util.c 22.9 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
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>
Linus Torvalds's avatar
Linus Torvalds committed
11 12
 * Oct 2002 - One lock per IPC id. RCU ipc_free for lock-free grow_ary().
 *            Mingming Cao <cmm@us.ibm.com>
Steve Grubb's avatar
Steve Grubb committed
13 14
 * Mar 2006 - support for audit of ipc object properties
 *            Dustin Kirkland <dustin.kirkland@us.ibm.com>
15 16 17
 * Jun 2006 - namespaces ssupport
 *            OpenVZ, SWsoft Inc.
 *            Pavel Emelianov <xemul@openvz.org>
Linus Torvalds's avatar
Linus Torvalds committed
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>
Linus Torvalds's avatar
Linus Torvalds committed
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>
Steve Grubb's avatar
Steve Grubb committed
33
#include <linux/audit.h>
34
#include <linux/nsproxy.h>
Nadia Derbey's avatar
Nadia Derbey committed
35
#include <linux/rwsem.h>
36
#include <linux/memory.h>
37
#include <linux/ipc_namespace.h>
Linus Torvalds's avatar
Linus Torvalds committed
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;
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 */

Linus Torvalds's avatar
Linus Torvalds committed
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.
Linus Torvalds's avatar
Linus Torvalds committed
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);
Linus Torvalds's avatar
Linus Torvalds committed
107 108 109 110 111 112 113 114
	return 0;
}
__initcall(ipc_init);

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

	ids->in_use = 0;
	ids->seq = 0;
	{
		int seq_limit = INT_MAX/SEQ_MULTIPLIER;
127 128
		if (seq_limit > USHORT_MAX)
			ids->seq_max = USHORT_MAX;
Linus Torvalds's avatar
Linus Torvalds committed
129 130 131 132
		 else
		 	ids->seq_max = seq_limit;
	}

Nadia Derbey's avatar
Nadia Derbey committed
133
	idr_init(&ids->ipcs_idr);
Linus Torvalds's avatar
Linus Torvalds committed
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,
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

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

Nadia Derbey's avatar
Nadia Derbey committed
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)
Linus Torvalds's avatar
Linus Torvalds committed
191
			continue;
Nadia Derbey's avatar
Nadia Derbey committed
192 193 194 195 196 197 198 199

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

		ipc_lock_by_ptr(ipc);
		return ipc;
Linus Torvalds's avatar
Linus Torvalds committed
200
	}
Nadia Derbey's avatar
Nadia Derbey committed
201 202

	return NULL;
Linus Torvalds's avatar
Linus Torvalds committed
203 204
}

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

Nadia Derbey's avatar
Nadia Derbey committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
220

Nadia Derbey's avatar
Nadia Derbey committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
234 235 236 237 238 239
}

/**
 *	ipc_addid 	-	add an IPC identifier
 *	@ids: IPC identifier set
 *	@new: new IPC permission set
Nadia Derbey's avatar
Nadia Derbey committed
240
 *	@size: limit for the number of used ids
Linus Torvalds's avatar
Linus Torvalds committed
241
 *
Nadia Derbey's avatar
Nadia Derbey committed
242
 *	Add an entry 'new' to the IPC ids idr. The permissions object is
Linus Torvalds's avatar
Linus Torvalds committed
243
 *	initialised and the first free entry is set up and the id assigned
Nadia Derbey's avatar
Nadia Derbey committed
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.
Linus Torvalds's avatar
Linus Torvalds committed
246
 *
Nadia Derbey's avatar
Nadia Derbey committed
247
 *	Called with ipc_ids.rw_mutex held as a writer.
Linus Torvalds's avatar
Linus Torvalds committed
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;
Nadia Derbey's avatar
Nadia Derbey committed
254
	int id, err;
Linus Torvalds's avatar
Linus Torvalds committed
255

