 72
		 * In order not to keep the lock on the hotplug memory chain
		 * for too long, queue a work item that will, when waken up,
		 * activate the ipcns notification chain.
		 * No need to keep several ipc work items on the queue.
73
		 */
74 75
		if (!work_pending(&ipc_memory_wq))
			schedule_work(&ipc_memory_wq);
76 77 78 79 80 81 82 83 84 85 86 87 88 89
		break;
	case MEM_GOING_ONLINE:
	case MEM_GOING_OFFLINE:
	case MEM_CANCEL_ONLINE:
	case MEM_CANCEL_OFFLINE:
	default:
		break;
	}

	return NOTIFY_OK;
}

#endif /* CONFIG_MEMORY_HOTPLUG */

L
Linus Torvalds 已提交
90 91 92 93
/**
 *	ipc_init	-	initialise IPC subsystem
 *
 *	The various system5 IPC resources (semaphores, messages and shared
94
 *	memory) are initialised
95 96 97
 *	A callback routine is registered into the memory hotplug notifier
 *	chain: since msgmni scales to lowmem this callback routine will be
 *	called upon successful memory add / remove to recompute msmgni.
L
Linus Torvalds 已提交
98 99 100 101 102 103 104
 */
 
static int __init ipc_init(void)
{
	sem_init();
	msg_init();
	shm_init();
105 106
	hotplug_memory_notifier(ipc_memory_callback, IPC_CALLBACK_PRI);
	register_ipcns_notifier(&init_ipc_ns);
L
Linus Torvalds 已提交
107 108 109 110 111 112 113 114
	return 0;
}
__initcall(ipc_init);

/**
 *	ipc_init_ids		-	initialise IPC identifiers
 *	@ids: Identifier set
 *
N
Nadia Derbey 已提交
115 116
 *	Set up the sequence range to use for the ipc identifier range (limited
 *	below IPCMNI) then initialise the ids idr.
L
Linus Torvalds 已提交
117 118
 */
 
N
Nadia Derbey 已提交
119
void ipc_init_ids(struct ipc_ids *ids)
L
Linus Torvalds 已提交
120
{
N
Nadia Derbey 已提交
121
	init_rwsem(&ids->rw_mutex);
L
Linus Torvalds 已提交
122 123 124 125 126

	ids->in_use = 0;
	ids->seq = 0;
	{
		int seq_limit = INT_MAX/SEQ_MULTIPLIER;
127 128
		if (seq_limit > USHRT_MAX)
			ids->seq_max = USHRT_MAX;
L
Linus Torvalds 已提交
129 130 131 132
		 else
		 	ids->seq_max = seq_limit;
	}

N
Nadia Derbey 已提交
133
	idr_init(&ids->ipcs_idr);
L
Linus Torvalds 已提交
134 135
}

136
#ifdef CONFIG_PROC_FS
137
static const struct file_operations sysvipc_proc_fops;
138
/**
139
 *	ipc_init_proc_interface	-  Create a proc interface for sysipc types using a seq_file interface.
140 141 142 143 144 145
 *	@path: Path in procfs
 *	@header: Banner to be printed at the beginning of the file.
 *	@ids: ipc id table to iterate.
 *	@show: show routine.
 */
void __init ipc_init_proc_interface(const char *path, const char *header,
K
Kirill Korotaev 已提交
146
		int ids, int (*show)(struct seq_file *, void *))
147 148 149 150 151 152 153 154 155 156 157 158
{
	struct proc_dir_entry *pde;
	struct ipc_proc_iface *iface;

	iface = kmalloc(sizeof(*iface), GFP_KERNEL);
	if (!iface)
		return;
	iface->path	= path;
	iface->header	= header;
	iface->ids	= ids;
	iface->show	= show;

159 160 161 162 163 164
	pde = proc_create_data(path,
			       S_IRUGO,        /* world readable */
			       NULL,           /* parent dir */
			       &sysvipc_proc_fops,
			       iface);
	if (!pde) {
165 166 167 168 169
		kfree(iface);
	}
}
#endif

L
Linus Torvalds 已提交
170 171 172 173 174
/**
 *	ipc_findkey	-	find a key in an ipc identifier set	
 *	@ids: Identifier set
 *	@key: The key to find
 *	
N
Nadia Derbey 已提交
175
 *	Requires ipc_ids.rw_mutex locked.
N
Nadia Derbey 已提交
176 177
 *	Returns the LOCKED pointer to the ipc structure if found or NULL
 *	if not.
N
Nadia Derbey 已提交
178
 *	If key is found ipc points to the owning ipc structure
L
Linus Torvalds 已提交
179 180
 */
 
