sem.c 61.5 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
L
Linus Torvalds 已提交
2 3 4 5 6 7 8 9
/*
 * 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
10
 * (c) 1999 Manfred Spraul <manfred@colorfullife.com>
L
Linus Torvalds 已提交
11
 * Enforced range limit on SEM_UNDO
A
Alan Cox 已提交
12
 * (c) 2001 Red Hat Inc
L
Linus Torvalds 已提交
13 14
 * Lockless wakeup
 * (c) 2003 Manfred Spraul <manfred@colorfullife.com>
D
Davidlohr Bueso 已提交
15
 * (c) 2016 Davidlohr Bueso <dave@stgolabs.net>
16 17
 * Further wakeup optimizations, documentation
 * (c) 2010 Manfred Spraul <manfred@colorfullife.com>
S
Steve Grubb 已提交
18 19 20
 *
 * support for audit of ipc object properties and permission changes
 * Dustin Kirkland <dustin.kirkland@us.ibm.com>
K
Kirill Korotaev 已提交
21 22 23 24
 *
 * namespaces support
 * OpenVZ, SWsoft Inc.
 * Pavel Emelianov <xemul@openvz.org>
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
 *
 * 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.
52
 * - semncnt and semzcnt are calculated on demand in count_semcnt()
53 54 55 56 57
 * - 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
D
Davidlohr Bueso 已提交
58
 *   dropping all locks. (see wake_up_sem_queue_prepare())
59 60 61 62 63 64 65 66 67 68 69 70
 * - 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).
 * - 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
71 72
 */

73
#include <linux/compat.h>
L
Linus Torvalds 已提交
74 75 76 77 78 79 80 81
#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>
82
#include <linux/capability.h>
83
#include <linux/seq_file.h>
N
Nadia Derbey 已提交
84
#include <linux/rwsem.h>
K
Kirill Korotaev 已提交
85
#include <linux/nsproxy.h>
86
#include <linux/ipc_namespace.h>
87
#include <linux/sched/wake_q.h>
88
#include <linux/nospec.h>
N
NeilBrown 已提交
89
#include <linux/rhashtable.h>
I
Ingo Molnar 已提交
90

P
Paul McQuade 已提交
91
#include <linux/uaccess.h>
L
Linus Torvalds 已提交
92 93
#include "util.h"

94 95 96 97 98 99 100 101 102 103
/* One semaphore structure for each semaphore in the system. */
struct sem {
	int	semval;		/* current value */
	/*
	 * PID of the process that last modified the semaphore. For
	 * Linux, specifically these are:
	 *  - semop
	 *  - semctl, via SETVAL and SETALL.
	 *  - at task exit when performing undo adjustments (see exit_sem).
	 */
104
	struct pid *sempid;
105 106 107 108 109
	spinlock_t	lock;	/* spinlock for fine-grained semtimedop */
	struct list_head pending_alter; /* pending single-sop operations */
					/* that alter the semaphore */
	struct list_head pending_const; /* pending single-sop operations */
					/* that do not alter the semaphore*/
110
	time64_t	 sem_otime;	/* candidate for sem_otime */
111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127
} ____cacheline_aligned_in_smp;

/* One sem_array data structure for each set of semaphores in the system. */
struct sem_array {
	struct kern_ipc_perm	sem_perm;	/* permissions .. see ipc.h */
	time64_t		sem_ctime;	/* create/last semctl() time */
	struct list_head	pending_alter;	/* pending operations */
						/* that alter the array */
	struct list_head	pending_const;	/* pending complex operations */
						/* that do not alter semvals */
	struct list_head	list_id;	/* undo requests on this array */
	int			sem_nsems;	/* no. of semaphores in array */
	int			complex_count;	/* pending complex operations */
	unsigned int		use_global_lock;/* >0: global lock required */

	struct sem		sems[];
} __randomize_layout;
128 129 130 131 132 133

/* One queue for each sleeping process in the system. */
struct sem_queue {
	struct list_head	list;	 /* queue of pending operations */
	struct task_struct	*sleeper; /* this process */
	struct sem_undo		*undo;	 /* undo structure */
134
	struct pid		*pid;	 /* process id of requesting process */
135 136
	int			status;	 /* completion status of operation */
	struct sembuf		*sops;	 /* array of pending operations */
137
	struct sembuf		*blocking; /* the operation that blocked */
138
	int			nsops;	 /* number of operations */
139 140
	bool			alter;	 /* does *sops alter the array? */
	bool                    dupsop;	 /* sops on more than one sem_num */
141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162
};

