sem.c 42.7 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8
/*
 * linux/ipc/sem.c
 * Copyright (C) 1992 Krishna Balasubramanian
 * Copyright (C) 1995 Eric Schenk, Bruno Haible
 *
 * /proc/sysvipc/sem support (c) 1999 Dragos Acostachioaie <dragos@iname.com>
 *
 * SMP-threaded, sysctl's added
9
 * (c) 1999 Manfred Spraul <manfred@colorfullife.com>
L
Linus Torvalds 已提交
10
 * Enforced range limit on SEM_UNDO
A
Alan Cox 已提交
11
 * (c) 2001 Red Hat Inc
L
Linus Torvalds 已提交
12 13
 * Lockless wakeup
 * (c) 2003 Manfred Spraul <manfred@colorfullife.com>
14 15
 * Further wakeup optimizations, documentation
 * (c) 2010 Manfred Spraul <manfred@colorfullife.com>
S
Steve Grubb 已提交
16 17 18
 *
 * support for audit of ipc object properties and permission changes
 * Dustin Kirkland <dustin.kirkland@us.ibm.com>
K
Kirill Korotaev 已提交
19 20 21 22
 *
 * namespaces support
 * OpenVZ, SWsoft Inc.
 * Pavel Emelianov <xemul@openvz.org>
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 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
 *
 * Implementation notes: (May 2010)
 * This file implements System V semaphores.
 *
 * User space visible behavior:
 * - FIFO ordering for semop() operations (just FIFO, not starvation
 *   protection)
 * - multiple semaphore operations that alter the same semaphore in
 *   one semop() are handled.
 * - sem_ctime (time of last semctl()) is updated in the IPC_SET, SETVAL and
 *   SETALL calls.
 * - two Linux specific semctl() commands: SEM_STAT, SEM_INFO.
 * - undo adjustments at process exit are limited to 0..SEMVMX.
 * - namespace are supported.
 * - SEMMSL, SEMMNS, SEMOPM and SEMMNI can be configured at runtine by writing
 *   to /proc/sys/kernel/sem.
 * - statistics about the usage are reported in /proc/sysvipc/sem.
 *
 * Internals:
 * - scalability:
 *   - all global variables are read-mostly.
 *   - semop() calls and semctl(RMID) are synchronized by RCU.
 *   - most operations do write operations (actually: spin_lock calls) to
 *     the per-semaphore array structure.
 *   Thus: Perfect SMP scaling between independent semaphore arrays.
 *         If multiple semaphores in one array are used, then cache line
 *         trashing on the semaphore array spinlock will limit the scaling.
 * - semncnt and semzcnt are calculated on demand in count_semncnt() and
 *   count_semzcnt()
 * - the task that performs a successful semop() scans the list of all
 *   sleeping tasks and completes any pending operations that can be fulfilled.
 *   Semaphores are actively given to waiting tasks (necessary for FIFO).
 *   (see update_queue())
 * - To improve the scalability, the actual wake-up calls are performed after
 *   dropping all locks. (see wake_up_sem_queue_prepare(),
 *   wake_up_sem_queue_do())
 * - All work is done by the waker, the woken up task does not have to do
 *   anything - not even acquiring a lock or dropping a refcount.
 * - A woken up task may not even touch the semaphore array anymore, it may
 *   have been destroyed already by a semctl(RMID).
 * - The synchronizations between wake-ups due to a timeout/signal and a
 *   wake-up due to a completed semaphore operation is achieved by using an
 *   intermediate state (IN_WAKEUP).
 * - UNDO values are stored in an array (one per process and per
 *   semaphore array, lazily allocated). For backwards compatibility, multiple
 *   modes for the UNDO variables are supported (per process, per thread)
 *   (see copy_semundo, CLONE_SYSVSEM)
 * - There are two lists of the pending operations: a per-array list
 *   and per-semaphore list (stored in the array). This allows to achieve FIFO
 *   ordering without always scanning all pending operations.
 *   The worst-case behavior is nevertheless O(N^2) for N wakeups.
L
Linus Torvalds 已提交
74 75 76 77 78 79 80 81 82 83
 */

#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/init.h>
#include <linux/proc_fs.h>
#include <linux/time.h>
#include <linux/security.h>
#include <linux/syscalls.h>
#include <linux/audit.h>
84
#include <linux/capability.h>
85
#include <linux/seq_file.h>
N
Nadia Derbey 已提交
86
#include <linux/rwsem.h>
K
Kirill Korotaev 已提交
87
#include <linux/nsproxy.h>
88
#include <linux/ipc_namespace.h>
I
Ingo Molnar 已提交
89

L
Linus Torvalds 已提交
90 91 92
#include <asm/uaccess.h>
#include "util.h"

93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138
/* One semaphore structure for each semaphore in the system. */
struct sem {
	int	semval;		/* current value */
	int	sempid;		/* pid of last operation */
	struct list_head sem_pending; /* pending single-sop operations */
};

/* One queue for each sleeping process in the system. */
struct sem_queue {
	struct list_head	simple_list; /* queue of pending operations */
	struct list_head	list;	 /* queue of pending operations */
	struct task_struct	*sleeper; /* this process */
	struct sem_undo		*undo;	 /* undo structure */
	int			pid;	 /* process id of requesting process */
	int			status;	 /* completion status of operation */
	struct sembuf		*sops;	 /* array of pending operations */
	int			nsops;	 /* number of operations */
	int			alter;	 /* does *sops alter the array? */
};

/* Each task has a list of undo requests. They are executed automatically
 * when the process exits.
 */
struct sem_undo {
	struct list_head	list_proc;	/* per-process list: *
						 * all undos from one process
						 * rcu protected */
	struct rcu_head		rcu;		/* rcu struct for sem_undo */
	struct sem_undo_list	*ulp;		/* back ptr to sem_undo_list */
	struct list_head	list_id;	/* per semaphore array list:
						 * all undos for one array */
	int			semid;		/* semaphore set identifier */
	short			*semadj;	/* array of adjustments */
						/* one per semaphore */
};

/* sem_undo_list controls shared access to the list of sem_undo structures
 * that may be shared among all a CLONE_SYSVSEM task group.
 */
struct sem_undo_list {
	atomic_t		refcnt;
	spinlock_t		lock;
	struct list_head	list_proc;
};


139
#define sem_ids(ns)	((ns)->ids[IPC_SEM_IDS])
K
Kirill Korotaev 已提交
140 141

#define sem_unlock(sma)		ipc_unlock(&(sma)->sem_perm)
N
Nadia Derbey 已提交
142
#define sem_checkid(sma, semid)	ipc_checkid(&sma->sem_perm, semid)
L
Linus Torvalds 已提交
143

N
Nadia Derbey 已提交
144
static int newary(struct ipc_namespace *, struct ipc_params *);
145
static void freeary(struct ipc_namespace *, struct kern_ipc_perm *);
L
Linus Torvalds 已提交
146
#ifdef CONFIG_PROC_FS
147
static int sysvipc_sem_proc_show(struct seq_file *s, void *it);
L
Linus Torvalds 已提交
148 149 150 151 152 153 154 155 156 157 158 159 160 161
#endif

#define SEMMSL_FAST	256 /* 512 bytes on stack */
#define SEMOPM_FAST	64  /* ~ 372 bytes on stack */

/*
 * linked list protection:
 *	sem_undo.id_next,
 *	sem_array.sem_pending{,last},
 *	sem_array.sem_undo: sem_lock() for read/write
 *	sem_undo.proc_next: only "current" is allowed to read/write that field.
 *	
 */

K
Kirill Korotaev 已提交
162 163 164 165 166
#define sc_semmsl	sem_ctls[0]
#define sc_semmns	sem_ctls[1]
#define sc_semopm	sem_ctls[2]
#define sc_semmni	sem_ctls[3]

167
void sem_init_ns(struct ipc_namespace *ns)
K
Kirill Korotaev 已提交
168 169 170 171 172 173
{
	ns->sc_semmsl = SEMMSL;
	ns->sc_semmns = SEMMNS;
	ns->sc_semopm = SEMOPM;
	ns->sc_semmni = SEMMNI;
	ns->used_sems = 0;
174
	ipc_init_ids(&ns->ids[IPC_SEM_IDS]);
K
Kirill Korotaev 已提交
175 176
}