N
Nadia Derbey 已提交
181
static struct kern_ipc_perm *ipc_findkey(struct ipc_ids *ids, key_t key)
L
Linus Torvalds 已提交
182
{
N
Nadia Derbey 已提交
183 184 185
	struct kern_ipc_perm *ipc;
	int next_id;
	int total;
L
Linus Torvalds 已提交
186

N
Nadia Derbey 已提交
187 188 189 190
	for (total = 0, next_id = 0; total < ids->in_use; next_id++) {
		ipc = idr_find(&ids->ipcs_idr, next_id);

		if (ipc == NULL)
L
Linus Torvalds 已提交
191
			continue;
N
Nadia Derbey 已提交
192 193 194 195 196 197 198 199

		if (ipc->key != key) {
			total++;
			continue;
		}

		ipc_lock_by_ptr(ipc);
		return ipc;
L
Linus Torvalds 已提交
200
	}
N
Nadia Derbey 已提交
201 202

	return NULL;
L
Linus Torvalds 已提交
203 204
}

N
Nadia Derbey 已提交
205 206 207 208
/**
 *	ipc_get_maxid 	-	get the last assigned id
 *	@ids: IPC identifier set
 *
N
Nadia Derbey 已提交
209
 *	Called with ipc_ids.rw_mutex held.
L
Linus Torvalds 已提交
210 211
 */

N
Nadia Derbey 已提交
212 213 214 215 216 217 218 219
int ipc_get_maxid(struct ipc_ids *ids)
{
	struct kern_ipc_perm *ipc;
	int max_id = -1;
	int total, id;

	if (ids->in_use == 0)
		return -1;
L
Linus Torvalds 已提交
220

N
Nadia Derbey 已提交
221 222 223 224 225 226 227 228 229 230 231 232 233
	if (ids->in_use == IPCMNI)
		return IPCMNI - 1;

	/* Look for the last assigned id */
	total = 0;
	for (id = 0; id < IPCMNI && total < ids->in_use; id++) {
		ipc = idr_find(&ids->ipcs_idr, id);
		if (ipc != NULL) {
			max_id = id;
			total++;
		}
	}
	return max_id;
L
Linus Torvalds 已提交
234 235 236 237 238 239
}

/**
 *	ipc_addid 	-	add an IPC identifier
 *	@ids: IPC identifier set
 *	@new: new IPC permission set
N
Nadia Derbey 已提交
240
 *	@size: limit for the number of used ids
L
Linus Torvalds 已提交
241
 *
N
Nadia Derbey 已提交
242
 *	Add an entry 'new' to the IPC ids idr. The permissions object is
L
Linus Torvalds 已提交
243
 *	initialised and the first free entry is set up and the id assigned
N
Nadia Derbey 已提交
244
 *	is returned. The 'new' entry is returned in a locked state on success.
245
 *	On failure the entry is not locked and a negative err-code is returned.
L
Linus Torvalds 已提交
246
 *
N
Nadia Derbey 已提交
247
 *	Called with ipc_ids.rw_mutex held as a writer.
L
Linus Torvalds 已提交
248 249 250 251
 */
 
int ipc_addid(struct ipc_ids* ids, struct kern_ipc_perm* new, int size)
{
252 253
	uid_t euid;
	gid_t egid;
N
Nadia Derbey 已提交
254
	int id, err;
L
Linus Torvalds 已提交
255

N
Nadia Derbey 已提交
256 257 258 259
	if (size > IPCMNI)
		size = IPCMNI;

	if (ids->in_use >= size)
260
		return -ENOSPC;
N
Nadia Derbey 已提交
261

262 263 264 265 266
	spin_lock_init(&new->lock);
	new->deleted = 0;
	rcu_read_lock();
	spin_lock(&new->lock);

N
Nadia Derbey 已提交
267
	err = idr_get_new(&ids->ipcs_idr, new, &id);
268 269 270
	if (err) {
		spin_unlock(&new->lock);
		rcu_read_unlock();
271
		return err;
272
	}
N
Nadia Derbey 已提交
273

L
Linus Torvalds 已提交
274 275
	ids->in_use++;

276 277 278
	current_euid_egid(&euid, &egid);
	new->cuid = new->uid = euid;
	new->gid = new->cgid = egid;
L
Linus Torvalds 已提交
279 280 281 282 283

	new->seq = ids->seq++;
	if(ids->seq > ids->seq_max)
		ids->seq = 0;

284
	new->id = ipc_buildid(id, new->seq);
L
Linus Torvalds 已提交
285 286 287
	return id;
}