Nadia Derbey's avatar
Nadia Derbey committed
256 257 258 259
	if (size > IPCMNI)
		size = IPCMNI;

	if (ids->in_use >= size)
260
		return -ENOSPC;
Nadia Derbey's avatar
Nadia Derbey committed
261

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

Nadia Derbey's avatar
Nadia Derbey committed
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
	}
Nadia Derbey's avatar
Nadia Derbey committed
273

Linus Torvalds's avatar
Linus Torvalds committed
274 275
	ids->in_use++;

276 277 278
	current_euid_egid(&euid, &egid);
	new->cuid = new->uid = euid;
	new->gid = new->cgid = egid;
Linus Torvalds's avatar
Linus Torvalds committed
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);
Linus Torvalds's avatar
Linus Torvalds committed
285 286 287
	return id;
}

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

	if (!err)
		return -ENOMEM;

Nadia Derbey's avatar
Nadia Derbey committed
308
	down_write(&ids->rw_mutex);
Nadia Derbey's avatar
Nadia Derbey committed
309
	err = ops->getnew(ns, params);
Nadia Derbey's avatar
Nadia Derbey committed
310
	up_write(&ids->rw_mutex);
Nadia Derbey's avatar
Nadia Derbey committed
311

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

Nadia Derbey's avatar
Nadia Derbey committed
315 316 317 318 319 320 321 322
	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
Nadia Derbey's avatar
Nadia Derbey committed
323 324 325 326 327 328 329
 *
 *	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.
 *
Nadia Derbey's avatar
Nadia Derbey committed
330
 *	It is called with ipc_ids.rw_mutex and ipcp->lock held.
Nadia Derbey's avatar
Nadia Derbey committed
331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350
 */
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
Nadia Derbey's avatar
Nadia Derbey committed
351
 *	@ids: IPC identifer set
Nadia Derbey's avatar
Nadia Derbey committed
352 353 354
 *	@ops: the actual creation routine to call
 *	@params: its parameters
 *
Nadia Derbey's avatar
Nadia Derbey committed
355 356 357 358 359 360
 *	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.
Nadia Derbey's avatar
Nadia Derbey committed
361
 */
362
static int ipcget_public(struct ipc_namespace *ns, struct ipc_ids *ids,
Nadia Derbey's avatar
Nadia Derbey committed
363 364 365 366 367
		struct ipc_ops *ops, struct ipc_params *params)
{
	struct kern_ipc_perm *ipcp;
	int flg = params->flg;
	int err;
368
retry:
Nadia Derbey's avatar
Nadia Derbey committed
369 370
	err = idr_pre_get(&ids->ipcs_idr, GFP_KERNEL);

Nadia Derbey's avatar
Nadia Derbey committed
371 372 373 374 375
	/*
	 * 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);
Nadia Derbey's avatar
Nadia Derbey committed
376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394
	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)
Nadia Derbey's avatar
Nadia Derbey committed
395 396 397 398
				/*
				 * ipc_check_perms returns the IPC id on
				 * success
				 */
Nadia Derbey's avatar
Nadia Derbey committed
399 400 401 402
				err = ipc_check_perms(ipcp, ops, params);
		}
		ipc_unlock(ipcp);
	}
Nadia Derbey's avatar
Nadia Derbey committed
403
	up_write(&ids->rw_mutex);
Nadia Derbey's avatar
Nadia Derbey committed
404

405 406 407
	if (err == -EAGAIN)
		goto retry;

Nadia Derbey's avatar
Nadia Derbey committed
408 409 410 411
	return err;
}


Linus Torvalds's avatar
Linus Torvalds committed
412 413
/**
 *	ipc_rmid	-	remove an IPC identifier
Nadia Derbey's avatar
Nadia Derbey committed
414 415
 *	@ids: IPC identifier set
 *	@ipcp: ipc perm structure containing the identifier to remove
Linus Torvalds's avatar
Linus Torvalds committed
416
 *
Nadia Derbey's avatar
Nadia Derbey committed
417 418
 *	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.
Linus Torvalds's avatar
Linus Torvalds committed
419 420
 */
 