177
#ifdef CONFIG_IPC_NS
K
Kirill Korotaev 已提交
178 179
void sem_exit_ns(struct ipc_namespace *ns)
{
180
	free_ipcs(ns, &sem_ids(ns), freeary);
S
Serge E. Hallyn 已提交
181
	idr_destroy(&ns->ids[IPC_SEM_IDS].ipcs_idr);
K
Kirill Korotaev 已提交
182
}
183
#endif
L
Linus Torvalds 已提交
184 185 186

void __init sem_init (void)
{
187
	sem_init_ns(&init_ipc_ns);
188 189
	ipc_init_proc_interface("sysvipc/sem",
				"       key      semid perms      nsems   uid   gid  cuid  cgid      otime      ctime\n",
K
Kirill Korotaev 已提交
190
				IPC_SEM_IDS, sysvipc_sem_proc_show);
L
Linus Torvalds 已提交
191 192
}

N
Nadia Derbey 已提交
193 194 195 196
/*
 * sem_lock_(check_) routines are called in the paths where the rw_mutex
 * is not held.
 */
197 198
static inline struct sem_array *sem_lock(struct ipc_namespace *ns, int id)
{
N
Nadia Derbey 已提交
199 200
	struct kern_ipc_perm *ipcp = ipc_lock(&sem_ids(ns), id);

201 202 203
	if (IS_ERR(ipcp))
		return (struct sem_array *)ipcp;

N
Nadia Derbey 已提交
204
	return container_of(ipcp, struct sem_array, sem_perm);
205 206 207 208 209
}

static inline struct sem_array *sem_lock_check(struct ipc_namespace *ns,
						int id)
{
N
Nadia Derbey 已提交
210 211
	struct kern_ipc_perm *ipcp = ipc_lock_check(&sem_ids(ns), id);

212 213 214
	if (IS_ERR(ipcp))
		return (struct sem_array *)ipcp;

N
Nadia Derbey 已提交
215
	return container_of(ipcp, struct sem_array, sem_perm);
216 217
}

218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236
static inline void sem_lock_and_putref(struct sem_array *sma)
{
	ipc_lock_by_ptr(&sma->sem_perm);
	ipc_rcu_putref(sma);
}

static inline void sem_getref_and_unlock(struct sem_array *sma)
{
	ipc_rcu_getref(sma);
	ipc_unlock(&(sma)->sem_perm);
}

static inline void sem_putref(struct sem_array *sma)
{
	ipc_lock_by_ptr(&sma->sem_perm);
	ipc_rcu_putref(sma);
	ipc_unlock(&(sma)->sem_perm);
}

N
Nadia Derbey 已提交
237 238 239 240 241
static inline void sem_rmid(struct ipc_namespace *ns, struct sem_array *s)
{
	ipc_rmid(&sem_ids(ns), &s->sem_perm);
}

L
Linus Torvalds 已提交
242 243 244 245 246 247 248 249 250 251 252 253 254 255 256
/*
 * Lockless wakeup algorithm:
 * Without the check/retry algorithm a lockless wakeup is possible:
 * - queue.status is initialized to -EINTR before blocking.
 * - wakeup is performed by
 *	* unlinking the queue entry from sma->sem_pending
 *	* setting queue.status to IN_WAKEUP
 *	  This is the notification for the blocked thread that a
 *	  result value is imminent.
 *	* call wake_up_process
 *	* set queue.status to the final value.
 * - the previously blocked thread checks queue.status:
 *   	* if it's IN_WAKEUP, then it must wait until the value changes
 *   	* if it's not -EINTR, then the operation was completed by
 *   	  update_queue. semtimedop can return queue.status without
I
Ingo Molnar 已提交
257
 *   	  performing any operation on the sem array.
L
Linus Torvalds 已提交
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275
 *   	* otherwise it must acquire the spinlock and check what's up.
 *
 * The two-stage algorithm is necessary to protect against the following
 * races:
 * - if queue.status is set after wake_up_process, then the woken up idle
 *   thread could race forward and try (and fail) to acquire sma->lock
 *   before update_queue had a chance to set queue.status
 * - if queue.status is written before wake_up_process and if the
 *   blocked process is woken up by a signal between writing
 *   queue.status and the wake_up_process, then the woken up
 *   process could return from semtimedop and die by calling
 *   sys_exit before wake_up_process is called. Then wake_up_process
 *   will oops, because the task structure is already invalid.
 *   (yes, this happened on s390 with sysv msg).
 *
 */
#define IN_WAKEUP	1

N
Nadia Derbey 已提交
276 277 278 279 280
/**
 * newary - Create a new semaphore set
 * @ns: namespace
 * @params: ptr to the structure that contains key, semflg and nsems
 *
N
Nadia Derbey 已提交
281
 * Called with sem_ids.rw_mutex held (as a writer)
N
Nadia Derbey 已提交
282 283
 */

N
Nadia Derbey 已提交
284
static int newary(struct ipc_namespace *ns, struct ipc_params *params)
L
Linus Torvalds 已提交
285 286 287 288 289
{
	int id;
	int retval;
	struct sem_array *sma;
	int size;
N
Nadia Derbey 已提交
290 291 292
	key_t key = params->key;
	int nsems = params->u.nsems;
	int semflg = params->flg;
293
	int i;
L
Linus Torvalds 已提交
294 295 296

	if (!nsems)
		return -EINVAL;
K
Kirill Korotaev 已提交
297
	if (ns->used_sems + nsems > ns->sc_semmns)
L
Linus Torvalds 已提交
298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316
		return -ENOSPC;

	size = sizeof (*sma) + nsems * sizeof (struct sem);
	sma = ipc_rcu_alloc(size);
	if (!sma) {
		return -ENOMEM;
	}
	memset (sma, 0, size);

	sma->sem_perm.mode = (semflg & S_IRWXUGO);
	sma->sem_perm.key = key;

	sma->sem_perm.security = NULL;
	retval = security_sem_alloc(sma);
	if (retval) {
		ipc_rcu_putref(sma);
		return retval;
	}

K
Kirill Korotaev 已提交
317
	id = ipc_addid(&sem_ids(ns), &sma->sem_perm, ns->sc_semmni);
318
	if (id < 0) {
L
Linus Torvalds 已提交
319 320
		security_sem_free(sma);
		ipc_rcu_putref(sma);
321
		return id;
L
Linus Torvalds 已提交
322
	}
K
Kirill Korotaev 已提交
323
	ns->used_sems += nsems;
L
Linus Torvalds 已提交
324 325

	sma->sem_base = (struct sem *) &sma[1];
326 327 328 329 330

	for (i = 0; i < nsems; i++)
		INIT_LIST_HEAD(&sma->sem_base[i].sem_pending);

	sma->complex_count = 0;
331
	INIT_LIST_HEAD(&sma->sem_pending);
332
	INIT_LIST_HEAD(&sma->list_id);
L
Linus Torvalds 已提交
333 334 335 336
	sma->sem_nsems = nsems;
	sma->sem_ctime = get_seconds();
	sem_unlock(sma);

N
Nadia Derbey 已提交
337
	return sma->sem_perm.id;
L
Linus Torvalds 已提交
338 339
}

N
Nadia Derbey 已提交
340

N
Nadia Derbey 已提交
341
/*
N
Nadia Derbey 已提交
342
 * Called with sem_ids.rw_mutex and ipcp locked.
N
Nadia Derbey 已提交
343
 */
N
Nadia Derbey 已提交
344
static inline int sem_security(struct kern_ipc_perm *ipcp, int semflg)
N
Nadia Derbey 已提交
345
{
N
Nadia Derbey 已提交
346 347 348 349
	struct sem_array *sma;

	sma = container_of(ipcp, struct sem_array, sem_perm);
	return security_sem_associate(sma, semflg);
N
Nadia Derbey 已提交
350 351
}

N
Nadia Derbey 已提交
352
/*
N
Nadia Derbey 已提交
353
 * Called with sem_ids.rw_mutex and ipcp locked.
N
Nadia Derbey 已提交
354
 */
N
Nadia Derbey 已提交
355 356
static inline int sem_more_checks(struct kern_ipc_perm *ipcp,
				struct ipc_params *params)
N
Nadia Derbey 已提交
357
{
N
Nadia Derbey 已提交
358 359 360 361
	struct sem_array *sma;

	sma = container_of(ipcp, struct sem_array, sem_perm);
	if (params->u.nsems > sma->sem_nsems)
N
Nadia Derbey 已提交
362 363 364 365 366
		return -EINVAL;

	return 0;
}

367
SYSCALL_DEFINE3(semget, key_t, key, int, nsems, int, semflg)
L
Linus Torvalds 已提交
368
{
K
Kirill Korotaev 已提交
369
	struct ipc_namespace *ns;
N
Nadia Derbey 已提交
370 371
	struct ipc_ops sem_ops;
	struct ipc_params sem_params;
K
Kirill Korotaev 已提交
372 373

	ns = current->nsproxy->ipc_ns;
L
Linus Torvalds 已提交
374

K
Kirill Korotaev 已提交
375
	if (nsems < 0 || nsems > ns->sc_semmsl)
L
Linus Torvalds 已提交
376
		return -EINVAL;
N
Nadia Derbey 已提交
377