N
Nadia Derbey 已提交
288 289 290
/**
 *	ipcget_new	-	create a new ipc object
 *	@ns: namespace
N
Nadia Derbey 已提交
291
 *	@ids: IPC identifer set
N
Nadia Derbey 已提交
292 293 294
 *	@ops: the actual creation routine to call
 *	@params: its parameters
 *
N
Nadia Derbey 已提交
295 296
 *	This routine is called by sys_msgget, sys_semget() and sys_shmget()
 *	when the key is IPC_PRIVATE.
N
Nadia Derbey 已提交
297
 */
298
static int ipcget_new(struct ipc_namespace *ns, struct ipc_ids *ids,
N
Nadia Derbey 已提交
299 300 301
		struct ipc_ops *ops, struct ipc_params *params)
{
	int err;
302
retry:
N
Nadia Derbey 已提交
303 304 305 306 307
	err = idr_pre_get(&ids->ipcs_idr, GFP_KERNEL);

	if (!err)
		return -ENOMEM;

N
Nadia Derbey 已提交
308
	down_write(&ids->rw_mutex);
N
Nadia Derbey 已提交
309
	err = ops->getnew(ns, params);
N
Nadia Derbey 已提交
310
	up_write(&ids->rw_mutex);
N
Nadia Derbey 已提交
311

312 313 314
	if (err == -EAGAIN)
		goto retry;

N
Nadia Derbey 已提交
315 316 317 318 319
	return err;
}

/**
 *	ipc_check_perms	-	check security and permissions for an IPC
320
 *	@ns: IPC namespace
N
Nadia Derbey 已提交
321 322 323
 *	@ipcp: ipc permission set
 *	@ops: the actual security routine to call
 *	@params: its parameters
N
Nadia Derbey 已提交
324 325 326 327 328 329 330
 *
 *	This routine is called by sys_msgget(), sys_semget() and sys_shmget()
 *      when the key is not IPC_PRIVATE and that key already exists in the
 *      ids IDR.
 *
 *	On success, the IPC id is returned.
 *
N
Nadia Derbey 已提交
331
 *	It is called with ipc_ids.rw_mutex and ipcp->lock held.
N
Nadia Derbey 已提交
332
 */
333 334 335 336
static int ipc_check_perms(struct ipc_namespace *ns,
			   struct kern_ipc_perm *ipcp,
			   struct ipc_ops *ops,
			   struct ipc_params *params)
N
Nadia Derbey 已提交
337 338 339
{
	int err;

340
	if (ipcperms(ns, ipcp, params->flg))
N
Nadia Derbey 已提交
341 342 343 344 345 346 347 348 349 350 351 352 353
		err = -EACCES;
	else {
		err = ops->associate(ipcp, params->flg);
		if (!err)
			err = ipcp->id;
	}

	return err;
}

/**
 *	ipcget_public	-	get an ipc object or create a new one
 *	@ns: namespace
N
Nadia Derbey 已提交
354
 *	@ids: IPC identifer set
N
Nadia Derbey 已提交
355 356 357
 *	@ops: the actual creation routine to call
 *	@params: its parameters
 *
N
Nadia Derbey 已提交
358 359 360 361 362 363
 *	This routine is called by sys_msgget, sys_semget() and sys_shmget()
 *	when the key is not IPC_PRIVATE.
 *	It adds a new entry if the key is not found and does some permission
 *      / security checkings if the key is found.
 *
 *	On success, the ipc id is returned.
N
Nadia Derbey 已提交
364
 */
365
static int ipcget_public(struct ipc_namespace *ns, struct ipc_ids *ids,
N
Nadia Derbey 已提交
366 367 368 369 370
		struct ipc_ops *ops, struct ipc_params *params)
{
	struct kern_ipc_perm *ipcp;
	int flg = params->flg;
	int err;
371
retry:
N
Nadia Derbey 已提交
372 373
	err = idr_pre_get(&ids->ipcs_idr, GFP_KERNEL);

N
Nadia Derbey 已提交
374 375 376 377 378
	/*
	 * Take the lock as a writer since we are potentially going to add
	 * a new entry + read locks are not "upgradable"
	 */
	down_write(&ids->rw_mutex);
N
Nadia Derbey 已提交
379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397
	ipcp = ipc_findkey(ids, params->key);
	if (ipcp == NULL) {
		/* key not used */
		if (!(flg & IPC_CREAT))
			err = -ENOENT;
		else if (!err)
			err = -ENOMEM;
		else
			err = ops->getnew(ns, params);
	} else {
		/* ipc object has been locked by ipc_findkey() */

		if (flg & IPC_CREAT && flg & IPC_EXCL)
			err = -EEXIST;
		else {
			err = 0;
			if (ops->more_checks)
				err = ops->more_checks(ipcp, params);
			if (!err)
N
Nadia Derbey 已提交
398 399 400 401
				/*
				 * ipc_check_perms returns the IPC id on
				 * success
				 */
402
				err = ipc_check_perms(ns, ipcp, ops, params);
N
Nadia Derbey 已提交
403 404 405
		}
		ipc_unlock(ipcp);
	}
N
Nadia Derbey 已提交
406
	up_write(&ids->rw_mutex);
N
Nadia Derbey 已提交
407