/* 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 {
163
	refcount_t		refcnt;
164 165 166 167 168
	spinlock_t		lock;
	struct list_head	list_proc;
};


169
#define sem_ids(ns)	((ns)->ids[IPC_SEM_IDS])
K
Kirill Korotaev 已提交
170

N
Nadia Derbey 已提交
171
static int newary(struct ipc_namespace *, struct ipc_params *);
172
static void freeary(struct ipc_namespace *, struct kern_ipc_perm *);
L
Linus Torvalds 已提交
173
#ifdef CONFIG_PROC_FS
174
static int sysvipc_sem_proc_show(struct seq_file *s, void *it);
L
Linus Torvalds 已提交
175 176 177 178 179
#endif

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

M
Manfred Spraul 已提交
180 181 182 183 184 185 186
/*
 * Switching from the mode suitable for simple ops
 * to the mode for complex ops is costly. Therefore:
 * use some hysteresis
 */
#define USE_GLOBAL_LOCK_HYSTERESIS	10

L
Linus Torvalds 已提交
187
/*
188
 * Locking:
189
 * a) global sem_lock() for read/write
L
Linus Torvalds 已提交
190
 *	sem_undo.id_next,
191
 *	sem_array.complex_count,
192 193
 *	sem_array.pending{_alter,_const},
 *	sem_array.sem_undo
P
Paul McQuade 已提交
194
 *
195
 * b) global or semaphore sem_lock() for read/write:
196
 *	sem_array.sems[i].pending_{const,alter}:
197 198 199 200 201
 *
 * c) special:
 *	sem_undo_list.list_proc:
 *	* undo_list->lock for write
 *	* rcu for read
M
Manfred Spraul 已提交
202 203 204 205 206 207 208 209 210 211 212 213 214 215 216
 *	use_global_lock:
 *	* global sem_lock() for write
 *	* either local or global sem_lock() for read.
 *
 * Memory ordering:
 * Most ordering is enforced by using spin_lock() and spin_unlock().
 * The special case is use_global_lock:
 * Setting it from non-zero to 0 is a RELEASE, this is ensured by
 * using smp_store_release().
 * Testing if it is non-zero is an ACQUIRE, this is ensured by using
 * smp_load_acquire().
 * Setting it from 0 to non-zero must be ordered with regards to
 * this smp_load_acquire(), this is guaranteed because the smp_load_acquire()
 * is inside a spin_lock() and after a write from 0 to non-zero a
 * spin_lock()+spin_unlock() is done.
L
Linus Torvalds 已提交
217 218
 */

K
Kirill Korotaev 已提交
219 220 221 222 223
#define sc_semmsl	sem_ctls[0]
#define sc_semmns	sem_ctls[1]
#define sc_semopm	sem_ctls[2]
#define sc_semmni	sem_ctls[3]

224
int sem_init_ns(struct ipc_namespace *ns)
K
Kirill Korotaev 已提交
225 226 227 228 229 230
{
	ns->sc_semmsl = SEMMSL;
	ns->sc_semmns = SEMMNS;
	ns->sc_semopm = SEMOPM;
	ns->sc_semmni = SEMMNI;
	ns->used_sems = 0;
231
	return ipc_init_ids(&ns->ids[IPC_SEM_IDS]);
K
Kirill Korotaev 已提交
232 233
}