N
Nadia Derbey 已提交
378 379 380 381 382 383 384
	sem_ops.getnew = newary;
	sem_ops.associate = sem_security;
	sem_ops.more_checks = sem_more_checks;

	sem_params.key = key;
	sem_params.flg = semflg;
	sem_params.u.nsems = nsems;
L
Linus Torvalds 已提交
385

N
Nadia Derbey 已提交
386
	return ipcget(ns, &sem_ids(ns), &sem_ops, &sem_params);
L
Linus Torvalds 已提交
387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454
}

/*
 * Determine whether a sequence of semaphore operations would succeed
 * all at once. Return 0 if yes, 1 if need to sleep, else return error code.
 */

static int try_atomic_semop (struct sem_array * sma, struct sembuf * sops,
			     int nsops, struct sem_undo *un, int pid)
{
	int result, sem_op;
	struct sembuf *sop;
	struct sem * curr;

	for (sop = sops; sop < sops + nsops; sop++) {
		curr = sma->sem_base + sop->sem_num;
		sem_op = sop->sem_op;
		result = curr->semval;
  
		if (!sem_op && result)
			goto would_block;

		result += sem_op;
		if (result < 0)
			goto would_block;
		if (result > SEMVMX)
			goto out_of_range;
		if (sop->sem_flg & SEM_UNDO) {
			int undo = un->semadj[sop->sem_num] - sem_op;
			/*
	 		 *	Exceeding the undo range is an error.
			 */
			if (undo < (-SEMAEM - 1) || undo > SEMAEM)
				goto out_of_range;
		}
		curr->semval = result;
	}

	sop--;
	while (sop >= sops) {
		sma->sem_base[sop->sem_num].sempid = pid;
		if (sop->sem_flg & SEM_UNDO)
			un->semadj[sop->sem_num] -= sop->sem_op;
		sop--;
	}
	
	return 0;

out_of_range:
	result = -ERANGE;
	goto undo;

would_block:
	if (sop->sem_flg & IPC_NOWAIT)
		result = -EAGAIN;
	else
		result = 1;

undo:
	sop--;
	while (sop >= sops) {
		sma->sem_base[sop->sem_num].semval -= sop->sem_op;
		sop--;
	}

	return result;
}

455 456 457 458 459
/** wake_up_sem_queue_prepare(q, error): Prepare wake-up
 * @q: queue entry that must be signaled
 * @error: Error value for the signal
 *
 * Prepare the wake-up of the queue entry q.
N
Nick Piggin 已提交
460
 */
461 462
static void wake_up_sem_queue_prepare(struct list_head *pt,
				struct sem_queue *q, int error)
N
Nick Piggin 已提交
463
{
464 465 466 467 468 469 470
	if (list_empty(pt)) {
		/*
		 * Hold preempt off so that we don't get preempted and have the
		 * wakee busy-wait until we're scheduled back on.
		 */
		preempt_disable();
	}
N
Nick Piggin 已提交
471
	q->status = IN_WAKEUP;
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
	q->pid = error;

	list_add_tail(&q->simple_list, pt);
}

/**
 * wake_up_sem_queue_do(pt) - do the actual wake-up
 * @pt: list of tasks to be woken up
 *
 * Do the actual wake-up.
 * The function is called without any locks held, thus the semaphore array
 * could be destroyed already and the tasks can disappear as soon as the
 * status is set to the actual return code.
 */
static void wake_up_sem_queue_do(struct list_head *pt)
{
	struct sem_queue *q, *t;
	int did_something;

	did_something = !list_empty(pt);
	list_for_each_entry_safe(q, t, pt, simple_list) {
		wake_up_process(q->sleeper);
		/* q can disappear immediately after writing q->status. */
		smp_wmb();
		q->status = q->pid;
	}
	if (did_something)
		preempt_enable();
N
Nick Piggin 已提交
500 501
}

502 503 504 505 506 507 508 509 510
static void unlink_queue(struct sem_array *sma, struct sem_queue *q)
{
	list_del(&q->list);
	if (q->nsops == 1)
		list_del(&q->simple_list);
	else
		sma->complex_count--;
}

511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573
/** check_restart(sma, q)
 * @sma: semaphore array
 * @q: the operation that just completed
 *
 * update_queue is O(N^2) when it restarts scanning the whole queue of
 * waiting operations. Therefore this function checks if the restart is
 * really necessary. It is called after a previously waiting operation
 * was completed.
 */
static int check_restart(struct sem_array *sma, struct sem_queue *q)
{
	struct sem *curr;
	struct sem_queue *h;

	/* if the operation didn't modify the array, then no restart */
	if (q->alter == 0)
		return 0;

	/* pending complex operations are too difficult to analyse */
	if (sma->complex_count)
		return 1;

	/* we were a sleeping complex operation. Too difficult */
	if (q->nsops > 1)
		return 1;

	curr = sma->sem_base + q->sops[0].sem_num;

	/* No-one waits on this queue */
	if (list_empty(&curr->sem_pending))
		return 0;

	/* the new semaphore value */
	if (curr->semval) {
		/* It is impossible that someone waits for the new value:
		 * - q is a previously sleeping simple operation that
		 *   altered the array. It must be a decrement, because
		 *   simple increments never sleep.
		 * - The value is not 0, thus wait-for-zero won't proceed.
		 * - If there are older (higher priority) decrements
		 *   in the queue, then they have observed the original
		 *   semval value and couldn't proceed. The operation
		 *   decremented to value - thus they won't proceed either.
		 */
		BUG_ON(q->sops[0].sem_op >= 0);
		return 0;
	}
	/*
	 * semval is 0. Check if there are wait-for-zero semops.
	 * They must be the first entries in the per-semaphore simple queue
	 */
	h = list_first_entry(&curr->sem_pending, struct sem_queue, simple_list);
	BUG_ON(h->nsops != 1);
	BUG_ON(h->sops[0].sem_num != q->sops[0].sem_num);

	/* Yes, there is a wait-for-zero semop. Restart */
	if (h->sops[0].sem_op == 0)
		return 1;

	/* Again - no-one is waiting for the new value. */
	return 0;
}

574 575 576 577 578

/**
 * update_queue(sma, semnum): Look for tasks that can be completed.
 * @sma: semaphore array.
 * @semnum: semaphore that was modified.
579
 * @pt: list head for the tasks that must be woken up.
580 581 582 583
 *
 * update_queue must be called after a semaphore in a semaphore array
 * was modified. If multiple semaphore were modified, then @semnum
 * must be set to -1.
584 585 586
 * The tasks that must be woken up are added to @pt. The return code
 * is stored in q->pid.
 * The function return 1 if at least one semop was completed successfully.
L
Linus Torvalds 已提交
587
 */
588
static int update_queue(struct sem_array *sma, int semnum, struct list_head *pt)
L
Linus Torvalds 已提交
589
{
590 591 592 593
	struct sem_queue *q;
	struct list_head *walk;
	struct list_head *pending_list;
	int offset;
594
	int semop_completed = 0;
595 596 597 598 599 600 601 602 603 604 605 606 607 608 609

	/* if there are complex operations around, then knowing the semaphore
	 * that was modified doesn't help us. Assume that multiple semaphores
	 * were modified.
	 */
	if (sma->complex_count)
		semnum = -1;

	if (semnum == -1) {
		pending_list = &sma->sem_pending;
		offset = offsetof(struct sem_queue, list);
	} else {
		pending_list = &sma->sem_base[semnum].sem_pending;
		offset = offsetof(struct sem_queue, simple_list);
	}
N
Nick Piggin 已提交
610 611

again:
612 613
	walk = pending_list->next;
	while (walk != pending_list) {
614
		int error, restart;
615 616 617

		q = (struct sem_queue *)((char *)walk - offset);
		walk = walk->next;
L
Linus Torvalds 已提交
618

619 620 621 622 623 624 625 626 627 628 629
		/* If we are scanning the single sop, per-semaphore list of
		 * one semaphore and that semaphore is 0, then it is not
		 * necessary to scan the "alter" entries: simple increments
		 * that affect only one entry succeed immediately and cannot
		 * be in the  per semaphore pending queue, and decrements
		 * cannot be successful if the value is already 0.
		 */
		if (semnum != -1 && sma->sem_base[semnum].semval == 0 &&
				q->alter)
			break;

L
Linus Torvalds 已提交
630 631 632 633
		error = try_atomic_semop(sma, q->sops, q->nsops,
					 q->undo, q->pid);

