util.c 20.8 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>
L
Linus Torvalds 已提交
36 37 38 39 40

#include <asm/unistd.h>

#include "util.h"

41 42 43
struct ipc_proc_iface {
	const char *path;
	const char *header;
K
Kirill Korotaev 已提交
44
	int ids;
45 46 47
	int (*show)(struct seq_file *, void *);
};

K
Kirill Korotaev 已提交
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
struct ipc_namespace init_ipc_ns = {
	.kref = {
		.refcount	= ATOMIC_INIT(2),
	},
};

static struct ipc_namespace *clone_ipc_ns(struct ipc_namespace *old_ns)
{
	int err;
	struct ipc_namespace *ns;

	err = -ENOMEM;
	ns = kmalloc(sizeof(struct ipc_namespace), GFP_KERNEL);
	if (ns == NULL)
		goto err_mem;

	err = sem_init_ns(ns);
	if (err)
		goto err_sem;
	err = msg_init_ns(ns);
	if (err)
		goto err_msg;
	err = shm_init_ns(ns);
	if (err)
		goto err_shm;

	kref_init(&ns->kref);
	return ns;

err_shm:
	msg_exit_ns(ns);
err_msg:
	sem_exit_ns(ns);
err_sem:
	kfree(ns);
err_mem:
	return ERR_PTR(err);
}

87
struct ipc_namespace *copy_ipcs(unsigned long flags, struct ipc_namespace *ns)
K
Kirill Korotaev 已提交
88 89 90
{
	struct ipc_namespace *new_ns;

91 92
	BUG_ON(!ns);
	get_ipc_ns(ns);
K
Kirill Korotaev 已提交
93 94

	if (!(flags & CLONE_NEWIPC))
95 96 97 98 99 100
		return ns;

	new_ns = clone_ipc_ns(ns);

	put_ipc_ns(ns);
	return new_ns;
K
Kirill Korotaev 已提交
101 102 103 104 105 106 107 108 109 110 111 112 113
}

void free_ipc_ns(struct kref *kref)
{
	struct ipc_namespace *ns;

	ns = container_of(kref, struct ipc_namespace, kref);
	sem_exit_ns(ns);
	msg_exit_ns(ns);
	shm_exit_ns(ns);
	kfree(ns);
}

L
Linus Torvalds 已提交
114 115 116 117
/**
 *	ipc_init	-	initialise IPC subsystem
 *
 *	The various system5 IPC resources (semaphores, messages and shared
118
 *	memory) are initialised
L
Linus Torvalds 已提交
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133
 */
 
static int __init ipc_init(void)
{
	sem_init();
	msg_init();
	shm_init();
	return 0;
}
__initcall(ipc_init);

/**
 *	ipc_init_ids		-	initialise IPC identifiers
 *	@ids: Identifier set
 *
N
Nadia Derbey 已提交
134 135
 *	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
136 137
 */
 
N
Nadia Derbey 已提交
138
void ipc_init_ids(struct ipc_ids *ids)
L
Linus Torvalds 已提交
139
{
N
Nadia Derbey 已提交
140
	init_rwsem(&ids->rw_mutex);
L
Linus Torvalds 已提交
141 142 143 144 145 146 147 148 149 150 151

	ids->in_use = 0;
	ids->seq = 0;
	{
		int seq_limit = INT_MAX/SEQ_MULTIPLIER;
		if(seq_limit > USHRT_MAX)
			ids->seq_max = USHRT_MAX;
		 else
		 	ids->seq_max = seq_limit;
	}

N
Nadia Derbey 已提交
152
	idr_init(&ids->ipcs_idr);
L
Linus Torvalds 已提交
153 154
}

155
#ifdef CONFIG_PROC_FS
156
static const struct file_operations sysvipc_proc_fops;
157
/**
158
 *	ipc_init_proc_interface	-  Create a proc interface for sysipc types using a seq_file interface.
159 160 161 162 163 164
 *	@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
165
		int ids, int (*show)(struct seq_file *, void *))
166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189
{
	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;

	pde = create_proc_entry(path,
				S_IRUGO,        /* world readable */
				NULL            /* parent dir */);
	if (pde) {
		pde->data = iface;
		pde->proc_fops = &sysvipc_proc_fops;
	} else {
		kfree(iface);
	}
}
#endif