234
#ifdef CONFIG_IPC_NS
K
Kirill Korotaev 已提交
235 236
void sem_exit_ns(struct ipc_namespace *ns)
{
237
	free_ipcs(ns, &sem_ids(ns), freeary);
S
Serge E. Hallyn 已提交
238
	idr_destroy(&ns->ids[IPC_SEM_IDS].ipcs_idr);
239
	rhashtable_destroy(&ns->ids[IPC_SEM_IDS].key_ht);
K
Kirill Korotaev 已提交
240
}
241
#endif
L
Linus Torvalds 已提交
242

243
int __init sem_init(void)
L
Linus Torvalds 已提交
244
{
245 246
	const int err = sem_init_ns(&init_ipc_ns);

247 248
	ipc_init_proc_interface("sysvipc/sem",
				"       key      semid perms      nsems   uid   gid  cuid  cgid      otime      ctime\n",
K
Kirill Korotaev 已提交
249
				IPC_SEM_IDS, sysvipc_sem_proc_show);
250
	return err;
L
Linus Torvalds 已提交
251 252
}

253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273
/**
 * unmerge_queues - unmerge queues, if possible.
 * @sma: semaphore array
 *
 * The function unmerges the wait queues if complex_count is 0.
 * It must be called prior to dropping the global semaphore array lock.
 */
static void unmerge_queues(struct sem_array *sma)
{
	struct sem_queue *q, *tq;

	/* complex operations still around? */
	if (sma->complex_count)
		return;
	/*
	 * We will switch back to simple mode.
	 * Move all pending operation back into the per-semaphore
	 * queues.
	 */
	list_for_each_entry_safe(q, tq, &sma->pending_alter, list) {
		struct sem *curr;
274
		curr = &sma->sems[q->sops[0].sem_num];
275 276 277 278 279 280 281

		list_add_tail(&q->list, &curr->pending_alter);
	}
	INIT_LIST_HEAD(&sma->pending_alter);
}

/**
D
Davidlohr Bueso 已提交
282
 * merge_queues - merge single semop queues into global queue
283 284 285 286 287 288 289 290 291 292 293
 * @sma: semaphore array
 *
 * This function merges all per-semaphore queues into the global queue.
 * It is necessary to achieve FIFO ordering for the pending single-sop
 * operations when a multi-semop operation must sleep.
 * Only the alter operations must be moved, the const operations can stay.
 */
static void merge_queues(struct sem_array *sma)
{
	int i;
	for (i = 0; i < sma->sem_nsems; i++) {
294
		struct sem *sem = &sma->sems[i];
295 296 297 298 299

		list_splice_init(&sem->pending_alter, &sma->pending_alter);
	}
}

D
Davidlohr Bueso 已提交
300 301
static void sem_rcu_free(struct rcu_head *head)
{
302 303
	struct kern_ipc_perm *p = container_of(head, struct kern_ipc_perm, rcu);
	struct sem_array *sma = container_of(p, struct sem_array, sem_perm);
D
Davidlohr Bueso 已提交
304

305
	security_sem_free(&sma->sem_perm);
K
Kees Cook 已提交
306
	kvfree(sma);
D
Davidlohr Bueso 已提交
307 308
}

309
/*
310
 * Enter the mode suitable for non-simple operations:
311 312
 * Caller must own sem_perm.lock.
 */
313
static void complexmode_enter(struct sem_array *sma)
314 315 316 317
{
	int i;
	struct sem *sem;

M
Manfred Spraul 已提交
318 319 320 321 322 323 324
	if (sma->use_global_lock > 0)  {
		/*
		 * We are already in global lock mode.
		 * Nothing to do, just reset the
		 * counter until we return to simple mode.
		 */
		sma->use_global_lock = USE_GLOBAL_LOCK_HYSTERESIS;
M
Manfred Spraul 已提交
325 326
		return;
	}
M
Manfred Spraul 已提交
327
	sma->use_global_lock = USE_GLOBAL_LOCK_HYSTERESIS;
328

329
	for (i = 0; i < sma->sem_nsems; i++) {
330
		sem = &sma->sems[i];
331 332
		spin_lock(&sem->lock);
		spin_unlock(&sem->lock);
333
	}
334 335 336 337 338 339 340 341 342 343 344 345 346 347
}