		/* Does q->sleeper still need to sleep? */
N
Nick Piggin 已提交
634 635 636
		if (error > 0)
			continue;

637
		unlink_queue(sma, q);
N
Nick Piggin 已提交
638

639
		if (error) {
640
			restart = 0;
641 642
		} else {
			semop_completed = 1;
643
			restart = check_restart(sma, q);
644
		}
645

646
		wake_up_sem_queue_prepare(pt, q, error);
647
		if (restart)
N
Nick Piggin 已提交
648
			goto again;
L
Linus Torvalds 已提交
649
	}
650
	return semop_completed;
L
Linus Torvalds 已提交
651 652
}

653 654
/**
 * do_smart_update(sma, sops, nsops, otime, pt) - optimized update_queue
655 656 657
 * @sma: semaphore array
 * @sops: operations that were performed
 * @nsops: number of operations
658 659
 * @otime: force setting otime
 * @pt: list head of the tasks that must be woken up.
660 661 662
 *
 * do_smart_update() does the required called to update_queue, based on the
 * actual changes that were performed on the semaphore array.
663 664 665
 * Note that the function does not do the actual wake-up: the caller is
 * responsible for calling wake_up_sem_queue_do(@pt).
 * It is safe to perform this call after dropping all locks.
666
 */
667 668
static void do_smart_update(struct sem_array *sma, struct sembuf *sops, int nsops,
			int otime, struct list_head *pt)
669 670 671 672
{
	int i;

	if (sma->complex_count || sops == NULL) {
673 674 675
		if (update_queue(sma, -1, pt))
			otime = 1;
		goto done;
676 677 678 679 680 681
	}

	for (i = 0; i < nsops; i++) {
		if (sops[i].sem_op > 0 ||
			(sops[i].sem_op < 0 &&
				sma->sem_base[sops[i].sem_num].semval == 0))
682 683
			if (update_queue(sma, sops[i].sem_num, pt))
				otime = 1;
684
	}
685 686 687
done:
	if (otime)
		sma->sem_otime = get_seconds();
688 689 690
}


L
Linus Torvalds 已提交
691 692 693 694 695 696 697 698 699 700 701 702 703 704 705
/* The following counts are associated to each semaphore:
 *   semncnt        number of tasks waiting on semval being nonzero
 *   semzcnt        number of tasks waiting on semval being zero
 * This model assumes that a task waits on exactly one semaphore.
 * Since semaphore operations are to be performed atomically, tasks actually
 * wait on a whole sequence of semaphores simultaneously.
 * The counts we return here are a rough approximation, but still
 * warrant that semncnt+semzcnt>0 if the task is on the pending queue.
 */
static int count_semncnt (struct sem_array * sma, ushort semnum)
{
	int semncnt;
	struct sem_queue * q;

	semncnt = 0;
706
	list_for_each_entry(q, &sma->sem_pending, list) {
L
Linus Torvalds 已提交
707 708 709 710 711 712 713 714 715 716 717
		struct sembuf * sops = q->sops;
		int nsops = q->nsops;
		int i;
		for (i = 0; i < nsops; i++)
			if (sops[i].sem_num == semnum
			    && (sops[i].sem_op < 0)
			    && !(sops[i].sem_flg & IPC_NOWAIT))
				semncnt++;
	}
	return semncnt;
}
718

L
Linus Torvalds 已提交
719 720 721 722 723 724
static int count_semzcnt (struct sem_array * sma, ushort semnum)
{
	int semzcnt;
	struct sem_queue * q;

	semzcnt = 0;
725
	list_for_each_entry(q, &sma->sem_pending, list) {
L
Linus Torvalds 已提交
726 727 728 729 730 731 732 733 734 735 736 737
		struct sembuf * sops = q->sops;
		int nsops = q->nsops;
		int i;
		for (i = 0; i < nsops; i++)
			if (sops[i].sem_num == semnum
			    && (sops[i].sem_op == 0)
			    && !(sops[i].sem_flg & IPC_NOWAIT))
				semzcnt++;
	}
	return semzcnt;
}

N
Nadia Derbey 已提交
738 739 740
/* Free a semaphore set. freeary() is called with sem_ids.rw_mutex locked
 * as a writer and the spinlock for this semaphore set hold. sem_ids.rw_mutex
 * remains locked on exit.
L
Linus Torvalds 已提交
741
 */
742
static void freeary(struct ipc_namespace *ns, struct kern_ipc_perm *ipcp)
L
Linus Torvalds 已提交
743
{
744 745
	struct sem_undo *un, *tu;
	struct sem_queue *q, *tq;
746
	struct sem_array *sma = container_of(ipcp, struct sem_array, sem_perm);
747
	struct list_head tasks;
L
Linus Torvalds 已提交
748

749
	/* Free the existing undo structures for this semaphore set.  */
750
	assert_spin_locked(&sma->sem_perm.lock);
751 752 753
	list_for_each_entry_safe(un, tu, &sma->list_id, list_id) {
		list_del(&un->list_id);
		spin_lock(&un->ulp->lock);
L
Linus Torvalds 已提交
754
		un->semid = -1;
755 756
		list_del_rcu(&un->list_proc);
		spin_unlock(&un->ulp->lock);
757
		kfree_rcu(un, rcu);
758
	}
L
Linus Torvalds 已提交
759 760

	/* Wake up all pending processes and let them fail with EIDRM. */
761
	INIT_LIST_HEAD(&tasks);
762
	list_for_each_entry_safe(q, tq, &sma->sem_pending, list) {
763
		unlink_queue(sma, q);
764
		wake_up_sem_queue_prepare(&tasks, q, -EIDRM);
L
Linus Torvalds 已提交
765 766
	}

N
Nadia Derbey 已提交
767 768
	/* Remove the semaphore set from the IDR */
	sem_rmid(ns, sma);
L
Linus Torvalds 已提交
769 770
	sem_unlock(sma);

771
	wake_up_sem_queue_do(&tasks);
K
Kirill Korotaev 已提交
772
	ns->used_sems -= sma->sem_nsems;
L
Linus Torvalds 已提交
773 774 775 776 777 778 779 780 781 782 783 784 785
	security_sem_free(sma);
	ipc_rcu_putref(sma);
}

static unsigned long copy_semid_to_user(void __user *buf, struct semid64_ds *in, int version)
{
	switch(version) {
	case IPC_64:
		return copy_to_user(buf, in, sizeof(*in));
	case IPC_OLD:
	    {
		struct semid_ds out;

786 787
		memset(&out, 0, sizeof(out));

L
Linus Torvalds 已提交
788 789 790 791 792 793 794 795 796 797 798 799 800
		ipc64_perm_to_ipc_perm(&in->sem_perm, &out.sem_perm);

		out.sem_otime	= in->sem_otime;
		out.sem_ctime	= in->sem_ctime;
		out.sem_nsems	= in->sem_nsems;

		return copy_to_user(buf, &out, sizeof(out));
	    }
	default:
		return -EINVAL;
	}
}

801
static int semctl_nolock(struct ipc_namespace *ns, int semid,
802
			 int cmd, int version, void __user *p)
L
Linus Torvalds 已提交
803
{
804
	int err;
L
Linus Torvalds 已提交
805 806 807 808 809 810 811 812 813 814 815 816 817 818
	struct sem_array *sma;

	switch(cmd) {
	case IPC_INFO:
	case SEM_INFO:
	{
		struct seminfo seminfo;
		int max_id;

		err = security_sem_semctl(NULL, cmd);
		if (err)
			return err;
		
		memset(&seminfo,0,sizeof(seminfo));
K
Kirill Korotaev 已提交
819 820 821 822
		seminfo.semmni = ns->sc_semmni;
		seminfo.semmns = ns->sc_semmns;
		seminfo.semmsl = ns->sc_semmsl;
		seminfo.semopm = ns->sc_semopm;
L
Linus Torvalds 已提交
823 824 825 826
		seminfo.semvmx = SEMVMX;
		seminfo.semmnu = SEMMNU;
		seminfo.semmap = SEMMAP;
		seminfo.semume = SEMUME;
N
Nadia Derbey 已提交
827
		down_read(&sem_ids(ns).rw_mutex);
L
Linus Torvalds 已提交
828
		if (cmd == SEM_INFO) {
K
Kirill Korotaev 已提交
829 830
			seminfo.semusz = sem_ids(ns).in_use;
			seminfo.semaem = ns->used_sems;
L
Linus Torvalds 已提交
831 832 833 834
		} else {
			seminfo.semusz = SEMUSZ;
			seminfo.semaem = SEMAEM;
		}
N
Nadia Derbey 已提交
835
		max_id = ipc_get_maxid(&sem_ids(ns));
N
Nadia Derbey 已提交
836
		up_read(&sem_ids(ns).rw_mutex);
837
		if (copy_to_user(p, &seminfo, sizeof(struct seminfo))) 
L
Linus Torvalds 已提交
838 839 840
			return -EFAULT;
		return (max_id < 0) ? 0: max_id;
	}
841
	case IPC_STAT:
L
Linus Torvalds 已提交
842 843 844 845 846
	case SEM_STAT:
	{
		struct semid64_ds tbuf;
		int id;