L
Linus Torvalds 已提交
190 191 192 193 194
/**
 *	ipc_findkey	-	find a key in an ipc identifier set	
 *	@ids: Identifier set
 *	@key: The key to find
 *	
N
Nadia Derbey 已提交
195
 *	Requires ipc_ids.rw_mutex locked.
N
Nadia Derbey 已提交
196 197
 *	Returns the LOCKED pointer to the ipc structure if found or NULL
 *	if not.
N
Nadia Derbey 已提交
198
 *	If key is found ipc points to the owning ipc structure
L
Linus Torvalds 已提交
199 200
 */
 
N
Nadia Derbey 已提交
201
static struct kern_ipc_perm *ipc_findkey(struct ipc_ids *ids, key_t key)
L
Linus Torvalds 已提交
202
{
N
Nadia Derbey 已提交
203 204 205
	struct kern_ipc_perm *ipc;
	int next_id;
	int total;
L
Linus Torvalds 已提交
206

N
Nadia Derbey 已提交
207 208 209 210
	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
211
			continue;
N
Nadia Derbey 已提交
212 213 214 215 216 217 218 219

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

		ipc_lock_by_ptr(ipc);
		return ipc;
L
Linus Torvalds 已提交
220
	}
N
Nadia Derbey 已提交
221 222

	return NULL;
L
Linus Torvalds 已提交
223 224
}

N
Nadia Derbey 已提交
225 226 227 228
/**
 *	ipc_get_maxid 	-	get the last assigned id
 *	@ids: IPC identifier set
 *
N
Nadia Derbey 已提交
229
 *	Called with ipc_ids.rw_mutex held.
L
Linus Torvalds 已提交
230 231
 */

N
Nadia Derbey 已提交
232 233 234 235 236 237 238 239
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
240

N
Nadia Derbey 已提交
241 242 243 244 245 246 247 248 249 250 251 252 253
	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
254 255 256 257 258 259
}

/**
 *	ipc_addid 	-	add an IPC identifier
 *	@ids: IPC identifier set
 *	@new: new IPC permission set
N
Nadia Derbey 已提交
260
 *	@size: limit for the number of used ids
L
Linus Torvalds 已提交
261
 *
N
Nadia Derbey 已提交
262
 *	Add an entry 'new' to the IPC ids idr. The permissions object is
L
Linus Torvalds 已提交
263
 *	initialised and the first free entry is set up and the id assigned
N
Nadia Derbey 已提交
264 265
 *	is returned. The 'new' entry is returned in a locked state on success.
 *	On failure the entry is not locked and -1 is returned.
L
Linus Torvalds 已提交
266
 *
N
Nadia Derbey 已提交
267
 *	Called with ipc_ids.rw_mutex held as a writer.
L
Linus Torvalds 已提交
268 269 270 271
 */
 
int ipc_addid(struct ipc_ids* ids, struct kern_ipc_perm* new, int size)
{
N
Nadia Derbey 已提交
272
	int id, err;
L
Linus Torvalds 已提交
273

N
Nadia Derbey 已提交
274 275 276 277 278 279 280 281 282 283
	if (size > IPCMNI)
		size = IPCMNI;

	if (ids->in_use >= size)
		return -1;

	err = idr_get_new(&ids->ipcs_idr, new, &id);
	if (err)
		return -1;

L
Linus Torvalds 已提交
284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299
	ids->in_use++;

	new->cuid = new->uid = current->euid;
	new->gid = new->cgid = current->egid;

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

	spin_lock_init(&new->lock);
	new->deleted = 0;
	rcu_read_lock();
	spin_lock(&new->lock);
	return id;
}

N
Nadia Derbey 已提交
300 301 302
/**
 *	ipcget_new	-	create a new ipc object
 *	@ns: namespace
N
Nadia Derbey 已提交
303
 *	@ids: IPC identifer set
N
Nadia Derbey 已提交
304 305 306
 *	@ops: the actual creation routine to call
 *	@params: its parameters
 *
N
Nadia Derbey 已提交
307 308
 *	This routine is called by sys_msgget, sys_semget() and sys_shmget()
 *	when the key is IPC_PRIVATE.
N
Nadia Derbey 已提交
309 310 311 312 313 314 315 316 317 318 319
 */
int ipcget_new(struct ipc_namespace *ns, struct ipc_ids *ids,
		struct ipc_ops *ops, struct ipc_params *params)
{
	int err;

	err = idr_pre_get(&ids->ipcs_idr, GFP_KERNEL);

	if (!err)
		return -ENOMEM;