/*
 * Try to leave the mode that disallows simple operations:
 * Caller must own sem_perm.lock.
 */
static void complexmode_tryleave(struct sem_array *sma)
{
	if (sma->complex_count)  {
		/* Complex ops are sleeping.
		 * We must stay in complex mode
		 */
		return;
	}
M
Manfred Spraul 已提交
348 349 350 351 352 353 354 355 356 357 358
	if (sma->use_global_lock == 1) {
		/*
		 * Immediately after setting use_global_lock to 0,
		 * a simple op can start. Thus: all memory writes
		 * performed by the current operation must be visible
		 * before we set use_global_lock to 0.
		 */
		smp_store_release(&sma->use_global_lock, 0);
	} else {
		sma->use_global_lock--;
	}
359 360
}

361
#define SEM_GLOBAL_LOCK	(-1)
362 363 364 365 366 367 368 369 370 371
/*
 * If the request contains only one semaphore operation, and there are
 * no complex transactions pending, lock only the semaphore involved.
 * Otherwise, lock the entire semaphore array, since we either have
 * multiple semaphores in our own semops, or we need to look at
 * semaphores from other pending complex operations.
 */
static inline int sem_lock(struct sem_array *sma, struct sembuf *sops,
			      int nsops)
{
372
	struct sem *sem;
373
	int idx;
374

375 376 377
	if (nsops != 1) {
		/* Complex operation - acquire a full lock */
		ipc_lock_object(&sma->sem_perm);
378

379 380 381
		/* Prevent parallel simple ops */
		complexmode_enter(sma);
		return SEM_GLOBAL_LOCK;
382 383 384 385
	}

	/*
	 * Only one semaphore affected - try to optimize locking.
386 387 388
	 * Optimized locking is possible if no complex operation
	 * is either enqueued or processed right now.
	 *
M
Manfred Spraul 已提交
389
	 * Both facts are tracked by use_global_mode.
390
	 */
391 392
	idx = array_index_nospec(sops->sem_num, sma->sem_nsems);
	sem = &sma->sems[idx];
393

394
	/*
M
Manfred Spraul 已提交
395
	 * Initial check for use_global_lock. Just an optimization,
396 397
	 * no locking, no memory barrier.
	 */
M
Manfred Spraul 已提交
398
	if (!sma->use_global_lock) {
399
		/*
400 401
		 * It appears that no complex operation is around.
		 * Acquire the per-semaphore lock.
402
		 */
403 404
		spin_lock(&sem->lock);

M
Manfred Spraul 已提交
405 406
		/* pairs with smp_store_release() */
		if (!smp_load_acquire(&sma->use_global_lock)) {
407 408
			/* fast path successful! */
			return sops->sem_num;
409
		}
410 411 412 413 414
		spin_unlock(&sem->lock);
	}

	/* slow path: acquire the full lock */
	ipc_lock_object(&sma->sem_perm);
415

M
Manfred Spraul 已提交
416 417 418 419 420 421 422 423 424
	if (sma->use_global_lock == 0) {
		/*
		 * The use_global_lock mode ended while we waited for
		 * sma->sem_perm.lock. Thus we must switch to locking
		 * with sem->lock.
		 * Unlike in the fast path, there is no need to recheck
		 * sma->use_global_lock after we have acquired sem->lock:
		 * We own sma->sem_perm.lock, thus use_global_lock cannot
		 * change.
425 426
		 */
		spin_lock(&sem->lock);
M
Manfred Spraul 已提交
427

428 429
		ipc_unlock_object(&sma->sem_perm);
		return sops->sem_num;
430
	} else {
M
Manfred Spraul 已提交
431 432 433 434
		/*
		 * Not a false alarm, thus continue to use the global lock
		 * mode. No need for complexmode_enter(), this was done by
		 * the caller that has set use_global_mode to non-zero.
435
		 */
436
		return SEM_GLOBAL_LOCK;
437 438 439 440 441
	}
}

static inline void sem_unlock(struct sem_array *sma, int locknum)
{
442
	if (locknum == SEM_GLOBAL_LOCK) {
443
		unmerge_queues(sma);
444
		complexmode_tryleave(sma);
445
		ipc_unlock_object(&sma->sem_perm);
446
	} else {
447
		struct sem *sem = &sma->sems[locknum];
448 449 450 451
		spin_unlock(&sem->lock);
	}
}