847 848 849 850 851 852 853 854 855 856 857
		if (cmd == SEM_STAT) {
			sma = sem_lock(ns, semid);
			if (IS_ERR(sma))
				return PTR_ERR(sma);
			id = sma->sem_perm.id;
		} else {
			sma = sem_lock_check(ns, semid);
			if (IS_ERR(sma))
				return PTR_ERR(sma);
			id = 0;
		}
L
Linus Torvalds 已提交
858 859

		err = -EACCES;
860
		if (ipcperms(ns, &sma->sem_perm, S_IRUGO))
L
Linus Torvalds 已提交
861 862 863 864 865 866
			goto out_unlock;

		err = security_sem_semctl(sma, cmd);
		if (err)
			goto out_unlock;

867 868
		memset(&tbuf, 0, sizeof(tbuf));

L
Linus Torvalds 已提交
869 870 871 872 873
		kernel_to_ipc64_perm(&sma->sem_perm, &tbuf.sem_perm);
		tbuf.sem_otime  = sma->sem_otime;
		tbuf.sem_ctime  = sma->sem_ctime;
		tbuf.sem_nsems  = sma->sem_nsems;
		sem_unlock(sma);
874
		if (copy_semid_to_user(p, &tbuf, version))
L
Linus Torvalds 已提交
875 876 877 878 879 880 881 882 883 884 885
			return -EFAULT;
		return id;
	}
	default:
		return -EINVAL;
	}
out_unlock:
	sem_unlock(sma);
	return err;
}

886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944
static int semctl_setval(struct ipc_namespace *ns, int semid, int semnum,
		unsigned long arg)
{
	struct sem_undo *un;
	struct sem_array *sma;
	struct sem* curr;
	int err;
	int nsems;
	struct list_head tasks;
	int val;
#if defined(CONFIG_64BIT) && defined(__BIG_ENDIAN)
	/* big-endian 64bit */
	val = arg >> 32;
#else
	/* 32bit or little-endian 64bit */
	val = arg;
#endif

	sma = sem_lock_check(ns, semid);
	if (IS_ERR(sma))
		return PTR_ERR(sma);

	INIT_LIST_HEAD(&tasks);
	nsems = sma->sem_nsems;

	err = -EACCES;
	if (ipcperms(ns, &sma->sem_perm, S_IWUGO))
		goto out_unlock;

	err = security_sem_semctl(sma, SETVAL);
	if (err)
		goto out_unlock;

	err = -EINVAL;
	if(semnum < 0 || semnum >= nsems)
		goto out_unlock;

	curr = &sma->sem_base[semnum];

	err = -ERANGE;
	if (val > SEMVMX || val < 0)
		goto out_unlock;

	assert_spin_locked(&sma->sem_perm.lock);
	list_for_each_entry(un, &sma->list_id, list_id)
		un->semadj[semnum] = 0;

	curr->semval = val;
	curr->sempid = task_tgid_vnr(current);
	sma->sem_ctime = get_seconds();
	/* maybe some queued-up processes were waiting for this */
	do_smart_update(sma, NULL, 0, 0, &tasks);
	err = 0;
out_unlock:
	sem_unlock(sma);
	wake_up_sem_queue_do(&tasks);
	return err;
}

K
Kirill Korotaev 已提交
945
static int semctl_main(struct ipc_namespace *ns, int semid, int semnum,
946
		int cmd, void __user *p)
L
Linus Torvalds 已提交
947 948 949 950 951 952 953
{
	struct sem_array *sma;
	struct sem* curr;
	int err;
	ushort fast_sem_io[SEMMSL_FAST];
	ushort* sem_io = fast_sem_io;
	int nsems;
954
	struct list_head tasks;
L
Linus Torvalds 已提交
955

956 957 958
	sma = sem_lock_check(ns, semid);
	if (IS_ERR(sma))
		return PTR_ERR(sma);
L
Linus Torvalds 已提交
959

960
	INIT_LIST_HEAD(&tasks);
L
Linus Torvalds 已提交
961 962 963
	nsems = sma->sem_nsems;

	err = -EACCES;
964
	if (ipcperms(ns, &sma->sem_perm,
965
			cmd == SETALL ? S_IWUGO : S_IRUGO))
L
Linus Torvalds 已提交
966 967 968 969 970 971 972 973 974 975
		goto out_unlock;

	err = security_sem_semctl(sma, cmd);
	if (err)
		goto out_unlock;

	err = -EACCES;
	switch (cmd) {
	case GETALL:
	{
976
		ushort __user *array = p;
L
Linus Torvalds 已提交
977 978 979
		int i;

		if(nsems > SEMMSL_FAST) {
980
			sem_getref_and_unlock(sma);
L
Linus Torvalds 已提交
981 982 983

			sem_io = ipc_alloc(sizeof(ushort)*nsems);
			if(sem_io == NULL) {
984
				sem_putref(sma);
L
Linus Torvalds 已提交
985 986 987
				return -ENOMEM;
			}

988
			sem_lock_and_putref(sma);
L
Linus Torvalds 已提交
989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008
			if (sma->sem_perm.deleted) {
				sem_unlock(sma);
				err = -EIDRM;
				goto out_free;
			}
		}

		for (i = 0; i < sma->sem_nsems; i++)
			sem_io[i] = sma->sem_base[i].semval;
		sem_unlock(sma);
		err = 0;
		if(copy_to_user(array, sem_io, nsems*sizeof(ushort)))
			err = -EFAULT;
		goto out_free;
	}
	case SETALL:
	{
		int i;
		struct sem_undo *un;

1009
		sem_getref_and_unlock(sma);
L
Linus Torvalds 已提交
1010 1011 1012 1013

		if(nsems > SEMMSL_FAST) {
			sem_io = ipc_alloc(sizeof(ushort)*nsems);
			if(sem_io == NULL) {
1014
				sem_putref(sma);
L
Linus Torvalds 已提交
1015 1016 1017 1018
				return -ENOMEM;
			}
		}

1019
		if (copy_from_user (sem_io, p, nsems*sizeof(ushort))) {
1020
			sem_putref(sma);
L
Linus Torvalds 已提交
1021 1022 1023 1024 1025 1026
			err = -EFAULT;
			goto out_free;
		}

		for (i = 0; i < nsems; i++) {
			if (sem_io[i] > SEMVMX) {
1027
				sem_putref(sma);
L
Linus Torvalds 已提交
1028 1029 1030 1031
				err = -ERANGE;
				goto out_free;
			}
		}
1032
		sem_lock_and_putref(sma);
L
Linus Torvalds 已提交
1033 1034 1035 1036 1037 1038 1039 1040
		if (sma->sem_perm.deleted) {
			sem_unlock(sma);
			err = -EIDRM;
			goto out_free;
		}

		for (i = 0; i < nsems; i++)
			sma->sem_base[i].semval = sem_io[i];
1041 1042 1043

		assert_spin_locked(&sma->sem_perm.lock);
		list_for_each_entry(un, &sma->list_id, list_id) {
L
Linus Torvalds 已提交
1044 1045
			for (i = 0; i < nsems; i++)
				un->semadj[i] = 0;
1046
		}
L
Linus Torvalds 已提交
1047 1048
		sma->sem_ctime = get_seconds();
		/* maybe some queued-up processes were waiting for this */
1049
		do_smart_update(sma, NULL, 0, 0, &tasks);
L
Linus Torvalds 已提交
1050 1051 1052
		err = 0;
		goto out_unlock;
	}
1053
	/* GETVAL, GETPID, GETNCTN, GETZCNT: fall-through */
L
Linus Torvalds 已提交
1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076
	}
	err = -EINVAL;
	if(semnum < 0 || semnum >= nsems)
		goto out_unlock;

	curr = &sma->sem_base[semnum];

	switch (cmd) {
	case GETVAL:
		err = curr->semval;
		goto out_unlock;
	case GETPID:
		err = curr->sempid;
		goto out_unlock;
	case GETNCNT:
		err = count_semncnt(sma,semnum);
		goto out_unlock;
	case GETZCNT:
		err = count_semzcnt(sma,semnum);
		goto out_unlock;
	}
out_unlock:
	sem_unlock(sma);
1077 1078
	wake_up_sem_queue_do(&tasks);