408 409 410
	if (err == -EAGAIN)
		goto retry;

N
Nadia Derbey 已提交
411 412 413 414
	return err;
}


L
Linus Torvalds 已提交
415 416
/**
 *	ipc_rmid	-	remove an IPC identifier
N
Nadia Derbey 已提交
417 418
 *	@ids: IPC identifier set
 *	@ipcp: ipc perm structure containing the identifier to remove
L
Linus Torvalds 已提交
419
 *
N
Nadia Derbey 已提交
420 421
 *	ipc_ids.rw_mutex (as a writer) and the spinlock for this ID are held
 *	before this function is called, and remain locked on the exit.
L
Linus Torvalds 已提交
422 423
 */
 
N
Nadia Derbey 已提交
424
void ipc_rmid(struct ipc_ids *ids, struct kern_ipc_perm *ipcp)
L
Linus Torvalds 已提交
425
{
426
	int lid = ipcid_to_idx(ipcp->id);
N
Nadia Derbey 已提交
427 428

	idr_remove(&ids->ipcs_idr, lid);
L
Linus Torvalds 已提交
429 430 431

	ids->in_use--;

N
Nadia Derbey 已提交
432 433 434
	ipcp->deleted = 1;

	return;
L
Linus Torvalds 已提交
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
}

/**
 *	ipc_alloc	-	allocate ipc space
 *	@size: size desired
 *
 *	Allocate memory from the appropriate pools and return a pointer to it.
 *	NULL is returned if the allocation fails
 */
 
void* ipc_alloc(int size)
{
	void* out;
	if(size > PAGE_SIZE)
		out = vmalloc(size);
	else
		out = kmalloc(size, GFP_KERNEL);
	return out;
}

/**
 *	ipc_free        -       free ipc space
 *	@ptr: pointer returned by ipc_alloc
 *	@size: size of block
 *
460
 *	Free a block created with ipc_alloc(). The caller must know the size
L
Linus Torvalds 已提交
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
 *	used in the allocation call.
 */

void ipc_free(void* ptr, int size)
{
	if(size > PAGE_SIZE)
		vfree(ptr);
	else
		kfree(ptr);
}

/*
 * rcu allocations:
 * There are three headers that are prepended to the actual allocation:
 * - during use: ipc_rcu_hdr.
 * - during the rcu grace period: ipc_rcu_grace.
 * - [only if vmalloc]: ipc_rcu_sched.
 * Their lifetime doesn't overlap, thus the headers share the same memory.
 * Unlike a normal union, they are right-aligned, thus some container_of
 * forward/backward casting is necessary:
 */
struct ipc_rcu_hdr
{
	int refcount;
	int is_vmalloc;
	void *data[0];
};


struct ipc_rcu_grace
{
	struct rcu_head rcu;
	/* "void *" makes sure alignment of following data is sane. */
	void *data[0];
};

struct ipc_rcu_sched
{
	struct work_struct work;
	/* "void *" makes sure alignment of following data is sane. */
	void *data[0];
};

#define HDRLEN_KMALLOC		(sizeof(struct ipc_rcu_grace) > sizeof(struct ipc_rcu_hdr) ? \
					sizeof(struct ipc_rcu_grace) : sizeof(struct ipc_rcu_hdr))
#define HDRLEN_VMALLOC		(sizeof(struct ipc_rcu_sched) > HDRLEN_KMALLOC ? \
					sizeof(struct ipc_rcu_sched) : HDRLEN_KMALLOC)

static inline int rcu_use_vmalloc(int size)
{
	/* Too big for a single page? */
	if (HDRLEN_KMALLOC + size > PAGE_SIZE)
		return 1;
	return 0;
}