Nadia Derbey's avatar
Nadia Derbey committed
421
void ipc_rmid(struct ipc_ids *ids, struct kern_ipc_perm *ipcp)
Linus Torvalds's avatar
Linus Torvalds committed
422
{
423
	int lid = ipcid_to_idx(ipcp->id);
Nadia Derbey's avatar
Nadia Derbey committed
424 425

	idr_remove(&ids->ipcs_idr, lid);
Linus Torvalds's avatar
Linus Torvalds committed
426 427 428

	ids->in_use--;

Nadia Derbey's avatar
Nadia Derbey committed
429 430 431
	ipcp->deleted = 1;

	return;
Linus Torvalds's avatar
Linus Torvalds committed
432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456
}

/**
 *	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
 *
457
 *	Free a block created with ipc_alloc(). The caller must know the size
Linus Torvalds's avatar
Linus Torvalds committed
458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553
 *	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++;
}

554 555 556 557 558
static void ipc_do_vfree(struct work_struct *work)
{
	vfree(container_of(work, struct ipc_rcu_sched, work));
}

Linus Torvalds's avatar
Linus Torvalds committed
559
/**
560 561
 * ipc_schedule_free - free ipc + rcu space
 * @head: RCU callback structure for queued work
Linus Torvalds's avatar
Linus Torvalds committed
562 563
 * 
 * Since RCU callback function is called in bh,
564
 * we need to defer the vfree to schedule_work().
Linus Torvalds's avatar
Linus Torvalds committed
565 566 567
 */
static void ipc_schedule_free(struct rcu_head *head)
{
Nadia Derbey's avatar
Nadia Derbey committed
568 569 570 571 572 573
	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]);
Linus Torvalds's avatar
Linus Torvalds committed
574

575
	INIT_WORK(&sched->work, ipc_do_vfree);
Linus Torvalds's avatar
Linus Torvalds committed
576 577 578 579
	schedule_work(&sched->work);
}

/**
580 581
 * ipc_immediate_free - free ipc + rcu space
 * @head: RCU callback structure that contains pointer to be freed
Linus Torvalds's avatar
Linus Torvalds committed
582
 *
583
 * Free from the RCU callback context.
Linus Torvalds's avatar
Linus Torvalds committed
584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616
 */
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> */
617
	uid_t euid = current_euid();
Al Viro's avatar
Al Viro committed
618
	int requested_mode, granted_mode;
Linus Torvalds's avatar
Linus Torvalds committed
619

Al Viro's avatar
Al Viro committed
620
	audit_ipc_obj(ipcp);
Linus Torvalds's avatar
Linus Torvalds committed
621 622
	requested_mode = (flag >> 6) | (flag >> 3) | flag;
	granted_mode = ipcp->mode;
623 624
	if (euid == ipcp->cuid ||
	    euid == ipcp->uid)
Linus Torvalds's avatar
Linus Torvalds committed
625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645
		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
 *
646 647
 *	Turn the kernel object @in into a set of permissions descriptions
 *	for returning to userspace (@out).
Linus Torvalds's avatar
Linus Torvalds committed
648 649 650 651 652 653 654 655 656 657 658 659 660 661 662
 */
 

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;
}

/**
Nadia Derbey's avatar
Nadia Derbey committed
663
 *	ipc64_perm_to_ipc_perm	-	convert new ipc permissions to old
Linus Torvalds's avatar
Linus Torvalds committed
664 665 666
 *	@in: new style IPC permissions
 *	@out: old style IPC permissions
 *
667 668
 *	Turn the new style permissions object @in into a compatibility
 *	object and store it into the @out pointer.
Linus Torvalds's avatar
Linus Torvalds committed
669 670 671 672 673 674 675 676 677 678 679 680 681
 */
 