L
Linus Torvalds 已提交
1079 1080 1081 1082 1083 1084
out_free:
	if(sem_io != fast_sem_io)
		ipc_free(sem_io, sizeof(ushort)*nsems);
	return err;
}

1085 1086
static inline unsigned long
copy_semid_from_user(struct semid64_ds *out, void __user *buf, int version)
L
Linus Torvalds 已提交
1087 1088 1089
{
	switch(version) {
	case IPC_64:
1090
		if (copy_from_user(out, buf, sizeof(*out)))
L
Linus Torvalds 已提交
1091 1092 1093 1094 1095 1096 1097 1098 1099
			return -EFAULT;
		return 0;
	case IPC_OLD:
	    {
		struct semid_ds tbuf_old;

		if(copy_from_user(&tbuf_old, buf, sizeof(tbuf_old)))
			return -EFAULT;

1100 1101 1102
		out->sem_perm.uid	= tbuf_old.sem_perm.uid;
		out->sem_perm.gid	= tbuf_old.sem_perm.gid;
		out->sem_perm.mode	= tbuf_old.sem_perm.mode;
L
Linus Torvalds 已提交
1103 1104 1105 1106 1107 1108 1109 1110

		return 0;
	    }
	default:
		return -EINVAL;
	}
}

1111 1112 1113 1114 1115
/*
 * This function handles some semctl commands which require the rw_mutex
 * to be held in write mode.
 * NOTE: no locks must be held, the rw_mutex is taken inside this function.
 */
1116
static int semctl_down(struct ipc_namespace *ns, int semid,
1117
		       int cmd, int version, void __user *p)
L
Linus Torvalds 已提交
1118 1119 1120
{
	struct sem_array *sma;
	int err;
1121
	struct semid64_ds semid64;
L
Linus Torvalds 已提交
1122 1123 1124
	struct kern_ipc_perm *ipcp;

	if(cmd == IPC_SET) {
1125
		if (copy_semid_from_user(&semid64, p, version))
L
Linus Torvalds 已提交
1126 1127
			return -EFAULT;
	}
S
Steve Grubb 已提交
1128

1129 1130
	ipcp = ipcctl_pre_down(ns, &sem_ids(ns), semid, cmd,
			       &semid64.sem_perm, 0);
1131 1132
	if (IS_ERR(ipcp))
		return PTR_ERR(ipcp);
S
Steve Grubb 已提交
1133

1134
	sma = container_of(ipcp, struct sem_array, sem_perm);
L
Linus Torvalds 已提交
1135 1136 1137 1138 1139 1140 1141

	err = security_sem_semctl(sma, cmd);
	if (err)
		goto out_unlock;

	switch(cmd){
	case IPC_RMID:
1142
		freeary(ns, ipcp);
1143
		goto out_up;
L
Linus Torvalds 已提交
1144
	case IPC_SET:
1145 1146 1147
		err = ipc_update_perm(&semid64.sem_perm, ipcp);
		if (err)
			goto out_unlock;
L
Linus Torvalds 已提交
1148 1149 1150 1151 1152 1153 1154 1155
		sma->sem_ctime = get_seconds();
		break;
	default:
		err = -EINVAL;
	}

out_unlock:
	sem_unlock(sma);
1156 1157
out_up:
	up_write(&sem_ids(ns).rw_mutex);
L
Linus Torvalds 已提交
1158 1159 1160
	return err;
}

1161
SYSCALL_DEFINE4(semctl, int, semid, int, semnum, int, cmd, unsigned long, arg)
L
Linus Torvalds 已提交
1162 1163
{
	int version;
K
Kirill Korotaev 已提交
1164
	struct ipc_namespace *ns;
1165
	void __user *p = (void __user *)arg;
L
Linus Torvalds 已提交
1166 1167 1168 1169 1170

	if (semid < 0)
		return -EINVAL;

	version = ipc_parse_version(&cmd);
K
Kirill Korotaev 已提交
1171
	ns = current->nsproxy->ipc_ns;
L
Linus Torvalds 已提交
1172 1173 1174 1175

	switch(cmd) {
	case IPC_INFO:
	case SEM_INFO:
1176
	case IPC_STAT:
L
Linus Torvalds 已提交
1177
	case SEM_STAT:
1178
		return semctl_nolock(ns, semid, cmd, version, p);
L
Linus Torvalds 已提交
1179 1180 1181 1182 1183 1184
	case GETALL:
	case GETVAL:
	case GETPID:
	case GETNCNT:
	case GETZCNT:
	case SETALL:
1185 1186 1187
		return semctl_main(ns, semid, semnum, cmd, p);
	case SETVAL:
		return semctl_setval(ns, semid, semnum, arg);
L
Linus Torvalds 已提交
1188 1189
	case IPC_RMID:
	case IPC_SET:
1190
		return semctl_down(ns, semid, cmd, version, p);
L
Linus Torvalds 已提交
1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212
	default:
		return -EINVAL;
	}
}

/* If the task doesn't already have a undo_list, then allocate one
 * here.  We guarantee there is only one thread using this undo list,
 * and current is THE ONE
 *
 * If this allocation and assignment succeeds, but later
 * portions of this code fail, there is no need to free the sem_undo_list.
 * Just let it stay associated with the task, and it'll be freed later
 * at exit time.
 *
 * This can block, so callers must hold no locks.
 */
static inline int get_undo_list(struct sem_undo_list **undo_listp)
{
	struct sem_undo_list *undo_list;

	undo_list = current->sysvsem.undo_list;
	if (!undo_list) {
1213
		undo_list = kzalloc(sizeof(*undo_list), GFP_KERNEL);
L
Linus Torvalds 已提交
1214 1215
		if (undo_list == NULL)
			return -ENOMEM;
I
Ingo Molnar 已提交
1216
		spin_lock_init(&undo_list->lock);
L
Linus Torvalds 已提交
1217
		atomic_set(&undo_list->refcnt, 1);
1218 1219
		INIT_LIST_HEAD(&undo_list->list_proc);

L
Linus Torvalds 已提交
1220 1221 1222 1223 1224 1225
		current->sysvsem.undo_list = undo_list;
	}
	*undo_listp = undo_list;
	return 0;
}

1226
static struct sem_undo *__lookup_undo(struct sem_undo_list *ulp, int semid)
L
Linus Torvalds 已提交
1227
{
1228
	struct sem_undo *un;
1229

1230 1231 1232
	list_for_each_entry_rcu(un, &ulp->list_proc, list_proc) {
		if (un->semid == semid)
			return un;
L
Linus Torvalds 已提交
1233
	}
1234
	return NULL;
L
Linus Torvalds 已提交
1235 1236
}

1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250
static struct sem_undo *lookup_undo(struct sem_undo_list *ulp, int semid)
{
	struct sem_undo *un;

  	assert_spin_locked(&ulp->lock);

	un = __lookup_undo(ulp, semid);
	if (un) {
		list_del_rcu(&un->list_proc);
		list_add_rcu(&un->list_proc, &ulp->list_proc);
	}
	return un;
}

1251 1252 1253 1254 1255 1256 1257 1258
/**
 * find_alloc_undo - Lookup (and if not present create) undo array
 * @ns: namespace
 * @semid: semaphore array id
 *
 * The function looks up (and if not present creates) the undo structure.
 * The size of the undo structure depends on the size of the semaphore
 * array, thus the alloc path is not that straightforward.
1259 1260
 * Lifetime-rules: sem_undo is rcu-protected, on success, the function
 * performs a rcu_read_lock().
1261 1262
 */
static struct sem_undo *find_alloc_undo(struct ipc_namespace *ns, int semid)
L
Linus Torvalds 已提交
1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273
{
	struct sem_array *sma;
	struct sem_undo_list *ulp;
	struct sem_undo *un, *new;
	int nsems;
	int error;

	error = get_undo_list(&ulp);
	if (error)
		return ERR_PTR(error);

1274
	rcu_read_lock();
1275
	spin_lock(&ulp->lock);
L
Linus Torvalds 已提交
1276
	un = lookup_undo(ulp, semid);
1277
	spin_unlock(&ulp->lock);
L
Linus Torvalds 已提交
1278 1279
	if (likely(un!=NULL))
		goto out;
1280
	rcu_read_unlock();
L
Linus Torvalds 已提交
1281 1282

	/* no undo structure around - allocate one. */
1283
	/* step 1: figure out the size of the semaphore array */
1284 1285
	sma = sem_lock_check(ns, semid);
	if (IS_ERR(sma))
J
Julia Lawall 已提交
1286
		return ERR_CAST(sma);
1287

L
Linus Torvalds 已提交
1288
	nsems = sma->sem_nsems;
1289
	sem_getref_and_unlock(sma);
L
Linus Torvalds 已提交
1290