/**
 *	ipc_rcu_alloc	-	allocate ipc and rcu space 
 *	@size: size desired
 *
 *	Allocate memory for the rcu header structure +  the object.
 *	Returns the pointer to the object.
 *	NULL is returned if the allocation fails. 
 */
 
void* ipc_rcu_alloc(int size)
{
	void* out;
	/* 
	 * We prepend the allocation with the rcu struct, and
	 * workqueue if necessary (for vmalloc). 
	 */
	if (rcu_use_vmalloc(size)) {
		out = vmalloc(HDRLEN_VMALLOC + size);
		if (out) {
			out += HDRLEN_VMALLOC;
			container_of(out, struct ipc_rcu_hdr, data)->is_vmalloc = 1;
			container_of(out, struct ipc_rcu_hdr, data)->refcount = 1;
		}
	} else {
		out = kmalloc(HDRLEN_KMALLOC + size, GFP_KERNEL);
		if (out) {
			out += HDRLEN_KMALLOC;
			container_of(out, struct ipc_rcu_hdr, data)->is_vmalloc = 0;
			container_of(out, struct ipc_rcu_hdr, data)->refcount = 1;
		}
	}

	return out;
}

void ipc_rcu_getref(void *ptr)
{
	container_of(ptr, struct ipc_rcu_hdr, data)->refcount++;
}

557 558 559 560 561
static void ipc_do_vfree(struct work_struct *work)
{
	vfree(container_of(work, struct ipc_rcu_sched, work));
}

L
Linus Torvalds 已提交
562
/**
563 564
 * ipc_schedule_free - free ipc + rcu space
 * @head: RCU callback structure for queued work
L
Linus Torvalds 已提交
565 566
 * 
 * Since RCU callback function is called in bh,
567
 * we need to defer the vfree to schedule_work().
L
Linus Torvalds 已提交
568 569 570
 */
static void ipc_schedule_free(struct rcu_head *head)
{
N
Nadia Derbey 已提交
571 572 573 574 575 576
	struct ipc_rcu_grace *grace;
	struct ipc_rcu_sched *sched;

	grace = container_of(head, struct ipc_rcu_grace, rcu);
	sched = container_of(&(grace->data[0]), struct ipc_rcu_sched,
				data[0]);
L
Linus Torvalds 已提交
577

578
	INIT_WORK(&sched->work, ipc_do_vfree);
L
Linus Torvalds 已提交
579 580 581 582 583 584 585 586 587 588 589 590
	schedule_work(&sched->work);
}

void ipc_rcu_putref(void *ptr)
{
	if (--container_of(ptr, struct ipc_rcu_hdr, data)->refcount > 0)
		return;

	if (container_of(ptr, struct ipc_rcu_hdr, data)->is_vmalloc) {
		call_rcu(&container_of(ptr, struct ipc_rcu_grace, data)->rcu,
				ipc_schedule_free);
	} else {
591
		kfree_rcu(container_of(ptr, struct ipc_rcu_grace, data), rcu);
L
Linus Torvalds 已提交
592 593 594 595 596
	}
}

/**
 *	ipcperms	-	check IPC permissions
597
 *	@ns: IPC namespace
L
Linus Torvalds 已提交
598 599 600 601 602
 *	@ipcp: IPC permission set
 *	@flag: desired permission set.
 *
 *	Check user, group, other permissions for access
 *	to ipc resources. return 0 if allowed
603 604
 *
 * 	@flag will most probably be 0 or S_...UGO from <linux/stat.h>
L
Linus Torvalds 已提交
605 606
 */
 
607 608
int ipcperms(struct ipc_namespace *ns, struct kern_ipc_perm *ipcp, short flag)
{
609
	uid_t euid = current_euid();
A
Al Viro 已提交
610
	int requested_mode, granted_mode;
L
Linus Torvalds 已提交
611

A
Al Viro 已提交
612
	audit_ipc_obj(ipcp);
L
Linus Torvalds 已提交
613 614
	requested_mode = (flag >> 6) | (flag >> 3) | flag;
	granted_mode = ipcp->mode;
615 616
	if (euid == ipcp->cuid ||
	    euid == ipcp->uid)
L
Linus Torvalds 已提交
617 618 619 620 621
		granted_mode >>= 6;
	else if (in_group_p(ipcp->cgid) || in_group_p(ipcp->gid))
		granted_mode >>= 3;
	/* is there some bit set in requested_mode but not in granted_mode? */
	if ((requested_mode & ~granted_mode & 0007) && 
622
	    !ns_capable(ns->user_ns, CAP_IPC_OWNER))
L
Linus Torvalds 已提交
623 624 625 626 627 628 629 630 631 632 633 634 635 636 637
		return -1;

	return security_ipc_permission(ipcp, flag);
}

/*
 * Functions to convert between the kern_ipc_perm structure and the
 * old/new ipc_perm structures
 */