N
Nadia Derbey 已提交
320
	down_write(&ids->rw_mutex);
N
Nadia Derbey 已提交
321
	err = ops->getnew(ns, params);
N
Nadia Derbey 已提交
322
	up_write(&ids->rw_mutex);
N
Nadia Derbey 已提交
323 324 325 326 327 328 329 330 331

	return err;
}

/**
 *	ipc_check_perms	-	check security and permissions for an IPC
 *	@ipcp: ipc permission set
 *	@ops: the actual security routine to call
 *	@params: its parameters
N
Nadia Derbey 已提交
332 333 334 335 336 337 338
 *
 *	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
339
 *	It is called with ipc_ids.rw_mutex and ipcp->lock held.
N
Nadia Derbey 已提交
340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359
 */
static int ipc_check_perms(struct kern_ipc_perm *ipcp, struct ipc_ops *ops,
			struct ipc_params *params)
{
	int err;

	if (ipcperms(ipcp, params->flg))
		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
360
 *	@ids: IPC identifer set
N
Nadia Derbey 已提交
361 362 363
 *	@ops: the actual creation routine to call
 *	@params: its parameters
 *
N
Nadia Derbey 已提交
364 365 366 367 368 369
 *	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
370 371 372 373 374 375 376 377 378 379
 */
int ipcget_public(struct ipc_namespace *ns, struct ipc_ids *ids,
		struct ipc_ops *ops, struct ipc_params *params)
{
	struct kern_ipc_perm *ipcp;
	int flg = params->flg;
	int err;

	err = idr_pre_get(&ids->ipcs_idr, GFP_KERNEL);

N
Nadia Derbey 已提交
380 381 382 383 384
	/*
	 * 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
385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403
	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
404 405 406 407
				/*
				 * ipc_check_perms returns the IPC id on
				 * success
				 */
N
Nadia Derbey 已提交
408 409 410 411
				err = ipc_check_perms(ipcp, ops, params);
		}
		ipc_unlock(ipcp);
	}
N
Nadia Derbey 已提交
412
	up_write(&ids->rw_mutex);
N
Nadia Derbey 已提交
413 414 415 416 417

	return err;
}


L
Linus Torvalds 已提交
418 419
/**
 *	ipc_rmid	-	remove an IPC identifier
N
Nadia Derbey 已提交
420 421
 *	@ids: IPC identifier set
 *	@ipcp: ipc perm structure containing the identifier to remove
L
Linus Torvalds 已提交
422
 *
N
Nadia Derbey 已提交
423 424
 *	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
425 426
 */
 
N
Nadia Derbey 已提交
427
void ipc_rmid(struct ipc_ids *ids, struct kern_ipc_perm *ipcp)
L
Linus Torvalds 已提交
428
{
429
	int lid = ipcid_to_idx(ipcp->id);
N
Nadia Derbey 已提交
430 431

	idr_remove(&ids->ipcs_idr, lid);
L
Linus Torvalds 已提交
432 433 434

	ids->in_use--;

N
Nadia Derbey 已提交
435 436 437
	ipcp->deleted = 1;

	return;
L
Linus Torvalds 已提交
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
}

/**
 *	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
 *
463
 *	Free a block created with ipc_alloc(). The caller must know the size
L
Linus Torvalds 已提交
464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559
 *	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++;
}

560 561 562 563 564
static void ipc_do_vfree(struct work_struct *work)
{
	vfree(container_of(work, struct ipc_rcu_sched, work));
}

L
Linus Torvalds 已提交
565
/**
566 567
 * ipc_schedule_free - free ipc + rcu space
 * @head: RCU callback structure for queued work
L
Linus Torvalds 已提交
568 569
 * 
 * Since RCU callback function is called in bh,
570
 * we need to defer the vfree to schedule_work().
L
Linus Torvalds 已提交
571 572 573
 */
static void ipc_schedule_free(struct rcu_head *head)
{
N
Nadia Derbey 已提交
574 575 576 577 578 579
	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
580

581
	INIT_WORK(&sched->work, ipc_do_vfree);
L
Linus Torvalds 已提交
582 583 584 585
	schedule_work(&sched->work);
}

/**
586 587
 * ipc_immediate_free - free ipc + rcu space
 * @head: RCU callback structure that contains pointer to be freed
L
Linus Torvalds 已提交
588
 *
589
 * Free from the RCU callback context.
L
Linus Torvalds 已提交
590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622
 */
static void ipc_immediate_free(struct rcu_head *head)
{
	struct ipc_rcu_grace *free =
		container_of(head, struct ipc_rcu_grace, rcu);
	kfree(free);
}

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 {
		call_rcu(&container_of(ptr, struct ipc_rcu_grace, data)->rcu,
				ipc_immediate_free);
	}
}