1291
	/* step 2: allocate new undo structure */
1292
	new = kzalloc(sizeof(struct sem_undo) + sizeof(short)*nsems, GFP_KERNEL);
L
Linus Torvalds 已提交
1293
	if (!new) {
1294
		sem_putref(sma);
L
Linus Torvalds 已提交
1295 1296 1297
		return ERR_PTR(-ENOMEM);
	}

1298
	/* step 3: Acquire the lock on semaphore array */
1299
	sem_lock_and_putref(sma);
L
Linus Torvalds 已提交
1300 1301 1302 1303 1304 1305
	if (sma->sem_perm.deleted) {
		sem_unlock(sma);
		kfree(new);
		un = ERR_PTR(-EIDRM);
		goto out;
	}
1306 1307 1308 1309 1310 1311 1312 1313 1314 1315
	spin_lock(&ulp->lock);

	/*
	 * step 4: check for races: did someone else allocate the undo struct?
	 */
	un = lookup_undo(ulp, semid);
	if (un) {
		kfree(new);
		goto success;
	}
1316 1317
	/* step 5: initialize & link new undo structure */
	new->semadj = (short *) &new[1];
1318
	new->ulp = ulp;
1319 1320
	new->semid = semid;
	assert_spin_locked(&ulp->lock);
1321
	list_add_rcu(&new->list_proc, &ulp->list_proc);
1322 1323
	assert_spin_locked(&sma->sem_perm.lock);
	list_add(&new->list_id, &sma->list_id);
1324
	un = new;
1325

1326
success:
1327
	spin_unlock(&ulp->lock);
1328 1329
	rcu_read_lock();
	sem_unlock(sma);
L
Linus Torvalds 已提交
1330 1331 1332 1333
out:
	return un;
}

1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360

/**
 * get_queue_result - Retrieve the result code from sem_queue
 * @q: Pointer to queue structure
 *
 * Retrieve the return code from the pending queue. If IN_WAKEUP is found in
 * q->status, then we must loop until the value is replaced with the final
 * value: This may happen if a task is woken up by an unrelated event (e.g.
 * signal) and in parallel the task is woken up by another task because it got
 * the requested semaphores.
 *
 * The function can be called with or without holding the semaphore spinlock.
 */
static int get_queue_result(struct sem_queue *q)
{
	int error;

	error = q->status;
	while (unlikely(error == IN_WAKEUP)) {
		cpu_relax();
		error = q->status;
	}

	return error;
}


1361 1362
SYSCALL_DEFINE4(semtimedop, int, semid, struct sembuf __user *, tsops,
		unsigned, nsops, const struct timespec __user *, timeout)
L
Linus Torvalds 已提交
1363 1364 1365 1366 1367 1368
{
	int error = -EINVAL;
	struct sem_array *sma;
	struct sembuf fast_sops[SEMOPM_FAST];
	struct sembuf* sops = fast_sops, *sop;
	struct sem_undo *un;
1369
	int undos = 0, alter = 0, max;
L
Linus Torvalds 已提交
1370 1371
	struct sem_queue queue;
	unsigned long jiffies_left = 0;
K
Kirill Korotaev 已提交
1372
	struct ipc_namespace *ns;
1373
	struct list_head tasks;
K
Kirill Korotaev 已提交
1374 1375

	ns = current->nsproxy->ipc_ns;
L
Linus Torvalds 已提交
1376 1377 1378

	if (nsops < 1 || semid < 0)
		return -EINVAL;
K
Kirill Korotaev 已提交
1379
	if (nsops > ns->sc_semopm)
L
Linus Torvalds 已提交
1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407
		return -E2BIG;
	if(nsops > SEMOPM_FAST) {
		sops = kmalloc(sizeof(*sops)*nsops,GFP_KERNEL);
		if(sops==NULL)
			return -ENOMEM;
	}
	if (copy_from_user (sops, tsops, nsops * sizeof(*tsops))) {
		error=-EFAULT;
		goto out_free;
	}
	if (timeout) {
		struct timespec _timeout;
		if (copy_from_user(&_timeout, timeout, sizeof(*timeout))) {
			error = -EFAULT;
			goto out_free;
		}
		if (_timeout.tv_sec < 0 || _timeout.tv_nsec < 0 ||
			_timeout.tv_nsec >= 1000000000L) {
			error = -EINVAL;
			goto out_free;
		}
		jiffies_left = timespec_to_jiffies(&_timeout);
	}
	max = 0;
	for (sop = sops; sop < sops + nsops; sop++) {
		if (sop->sem_num >= max)
			max = sop->sem_num;
		if (sop->sem_flg & SEM_UNDO)
1408 1409
			undos = 1;
		if (sop->sem_op != 0)
L
Linus Torvalds 已提交
1410 1411 1412 1413
			alter = 1;
	}

	if (undos) {
1414
		un = find_alloc_undo(ns, semid);
L
Linus Torvalds 已提交
1415 1416 1417 1418 1419 1420 1421
		if (IS_ERR(un)) {
			error = PTR_ERR(un);
			goto out_free;
		}
	} else
		un = NULL;

1422 1423
	INIT_LIST_HEAD(&tasks);

1424 1425
	sma = sem_lock_check(ns, semid);
	if (IS_ERR(sma)) {
1426 1427
		if (un)
			rcu_read_unlock();
1428
		error = PTR_ERR(sma);
L
Linus Torvalds 已提交
1429
		goto out_free;
1430 1431
	}

L
Linus Torvalds 已提交
1432
	/*
1433
	 * semid identifiers are not unique - find_alloc_undo may have
L
Linus Torvalds 已提交
1434
	 * allocated an undo structure, it was invalidated by an RMID
1435
	 * and now a new array with received the same id. Check and fail.
L
Lucas De Marchi 已提交
1436
	 * This case can be detected checking un->semid. The existence of
1437
	 * "un" itself is guaranteed by rcu.
L
Linus Torvalds 已提交
1438
	 */
1439
	error = -EIDRM;
1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455
	if (un) {
		if (un->semid == -1) {
			rcu_read_unlock();
			goto out_unlock_free;
		} else {
			/*
			 * rcu lock can be released, "un" cannot disappear:
			 * - sem_lock is acquired, thus IPC_RMID is
			 *   impossible.
			 * - exit_sem is impossible, it always operates on
			 *   current (or a dead task).
			 */

			rcu_read_unlock();
		}
	}
1456

L
Linus Torvalds 已提交
1457 1458 1459 1460 1461
	error = -EFBIG;
	if (max >= sma->sem_nsems)
		goto out_unlock_free;

	error = -EACCES;
1462
	if (ipcperms(ns, &sma->sem_perm, alter ? S_IWUGO : S_IRUGO))
L
Linus Torvalds 已提交
1463 1464 1465 1466 1467 1468
		goto out_unlock_free;

	error = security_sem_semop(sma, sops, nsops, alter);
	if (error)
		goto out_unlock_free;

1469
	error = try_atomic_semop (sma, sops, nsops, un, task_tgid_vnr(current));
L
Linus Torvalds 已提交
1470 1471
	if (error <= 0) {
		if (alter && error == 0)
1472
			do_smart_update(sma, sops, nsops, 1, &tasks);
1473

L
Linus Torvalds 已提交
1474 1475 1476 1477 1478 1479 1480 1481 1482 1483
		goto out_unlock_free;
	}

	/* We need to sleep on this operation, so we put the current
	 * task into the pending queue and go to sleep.
	 */
		
	queue.sops = sops;
	queue.nsops = nsops;
	queue.undo = un;
1484
	queue.pid = task_tgid_vnr(current);
L
Linus Torvalds 已提交
1485 1486
	queue.alter = alter;
	if (alter)
1487
		list_add_tail(&queue.list, &sma->sem_pending);
L
Linus Torvalds 已提交
1488
	else
1489
		list_add(&queue.list, &sma->sem_pending);
L
Linus Torvalds 已提交
1490

1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503
	if (nsops == 1) {
		struct sem *curr;
		curr = &sma->sem_base[sops->sem_num];

		if (alter)
			list_add_tail(&queue.simple_list, &curr->sem_pending);
		else
			list_add(&queue.simple_list, &curr->sem_pending);
	} else {
		INIT_LIST_HEAD(&queue.simple_list);
		sma->complex_count++;
	}

L
Linus Torvalds 已提交
1504 1505
	queue.status = -EINTR;
	queue.sleeper = current;
1506 1507

sleep_again:
L
Linus Torvalds 已提交
1508 1509 1510 1511 1512 1513 1514 1515
	current->state = TASK_INTERRUPTIBLE;
	sem_unlock(sma);

	if (timeout)
		jiffies_left = schedule_timeout(jiffies_left);
	else
		schedule();