/**
 *	kernel_to_ipc64_perm	-	convert kernel ipc permissions to user
 *	@in: kernel permissions
 *	@out: new style IPC permissions
 *
638 639
 *	Turn the kernel object @in into a set of permissions descriptions
 *	for returning to userspace (@out).
L
Linus Torvalds 已提交
640 641 642 643 644 645 646 647 648 649 650 651 652 653 654
 */
 

void kernel_to_ipc64_perm (struct kern_ipc_perm *in, struct ipc64_perm *out)
{
	out->key	= in->key;
	out->uid	= in->uid;
	out->gid	= in->gid;
	out->cuid	= in->cuid;
	out->cgid	= in->cgid;
	out->mode	= in->mode;
	out->seq	= in->seq;
}

/**
N
Nadia Derbey 已提交
655
 *	ipc64_perm_to_ipc_perm	-	convert new ipc permissions to old
L
Linus Torvalds 已提交
656 657 658
 *	@in: new style IPC permissions
 *	@out: old style IPC permissions
 *
659 660
 *	Turn the new style permissions object @in into a compatibility
 *	object and store it into the @out pointer.
L
Linus Torvalds 已提交
661 662 663 664 665 666 667 668 669 670 671 672 673
 */
 
void ipc64_perm_to_ipc_perm (struct ipc64_perm *in, struct ipc_perm *out)
{
	out->key	= in->key;
	SET_UID(out->uid, in->uid);
	SET_GID(out->gid, in->gid);
	SET_UID(out->cuid, in->cuid);
	SET_GID(out->cgid, in->cgid);
	out->mode	= in->mode;
	out->seq	= in->seq;
}

N
Nadia Derbey 已提交
674
/**
N
Nadia Derbey 已提交
675
 * ipc_lock - Lock an ipc structure without rw_mutex held
N
Nadia Derbey 已提交
676 677 678 679 680 681 682 683
 * @ids: IPC identifier set
 * @id: ipc id to look for
 *
 * Look for an id in the ipc ids idr and lock the associated ipc object.
 *
 * The ipc object is locked on exit.
 */

N
Nadia Derbey 已提交
684
struct kern_ipc_perm *ipc_lock(struct ipc_ids *ids, int id)
L
Linus Torvalds 已提交
685
{
N
Nadia Derbey 已提交
686
	struct kern_ipc_perm *out;
687
	int lid = ipcid_to_idx(id);
L
Linus Torvalds 已提交
688 689

	rcu_read_lock();
N
Nadia Derbey 已提交
690 691
	out = idr_find(&ids->ipcs_idr, lid);
	if (out == NULL) {
L
Linus Torvalds 已提交
692
		rcu_read_unlock();
693
		return ERR_PTR(-EINVAL);
L
Linus Torvalds 已提交
694
	}
N
Nadia Derbey 已提交
695

L
Linus Torvalds 已提交
696 697 698 699 700 701 702 703
	spin_lock(&out->lock);
	
	/* ipc_rmid() may have already freed the ID while ipc_lock
	 * was spinning: here verify that the structure is still valid
	 */
	if (out->deleted) {
		spin_unlock(&out->lock);
		rcu_read_unlock();
704
		return ERR_PTR(-EINVAL);
L
Linus Torvalds 已提交
705
	}
N
Nadia Derbey 已提交
706

L
Linus Torvalds 已提交
707 708 709
	return out;
}

710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744
struct kern_ipc_perm *ipc_lock_check(struct ipc_ids *ids, int id)
{
	struct kern_ipc_perm *out;

	out = ipc_lock(ids, id);
	if (IS_ERR(out))
		return out;

	if (ipc_checkid(out, id)) {
		ipc_unlock(out);
		return ERR_PTR(-EIDRM);
	}

	return out;
}

/**
 * ipcget - Common sys_*get() code
 * @ns : namsepace
 * @ids : IPC identifier set
 * @ops : operations to be called on ipc object creation, permission checks
 *        and further checks
 * @params : the parameters needed by the previous operations.
 *
 * Common routine called by sys_msgget(), sys_semget() and sys_shmget().
 */
int ipcget(struct ipc_namespace *ns, struct ipc_ids *ids,
			struct ipc_ops *ops, struct ipc_params *params)
{
	if (params->key == IPC_PRIVATE)
		return ipcget_new(ns, ids, ops, params);
	else
		return ipcget_public(ns, ids, ops, params);
}

745 746 747 748 749 750 751 752 753 754 755 756 757
/**
 * ipc_update_perm - update the permissions of an IPC.
 * @in:  the permission given as input.
 * @out: the permission of the ipc to set.
 */
void ipc_update_perm(struct ipc64_perm *in, struct kern_ipc_perm *out)
{
	out->uid = in->uid;
	out->gid = in->gid;
	out->mode = (out->mode & ~S_IRWXUGO)
		| (in->mode & S_IRWXUGO);
}