/**
 *	ipcperms	-	check IPC permissions
 *	@ipcp: IPC permission set
 *	@flag: desired permission set.
 *
 *	Check user, group, other permissions for access
 *	to ipc resources. return 0 if allowed
 */
 
int ipcperms (struct kern_ipc_perm *ipcp, short flag)
{	/* flag will most probably be 0 or S_...UGO from <linux/stat.h> */
S
Steve Grubb 已提交
623
	int requested_mode, granted_mode, err;
L
Linus Torvalds 已提交
624

S
Steve Grubb 已提交
625 626
	if (unlikely((err = audit_ipc_obj(ipcp))))
		return err;
L
Linus Torvalds 已提交
627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650
	requested_mode = (flag >> 6) | (flag >> 3) | flag;
	granted_mode = ipcp->mode;
	if (current->euid == ipcp->cuid || current->euid == ipcp->uid)
		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) && 
	    !capable(CAP_IPC_OWNER))
		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
 *
651 652
 *	Turn the kernel object @in into a set of permissions descriptions
 *	for returning to userspace (@out).
L
Linus Torvalds 已提交
653 654 655 656 657 658 659 660 661 662 663 664 665 666 667
 */
 

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
668
 *	ipc64_perm_to_ipc_perm	-	convert new ipc permissions to old
L
Linus Torvalds 已提交
669 670 671
 *	@in: new style IPC permissions
 *	@out: old style IPC permissions
 *
672 673
 *	Turn the new style permissions object @in into a compatibility
 *	object and store it into the @out pointer.
L
Linus Torvalds 已提交
674 675 676 677 678 679 680 681 682 683 684 685 686
 */
 
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
687
/**
N
Nadia Derbey 已提交
688
 * ipc_lock - Lock an ipc structure without rw_mutex held
N
Nadia Derbey 已提交
689 690 691 692 693 694
 * @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
695 696 697 698
 *
 * This is the routine that should be called when the rw_mutex is not already
 * held, i.e. idr tree not protected: it protects the idr tree in read mode
 * during the idr_find().
N
Nadia Derbey 已提交
699 700
 */

N
Nadia Derbey 已提交
701
struct kern_ipc_perm *ipc_lock(struct ipc_ids *ids, int id)
L
Linus Torvalds 已提交
702
{
N
Nadia Derbey 已提交
703
	struct kern_ipc_perm *out;
704
	int lid = ipcid_to_idx(id);
L
Linus Torvalds 已提交
705

N
Nadia Derbey 已提交
706 707
	down_read(&ids->rw_mutex);

L
Linus Torvalds 已提交
708
	rcu_read_lock();
N
Nadia Derbey 已提交
709 710
	out = idr_find(&ids->ipcs_idr, lid);
	if (out == NULL) {
L
Linus Torvalds 已提交
711
		rcu_read_unlock();
N
Nadia Derbey 已提交
712
		up_read(&ids->rw_mutex);
713
		return ERR_PTR(-EINVAL);
L
Linus Torvalds 已提交
714
	}
N
Nadia Derbey 已提交
715

N
Nadia Derbey 已提交
716 717
	up_read(&ids->rw_mutex);

L
Linus Torvalds 已提交
718 719 720 721 722 723 724 725
	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();
726
		return ERR_PTR(-EINVAL);
L
Linus Torvalds 已提交
727
	}
N
Nadia Derbey 已提交
728

L
Linus Torvalds 已提交
729 730 731
	return out;
}

N
Nadia Derbey 已提交
732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765
/**
 * ipc_lock_down - Lock an ipc structure with rw_sem held
 * @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.
 *
 * This is the routine that should be called when the rw_mutex is already
 * held, i.e. idr tree protected.
 */

struct kern_ipc_perm *ipc_lock_down(struct ipc_ids *ids, int id)
{
	struct kern_ipc_perm *out;
	int lid = ipcid_to_idx(id);

	rcu_read_lock();
	out = idr_find(&ids->ipcs_idr, lid);
	if (out == NULL) {
		rcu_read_unlock();
		return ERR_PTR(-EINVAL);
	}

	spin_lock(&out->lock);

	/*
	 * No need to verify that the structure is still valid since the
	 * rw_mutex is held.
	 */
	return out;
}

L
Linus Torvalds 已提交
766 767 768 769 770 771 772 773
#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. 
774
 *	The @cmd value is turned from an encoding command and version into
L
Linus Torvalds 已提交
775 776 777 778 779 780 781 782 783 784 785 786 787 788
 *	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 */
789 790