N
Nadia Derbey 已提交
452
/*
D
Davidlohr Bueso 已提交
453
 * sem_lock_(check_) routines are called in the paths where the rwsem
N
Nadia Derbey 已提交
454
 * is not held.
455 456
 *
 * The caller holds the RCU read lock.
N
Nadia Derbey 已提交
457
 */
458 459
static inline struct sem_array *sem_obtain_object(struct ipc_namespace *ns, int id)
{
D
Davidlohr Bueso 已提交
460
	struct kern_ipc_perm *ipcp = ipc_obtain_object_idr(&sem_ids(ns), id);
461 462 463 464 465 466 467 468 469 470 471 472 473 474

	if (IS_ERR(ipcp))
		return ERR_CAST(ipcp);

	return container_of(ipcp, struct sem_array, sem_perm);
}

static inline struct sem_array *sem_obtain_object_check(struct ipc_namespace *ns,
							int id)
{
	struct kern_ipc_perm *ipcp = ipc_obtain_object_check(&sem_ids(ns), id);

	if (IS_ERR(ipcp))
		return ERR_CAST(ipcp);
475

N
Nadia Derbey 已提交
476
	return container_of(ipcp, struct sem_array, sem_perm);
477 478
}

479 480
static inline void sem_lock_and_putref(struct sem_array *sma)
{
481
	sem_lock(sma, NULL, -1);
482
	ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
483 484
}

N
Nadia Derbey 已提交
485 486 487 488 489
static inline void sem_rmid(struct ipc_namespace *ns, struct sem_array *s)
{
	ipc_rmid(&sem_ids(ns), &s->sem_perm);
}

K
Kees Cook 已提交
490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507
static struct sem_array *sem_alloc(size_t nsems)
{
	struct sem_array *sma;
	size_t size;

	if (nsems > (INT_MAX - sizeof(*sma)) / sizeof(sma->sems[0]))
		return NULL;

	size = sizeof(*sma) + nsems * sizeof(sma->sems[0]);
	sma = kvmalloc(size, GFP_KERNEL);
	if (unlikely(!sma))
		return NULL;

	memset(sma, 0, size);

	return sma;
}

N
Nadia Derbey 已提交
508 509 510 511 512
/**
 * newary - Create a new semaphore set
 * @ns: namespace
 * @params: ptr to the structure that contains key, semflg and nsems
 *
D
Davidlohr Bueso 已提交
513
 * Called with sem_ids.rwsem held (as a writer)
N
Nadia Derbey 已提交
514
 */
N
Nadia Derbey 已提交
515
static int newary(struct ipc_namespace *ns, struct ipc_params *params)
L
Linus Torvalds 已提交
516 517 518
{
	int retval;
	struct sem_array *sma;
N
Nadia Derbey 已提交
519 520 521
	key_t key = params->key;
	int nsems = params->u.nsems;
	int semflg = params->flg;
522
	int i;
L
Linus Torvalds 已提交
523 524 525

	if (!nsems)
		return -EINVAL;
K
Kirill Korotaev 已提交
526
	if (ns->used_sems + nsems > ns->sc_semmns)
L
Linus Torvalds 已提交
527 528
		return -ENOSPC;

K
Kees Cook 已提交
529
	sma = sem_alloc(nsems);
530
	if (!sma)
L
Linus Torvalds 已提交
531
		return -ENOMEM;
532

L
Linus Torvalds 已提交
533 534 535 536
	sma->sem_perm.mode = (semflg & S_IRWXUGO);
	sma->sem_perm.key = key;

	sma->sem_perm.security = NULL;
537
	retval = security_sem_alloc(&sma->sem_perm);
L
Linus Torvalds 已提交
538
	if (retval) {
K
Kees Cook 已提交
539
		kvfree(sma);
L
Linus Torvalds 已提交
540 541 542
		return retval;
	}