758 759
/**
 * ipcctl_pre_down - retrieve an ipc and check permissions for some IPC_XXX cmd
760
 * @ns:  the ipc namespace
761 762 763 764 765 766 767 768 769 770 771 772 773 774
 * @ids:  the table of ids where to look for the ipc
 * @id:   the id of the ipc to retrieve
 * @cmd:  the cmd to check
 * @perm: the permission to set
 * @extra_perm: one extra permission parameter used by msq
 *
 * This function does some common audit and permissions check for some IPC_XXX
 * cmd and is called from semctl_down, shmctl_down and msgctl_down.
 * It must be called without any lock held and
 *  - retrieves the ipc with the given id in the given table.
 *  - performs some audit and permission check, depending on the given cmd
 *  - returns the ipc with both ipc and rw_mutex locks held in case of success
 *    or an err-code without any lock held otherwise.
 */
775 776
struct kern_ipc_perm *ipcctl_pre_down(struct ipc_namespace *ns,
				      struct ipc_ids *ids, int id, int cmd,
777 778 779
				      struct ipc64_perm *perm, int extra_perm)
{
	struct kern_ipc_perm *ipcp;
780
	uid_t euid;
781 782 783
	int err;

	down_write(&ids->rw_mutex);
N
Nadia Derbey 已提交
784
	ipcp = ipc_lock_check(ids, id);
785 786 787 788 789
	if (IS_ERR(ipcp)) {
		err = PTR_ERR(ipcp);
		goto out_up;
	}

A
Al Viro 已提交
790
	audit_ipc_obj(ipcp);
A
Al Viro 已提交
791 792
	if (cmd == IPC_SET)
		audit_ipc_set_perm(extra_perm, perm->uid,
793
					 perm->gid, perm->mode);
794 795

	euid = current_euid();
796 797
	if (euid == ipcp->cuid || euid == ipcp->uid  ||
	    ns_capable(ns->user_ns, CAP_SYS_ADMIN))
798 799 800 801 802 803 804 805 806
		return ipcp;

	err = -EPERM;
	ipc_unlock(ipcp);
out_up:
	up_write(&ids->rw_mutex);
	return ERR_PTR(err);
}

L
Linus Torvalds 已提交
807 808 809 810 811 812 813 814
#ifdef __ARCH_WANT_IPC_PARSE_VERSION


/**
 *	ipc_parse_version	-	IPC call version
 *	@cmd: pointer to command
 *
 *	Return IPC_64 for new style IPC and IPC_OLD for old style IPC. 
815
 *	The @cmd value is turned from an encoding command and version into
L
Linus Torvalds 已提交
816 817 818 819 820 821 822 823 824 825 826 827 828 829
 *	just the command code.
 */
 
int ipc_parse_version (int *cmd)
{
	if (*cmd & IPC_64) {
		*cmd ^= IPC_64;
		return IPC_64;
	} else {
		return IPC_OLD;
	}
}

#endif /* __ARCH_WANT_IPC_PARSE_VERSION */
830 831

#ifdef CONFIG_PROC_FS
832 833 834 835 836
struct ipc_proc_iter {
	struct ipc_namespace *ns;
	struct ipc_proc_iface *iface;
};

N
Nadia Derbey 已提交
837 838 839
/*
 * This routine locks the ipc structure found at least at position pos.
 */
840 841
static struct kern_ipc_perm *sysvipc_find_ipc(struct ipc_ids *ids, loff_t pos,
					      loff_t *new_pos)
842
{
N
Nadia Derbey 已提交
843 844
	struct kern_ipc_perm *ipc;
	int total, id;
K
Kirill Korotaev 已提交
845

N
Nadia Derbey 已提交
846 847 848 849 850 851
	total = 0;
	for (id = 0; id < pos && total < ids->in_use; id++) {
		ipc = idr_find(&ids->ipcs_idr, id);
		if (ipc != NULL)
			total++;
	}
852

N
Nadia Derbey 已提交
853 854
	if (total >= ids->in_use)
		return NULL;
855

N
Nadia Derbey 已提交
856 857 858 859 860
	for ( ; pos < IPCMNI; pos++) {
		ipc = idr_find(&ids->ipcs_idr, pos);
		if (ipc != NULL) {
			*new_pos = pos + 1;
			ipc_lock_by_ptr(ipc);
861 862 863 864 865 866 867 868
			return ipc;
		}
	}