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;
}

Nadia Derbey's avatar
Nadia Derbey committed
682
/**
Nadia Derbey's avatar
Nadia Derbey committed
683
 * ipc_lock - Lock an ipc structure without rw_mutex held
Nadia Derbey's avatar
Nadia Derbey committed
684 685 686 687 688 689 690 691
 * @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.
 */

Nadia Derbey's avatar
Nadia Derbey committed
692
struct kern_ipc_perm *ipc_lock(struct ipc_ids *ids, int id)
Linus Torvalds's avatar
Linus Torvalds committed
693
{
Nadia Derbey's avatar
Nadia Derbey committed
694
	struct kern_ipc_perm *out;
695
	int lid = ipcid_to_idx(id);
Linus Torvalds's avatar
Linus Torvalds committed
696 697

	rcu_read_lock();
Nadia Derbey's avatar
Nadia Derbey committed
698 699
	out = idr_find(&ids->ipcs_idr, lid);
	if (out == NULL) {
Linus Torvalds's avatar
Linus Torvalds committed
700
		rcu_read_unlock();
701
		return ERR_PTR(-EINVAL);
Linus Torvalds's avatar
Linus Torvalds committed
702
	}
Nadia Derbey's avatar
Nadia Derbey committed
703

Linus Torvalds's avatar
Linus Torvalds committed
704 705 706 707 708 709 710 711
	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();
712
		return ERR_PTR(-EINVAL);
Linus Torvalds's avatar
Linus Torvalds committed
713
	}
Nadia Derbey's avatar
Nadia Derbey committed
714

Linus Torvalds's avatar
Linus Torvalds committed
715 716 717
	return out;
}

718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752
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);
}

753 754 755 756 757 758 759 760 761 762 763 764 765
/**
 * 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);
}

766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785
/**
 * ipcctl_pre_down - retrieve an ipc and check permissions for some IPC_XXX cmd
 * @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.
 */
struct kern_ipc_perm *ipcctl_pre_down(struct ipc_ids *ids, int id, int cmd,
				      struct ipc64_perm *perm, int extra_perm)
{
	struct kern_ipc_perm *ipcp;
786
	uid_t euid;
787 788 789
	int err;

	down_write(&ids->rw_mutex);
Nadia Derbey's avatar
Nadia Derbey committed
790
	ipcp = ipc_lock_check(ids, id);
791 792 793 794 795
	if (IS_ERR(ipcp)) {
		err = PTR_ERR(ipcp);
		goto out_up;
	}

Al Viro's avatar
Al Viro committed
796
	audit_ipc_obj(ipcp);
Al Viro's avatar
Al Viro committed
797 798
	if (cmd == IPC_SET)
		audit_ipc_set_perm(extra_perm, perm->uid,
799
					 perm->gid, perm->mode);
800 801 802 803

	euid = current_euid();
	if (euid == ipcp->cuid ||
	    euid == ipcp->uid  || capable(CAP_SYS_ADMIN))
804 805 806 807 808 809 810 811 812
		return ipcp;

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

Linus Torvalds's avatar
Linus Torvalds committed
813 814 815 816 817 818 819 820
#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. 
821
 *	The @cmd value is turned from an encoding command and version into
Linus Torvalds's avatar
Linus Torvalds committed
822 823 824 825 826 827 828 829 830 831 832 833 834 835
 *	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 */
836 837

#ifdef CONFIG_PROC_FS
838 839 840 841 842
struct ipc_proc_iter {
	struct ipc_namespace *ns;
	struct ipc_proc_iface *iface;
};

Nadia Derbey's avatar
Nadia Derbey committed
843 844 845
/*
 * This routine locks the ipc structure found at least at position pos.
 */
846 847
static struct kern_ipc_perm *sysvipc_find_ipc(struct ipc_ids *ids, loff_t pos,
					      loff_t *new_pos)
848
{
Nadia Derbey's avatar
Nadia Derbey committed
849 850
	struct kern_ipc_perm *ipc;
	int total, id;
851