#ifdef CONFIG_PROC_FS
791 792 793 794 795
struct ipc_proc_iter {
	struct ipc_namespace *ns;
	struct ipc_proc_iface *iface;
};

N
Nadia Derbey 已提交
796 797 798 799 800
/*
 * This routine locks the ipc structure found at least at position pos.
 */
struct kern_ipc_perm *sysvipc_find_ipc(struct ipc_ids *ids, loff_t pos,
					loff_t *new_pos)
801
{
N
Nadia Derbey 已提交
802 803
	struct kern_ipc_perm *ipc;
	int total, id;
K
Kirill Korotaev 已提交
804

N
Nadia Derbey 已提交
805 806 807 808 809 810
	total = 0;
	for (id = 0; id < pos && total < ids->in_use; id++) {
		ipc = idr_find(&ids->ipcs_idr, id);
		if (ipc != NULL)
			total++;
	}
811

N
Nadia Derbey 已提交
812 813
	if (total >= ids->in_use)
		return NULL;
814

N
Nadia Derbey 已提交
815 816 817 818 819
	for ( ; pos < IPCMNI; pos++) {
		ipc = idr_find(&ids->ipcs_idr, pos);
		if (ipc != NULL) {
			*new_pos = pos + 1;
			ipc_lock_by_ptr(ipc);
820 821 822 823 824 825 826 827
			return ipc;
		}
	}

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

N
Nadia Derbey 已提交
828 829 830 831 832 833 834 835 836 837 838 839 840
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);

	return sysvipc_find_ipc(iter->ns->ids[iface->ids], *pos, pos);
}

841
/*
N
Nadia Derbey 已提交
842 843
 * File positions: pos 0 -> header, pos n -> ipc id = n - 1.
 * SeqFile iterator: iterator value locked ipc pointer or SEQ_TOKEN_START.
844 845 846
 */
static void *sysvipc_proc_start(struct seq_file *s, loff_t *pos)
{
847 848
	struct ipc_proc_iter *iter = s->private;
	struct ipc_proc_iface *iface = iter->iface;
K
Kirill Korotaev 已提交
849 850
	struct ipc_ids *ids;

851
	ids = iter->ns->ids[iface->ids];
852 853 854 855 856

	/*
	 * Take the lock - this will be released by the corresponding
	 * call to stop().
	 */
N
Nadia Derbey 已提交
857
	down_read(&ids->rw_mutex);
858 859 860 861 862 863 864 865 866 867

	/* 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
868
	return sysvipc_find_ipc(ids, *pos - 1, pos);
869 870 871 872 873
}

static void sysvipc_proc_stop(struct seq_file *s, void *it)
{
	struct kern_ipc_perm *ipc = it;
874 875
	struct ipc_proc_iter *iter = s->private;
	struct ipc_proc_iface *iface = iter->iface;
K
Kirill Korotaev 已提交
876
	struct ipc_ids *ids;
877

N
Nadia Derbey 已提交
878
	/* If we had a locked structure, release it */
879 880 881
	if (ipc && ipc != SEQ_START_TOKEN)
		ipc_unlock(ipc);

882
	ids = iter->ns->ids[iface->ids];
883
	/* Release the lock we took in start() */
N
Nadia Derbey 已提交
884
	up_read(&ids->rw_mutex);
885 886 887 888
}

static int sysvipc_proc_show(struct seq_file *s, void *it)
{
889 890
	struct ipc_proc_iter *iter = s->private;
	struct ipc_proc_iface *iface = iter->iface;
891 892 893 894 895 896 897 898 899 900 901 902 903 904

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

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

static struct seq_operations sysvipc_proc_seqops = {
	.start = sysvipc_proc_start,
	.stop  = sysvipc_proc_stop,
	.next  = sysvipc_proc_next,
	.show  = sysvipc_proc_show,
};

905 906
static int sysvipc_proc_open(struct inode *inode, struct file *file)
{
907 908
	int ret;
	struct seq_file *seq;
909 910 911 912 913 914
	struct ipc_proc_iter *iter;

	ret = -ENOMEM;
	iter = kmalloc(sizeof(*iter), GFP_KERNEL);
	if (!iter)
		goto out;
915 916

	ret = seq_open(file, &sysvipc_proc_seqops);
917 918 919 920 921 922 923 924 925
	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:
926
	return ret;
927 928 929 930 931 932 933 934 935 936 937
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);
938 939
}

940
static const struct file_operations sysvipc_proc_fops = {
941 942 943
	.open    = sysvipc_proc_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
944
	.release = sysvipc_proc_release,
945 946
};
#endif /* CONFIG_PROC_FS */