1516
	error = get_queue_result(&queue);
L
Linus Torvalds 已提交
1517 1518 1519

	if (error != -EINTR) {
		/* fast path: update_queue already obtained all requested
1520 1521 1522 1523 1524 1525 1526 1527
		 * resources.
		 * Perform a smp_mb(): User space could assume that semop()
		 * is a memory barrier: Without the mb(), the cpu could
		 * speculatively read in user space stale data that was
		 * overwritten by the previous owner of the semaphore.
		 */
		smp_mb();

L
Linus Torvalds 已提交
1528 1529 1530
		goto out_free;
	}

K
Kirill Korotaev 已提交
1531
	sma = sem_lock(ns, semid);
1532 1533 1534 1535 1536 1537 1538 1539 1540

	/*
	 * Wait until it's guaranteed that no wakeup_sem_queue_do() is ongoing.
	 */
	error = get_queue_result(&queue);

	/*
	 * Array removed? If yes, leave without sem_unlock().
	 */
1541
	if (IS_ERR(sma)) {
L
Linus Torvalds 已提交
1542 1543 1544
		goto out_free;
	}

1545

L
Linus Torvalds 已提交
1546
	/*
1547 1548
	 * If queue.status != -EINTR we are woken up by another process.
	 * Leave without unlink_queue(), but with sem_unlock().
L
Linus Torvalds 已提交
1549
	 */
1550

L
Linus Torvalds 已提交
1551 1552 1553 1554 1555 1556 1557 1558 1559
	if (error != -EINTR) {
		goto out_unlock_free;
	}

	/*
	 * If an interrupt occurred we have to clean up the queue
	 */
	if (timeout && jiffies_left == 0)
		error = -EAGAIN;
1560 1561 1562 1563 1564 1565 1566

	/*
	 * If the wakeup was spurious, just retry
	 */
	if (error == -EINTR && !signal_pending(current))
		goto sleep_again;

1567
	unlink_queue(sma, &queue);
L
Linus Torvalds 已提交
1568 1569 1570

out_unlock_free:
	sem_unlock(sma);
1571 1572

	wake_up_sem_queue_do(&tasks);
L
Linus Torvalds 已提交
1573 1574 1575 1576 1577 1578
out_free:
	if(sops != fast_sops)
		kfree(sops);
	return error;
}

1579 1580
SYSCALL_DEFINE3(semop, int, semid, struct sembuf __user *, tsops,
		unsigned, nsops)
L
Linus Torvalds 已提交
1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619
{
	return sys_semtimedop(semid, tsops, nsops, NULL);
}

/* If CLONE_SYSVSEM is set, establish sharing of SEM_UNDO state between
 * parent and child tasks.
 */

int copy_semundo(unsigned long clone_flags, struct task_struct *tsk)
{
	struct sem_undo_list *undo_list;
	int error;

	if (clone_flags & CLONE_SYSVSEM) {
		error = get_undo_list(&undo_list);
		if (error)
			return error;
		atomic_inc(&undo_list->refcnt);
		tsk->sysvsem.undo_list = undo_list;
	} else 
		tsk->sysvsem.undo_list = NULL;

	return 0;
}

/*
 * add semadj values to semaphores, free undo structures.
 * undo structures are not freed when semaphore arrays are destroyed
 * so some of them may be out of date.
 * IMPLEMENTATION NOTE: There is some confusion over whether the
 * set of adjustments that needs to be done should be done in an atomic
 * manner or not. That is, if we are attempting to decrement the semval
 * should we queue up and wait until we can do so legally?
 * The original implementation attempted to do this (queue and wait).
 * The current implementation does not do so. The POSIX standard
 * and SVID should be consulted to determine what behavior is mandated.
 */
void exit_sem(struct task_struct *tsk)
{
1620
	struct sem_undo_list *ulp;
L
Linus Torvalds 已提交
1621

1622 1623
	ulp = tsk->sysvsem.undo_list;
	if (!ulp)
L
Linus Torvalds 已提交
1624
		return;
1625
	tsk->sysvsem.undo_list = NULL;
L
Linus Torvalds 已提交
1626

1627
	if (!atomic_dec_and_test(&ulp->refcnt))
L
Linus Torvalds 已提交
1628 1629
		return;

1630
	for (;;) {
L
Linus Torvalds 已提交
1631
		struct sem_array *sma;
1632
		struct sem_undo *un;
1633
		struct list_head tasks;
1634
		int semid;
1635 1636
		int i;

1637
		rcu_read_lock();
1638 1639
		un = list_entry_rcu(ulp->list_proc.next,
				    struct sem_undo, list_proc);
1640 1641 1642 1643 1644
		if (&un->list_proc == &ulp->list_proc)
			semid = -1;
		 else
			semid = un->semid;
		rcu_read_unlock();
1645

1646 1647
		if (semid == -1)
			break;
L
Linus Torvalds 已提交
1648

1649
		sma = sem_lock_check(tsk->nsproxy->ipc_ns, un->semid);
L
Linus Torvalds 已提交
1650

1651 1652 1653
		/* exit_sem raced with IPC_RMID, nothing to do */
		if (IS_ERR(sma))
			continue;
L
Linus Torvalds 已提交
1654

1655
		un = __lookup_undo(ulp, semid);
1656 1657 1658 1659 1660 1661 1662 1663 1664
		if (un == NULL) {
			/* exit_sem raced with IPC_RMID+semget() that created
			 * exactly the same semid. Nothing to do.
			 */
			sem_unlock(sma);
			continue;
		}

		/* remove un from the linked lists */
1665 1666 1667
		assert_spin_locked(&sma->sem_perm.lock);
		list_del(&un->list_id);

1668 1669 1670 1671
		spin_lock(&ulp->lock);
		list_del_rcu(&un->list_proc);
		spin_unlock(&ulp->lock);

1672 1673
		/* perform adjustments registered in un */
		for (i = 0; i < sma->sem_nsems; i++) {
I
Ingo Molnar 已提交
1674
			struct sem * semaphore = &sma->sem_base[i];
1675 1676
			if (un->semadj[i]) {
				semaphore->semval += un->semadj[i];
L
Linus Torvalds 已提交
1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689
				/*
				 * Range checks of the new semaphore value,
				 * not defined by sus:
				 * - Some unices ignore the undo entirely
				 *   (e.g. HP UX 11i 11.22, Tru64 V5.1)
				 * - some cap the value (e.g. FreeBSD caps
				 *   at 0, but doesn't enforce SEMVMX)
				 *
				 * Linux caps the semaphore value, both at 0
				 * and at SEMVMX.
				 *
				 * 	Manfred <manfred@colorfullife.com>
				 */
I
Ingo Molnar 已提交
1690 1691 1692 1693
				if (semaphore->semval < 0)
					semaphore->semval = 0;
				if (semaphore->semval > SEMVMX)
					semaphore->semval = SEMVMX;
1694
				semaphore->sempid = task_tgid_vnr(current);
L
Linus Torvalds 已提交
1695 1696 1697
			}
		}
		/* maybe some queued-up processes were waiting for this */
1698 1699
		INIT_LIST_HEAD(&tasks);
		do_smart_update(sma, NULL, 0, 1, &tasks);
L
Linus Torvalds 已提交
1700
		sem_unlock(sma);
1701
		wake_up_sem_queue_do(&tasks);
1702

1703
		kfree_rcu(un, rcu);
L
Linus Torvalds 已提交
1704
	}
1705
	kfree(ulp);
L
Linus Torvalds 已提交
1706 1707 1708
}

#ifdef CONFIG_PROC_FS
1709
static int sysvipc_sem_proc_show(struct seq_file *s, void *it)
L
Linus Torvalds 已提交
1710
{
1711
	struct user_namespace *user_ns = seq_user_ns(s);
1712 1713 1714
	struct sem_array *sma = it;

	return seq_printf(s,
1715
			  "%10d %10d  %4o %10u %5u %5u %5u %5u %10lu %10lu\n",
1716
			  sma->sem_perm.key,
N
Nadia Derbey 已提交
1717
			  sma->sem_perm.id,
1718 1719
			  sma->sem_perm.mode,
			  sma->sem_nsems,
1720 1721 1722 1723
			  from_kuid_munged(user_ns, sma->sem_perm.uid),
			  from_kgid_munged(user_ns, sma->sem_perm.gid),
			  from_kuid_munged(user_ns, sma->sem_perm.cuid),
			  from_kgid_munged(user_ns, sma->sem_perm.cgid),
1724 1725
			  sma->sem_otime,
			  sma->sem_ctime);
L
Linus Torvalds 已提交
1726 1727
}
#endif