Nadia Derbey's avatar
Nadia Derbey committed
852 853 854 855 856 857
	total = 0;
	for (id = 0; id < pos && total < ids->in_use; id++) {
		ipc = idr_find(&ids->ipcs_idr, id);
		if (ipc != NULL)
			total++;
	}
858

Nadia Derbey's avatar
Nadia Derbey committed
859 860
	if (total >= ids->in_use)
		return NULL;
861

Nadia Derbey's avatar
Nadia Derbey committed
862 863 864 865 866
	for ( ; pos < IPCMNI; pos++) {
		ipc = idr_find(&ids->ipcs_idr, pos);
		if (ipc != NULL) {
			*new_pos = pos + 1;
			ipc_lock_by_ptr(ipc);
867 868 869 870 871 872 873 874
			return ipc;
		}
	}

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

Nadia Derbey's avatar
Nadia Derbey committed
875 876 877 878 879 880 881 882 883 884
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);

885
	return sysvipc_find_ipc(&iter->ns->ids[iface->ids], *pos, pos);
Nadia Derbey's avatar
Nadia Derbey committed
886 887
}

888
/*
Nadia Derbey's avatar
Nadia Derbey committed
889 890
 * File positions: pos 0 -> header, pos n -> ipc id = n - 1.
 * SeqFile iterator: iterator value locked ipc pointer or SEQ_TOKEN_START.
891 892 893
 */
static void *sysvipc_proc_start(struct seq_file *s, loff_t *pos)
{
894 895
	struct ipc_proc_iter *iter = s->private;
	struct ipc_proc_iface *iface = iter->iface;
896 897
	struct ipc_ids *ids;

898
	ids = &iter->ns->ids[iface->ids];
899 900 901 902 903

	/*
	 * Take the lock - this will be released by the corresponding
	 * call to stop().
	 */
Nadia Derbey's avatar
Nadia Derbey committed
904
	down_read(&ids->rw_mutex);
905 906 907 908 909 910 911 912 913 914

	/* 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 */
Nadia Derbey's avatar
Nadia Derbey committed
915
	return sysvipc_find_ipc(ids, *pos - 1, pos);
916 917 918 919 920
}

static void sysvipc_proc_stop(struct seq_file *s, void *it)
{
	struct kern_ipc_perm *ipc = it;
921 922
	struct ipc_proc_iter *iter = s->private;
	struct ipc_proc_iface *iface = iter->iface;
923
	struct ipc_ids *ids;
924

Nadia Derbey's avatar
Nadia Derbey committed
925
	/* If we had a locked structure, release it */
926 927 928
	if (ipc && ipc != SEQ_START_TOKEN)
		ipc_unlock(ipc);

929
	ids = &iter->ns->ids[iface->ids];
930
	/* Release the lock we took in start() */
Nadia Derbey's avatar
Nadia Derbey committed
931
	up_read(&ids->rw_mutex);
932 933 934 935
}

static int sysvipc_proc_show(struct seq_file *s, void *it)
{
936 937
	struct ipc_proc_iter *iter = s->private;
	struct ipc_proc_iface *iface = iter->iface;
938 939 940 941 942 943 944 945 946 947 948 949 950 951

	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,
};

952 953
static int sysvipc_proc_open(struct inode *inode, struct file *file)
{
954 955
	int ret;
	struct seq_file *seq;
956 957 958 959 960 961
	struct ipc_proc_iter *iter;

	ret = -ENOMEM;
	iter = kmalloc(sizeof(*iter), GFP_KERNEL);
	if (!iter)
		goto out;
962 963

	ret = seq_open(file, &sysvipc_proc_seqops);
964 965 966 967 968 969 970 971 972
	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:
973
	return ret;
974 975 976 977 978 979 980 981 982 983 984
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);
985 986
}

987
static const struct file_operations sysvipc_proc_fops = {
988 989 990
	.open    = sysvipc_proc_open,
	.read    = seq_read,
	.llseek  = seq_lseek,
991
	.release = sysvipc_proc_release,
992 993
};
#endif /* CONFIG_PROC_FS */