	/* Out of range - return NULL to terminate iteration */
	return NULL;
}

N
Nadia Derbey 已提交
869 870 871 872 873 874 875 876 877 878
static void *sysvipc_proc_next(struct seq_file *s, void *it, loff_t *pos)
{
	struct ipc_proc_iter *iter = s->private;
	struct ipc_proc_iface *iface = iter->iface;
	struct kern_ipc_perm *ipc = it;

	/* If we had an ipc id locked before, unlock it */
	if (ipc && ipc != SEQ_START_TOKEN)
		ipc_unlock(ipc);

879
	return sysvipc_find_ipc(&iter->ns->ids[iface->ids], *pos, pos);
N
Nadia Derbey 已提交
880 881
}

882
/*
N
Nadia Derbey 已提交
883 884
 * File positions: pos 0 -> header, pos n -> ipc id = n - 1.
 * SeqFile iterator: iterator value locked ipc pointer or SEQ_TOKEN_START.
885 886 887
 */
static void *sysvipc_proc_start(struct seq_file *s, loff_t *pos)
{
888 889
	struct ipc_proc_iter *iter = s->private;
	struct ipc_proc_iface *iface = iter->iface;
K
Kirill Korotaev 已提交
890 891
	struct ipc_ids *ids;

892
	ids = &iter->ns->ids[iface->ids];
893 894 895 896 897

	/*
	 * Take the lock - this will be released by the corresponding
	 * call to stop().
	 */
N
Nadia Derbey 已提交
898
	down_read(&ids->rw_mutex);
899 900 901 902 903 904 905 906 907 908

	/* pos < 0 is invalid */
	if (*pos < 0)
		return NULL;

	/* pos == 0 means header */
	if (*pos == 0)
		return SEQ_START_TOKEN;

	/* Find the (pos-1)th ipc */
N
Nadia Derbey 已提交
909
	return sysvipc_find_ipc(ids, *pos - 1, pos);
910 911 912 913 914
}

static void sysvipc_proc_stop(struct seq_file *s, void *it)
{
	struct kern_ipc_perm *ipc = it;
915 916
	struct ipc_proc_iter *iter = s->private;
	struct ipc_proc_iface *iface = iter->iface;
K
Kirill Korotaev 已提交
917
	struct ipc_ids *ids;
918

N
Nadia Derbey 已提交
919
	/* If we had a locked structure, release it */
920 921 922
	if (ipc && ipc != SEQ_START_TOKEN)
		ipc_unlock(ipc);

923
	ids = &iter->ns->ids[iface->ids];
924
	/* Release the lock we took in start() */
N
Nadia Derbey 已提交
925
	up_read(&ids->rw_mutex);
926 927 928 929
}

static int sysvipc_proc_show(struct seq_file *s, void *it)
{
930 931
	struct ipc_proc_iter *iter = s->private;
	struct ipc_proc_iface *iface = iter->iface;
932 933 934 935 936 937 938

	if (it == SEQ_START_TOKEN)
		return seq_puts(s, iface->header);

	return iface->show(s, it);
}

J
James Morris 已提交
939
static const struct seq_operations sysvipc_proc_seqops = {
940 941 942 943 944 945
	.start = sysvipc_proc_start,
	.stop  = sysvipc_proc_stop,
	.next  = sysvipc_proc_next,
	.show  = sysvipc_proc_show,
};

946 947
static int sysvipc_proc_open(struct inode *inode, struct file *file)
{
948 949
	int ret;
	struct seq_file *seq;
950 951 952 953 954 955
	struct ipc_proc_iter *iter;

	ret = -ENOMEM;
	iter = kmalloc(sizeof(*iter), GFP_KERNEL);
	if (!iter)
		goto out;
956 957

	ret = seq_open(file, &sysvipc_proc_seqops);
958 959 960 961 962 963 964 965 966
	if (ret)
		goto out_kfree;

	seq = file->private_data;
	seq->private = iter;

	iter->iface = PDE(inode)->data;
	iter->ns    = get_ipc_ns(current->nsproxy->ipc_ns);
out:
967
	return ret;
968 969 970 971 972 973 974 975 976 977 978
out_kfree:
	kfree(iter);
	goto out;
}

static int sysvipc_proc_release(struct inode *inode, struct file *file)
{
	struct seq_file *seq = file->private_data;
	struct ipc_proc_iter *iter = seq->private;
	put_ipc_ns(iter->ns);
	return seq_release_private(inode, file);
979 980
}

981
static const struct file_operations sysvipc_proc_fops = {
982 983 984
	.open    = sysvipc_proc_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
985
	.release = sysvipc_proc_release,
986 987
};
#endif /* CONFIG_PROC_FS */