543
	for (i = 0; i < nsems; i++) {
544 545 546
		INIT_LIST_HEAD(&sma->sems[i].pending_alter);
		INIT_LIST_HEAD(&sma->sems[i].pending_const);
		spin_lock_init(&sma->sems[i].lock);
547
	}
548 549

	sma->complex_count = 0;
M
Manfred Spraul 已提交
550
	sma->use_global_lock = USE_GLOBAL_LOCK_HYSTERESIS;
551 552
	INIT_LIST_HEAD(&sma->pending_alter);
	INIT_LIST_HEAD(&sma->pending_const);
553
	INIT_LIST_HEAD(&sma->list_id);
L
Linus Torvalds 已提交
554
	sma->sem_nsems = nsems;
555
	sma->sem_ctime = ktime_get_real_seconds();
556

557
	/* ipc_addid() locks sma upon success. */
558 559
	retval = ipc_addid(&sem_ids(ns), &sma->sem_perm, ns->sc_semmni);
	if (retval < 0) {
560
		ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
561
		return retval;
562 563 564
	}
	ns->used_sems += nsems;

565
	sem_unlock(sma, -1);
566
	rcu_read_unlock();
L
Linus Torvalds 已提交
567

N
Nadia Derbey 已提交
568
	return sma->sem_perm.id;
L
Linus Torvalds 已提交
569 570
}

N
Nadia Derbey 已提交
571

N
Nadia Derbey 已提交
572
/*
D
Davidlohr Bueso 已提交
573
 * Called with sem_ids.rwsem and ipcp locked.
N
Nadia Derbey 已提交
574
 */
N
Nadia Derbey 已提交
575 576
static inline int sem_more_checks(struct kern_ipc_perm *ipcp,
				struct ipc_params *params)
N
Nadia Derbey 已提交
577
{
N
Nadia Derbey 已提交
578 579 580 581
	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
582 583 584 585 586
		return -EINVAL;

	return 0;
}

587
long ksys_semget(key_t key, int nsems, int semflg)
L
Linus Torvalds 已提交
588
{
K
Kirill Korotaev 已提交
589
	struct ipc_namespace *ns;
M
Mathias Krause 已提交
590 591
	static const struct ipc_ops sem_ops = {
		.getnew = newary,
592
		.associate = security_sem_associate,
M
Mathias Krause 已提交
593 594
		.more_checks = sem_more_checks,
	};
N
Nadia Derbey 已提交
595
	struct ipc_params sem_params;
K
Kirill Korotaev 已提交
596 597

	ns = current->nsproxy->ipc_ns;
L
Linus Torvalds 已提交
598

K
Kirill Korotaev 已提交
599
	if (nsems < 0 || nsems > ns->sc_semmsl)
L
Linus Torvalds 已提交
600
		return -EINVAL;
N
Nadia Derbey 已提交
601

N
Nadia Derbey 已提交
602 603 604
	sem_params.key = key;
	sem_params.flg = semflg;
	sem_params.u.nsems = nsems;
L
Linus Torvalds 已提交
605

N
Nadia Derbey 已提交
606
	return ipcget(ns, &sem_ids(ns), &sem_ops, &sem_params);
L
Linus Torvalds 已提交
607 608
}

609 610 611 612 613
SYSCALL_DEFINE3(semget, key_t, key, int, nsems, int, semflg)
{
	return ksys_semget(key, nsems, semflg);
}

614
/**
615 616
 * perform_atomic_semop[_slow] - Attempt to perform semaphore
 *                               operations on a given array.
617
 * @sma: semaphore array
618
 * @q: struct sem_queue that describes the operation
619
 *
620 621 622 623 624 625 626
 * Caller blocking are as follows, based the value
 * indicated by the semaphore operation (sem_op):
 *
 *  (1) >0 never blocks.
 *  (2)  0 (wait-for-zero operation): semval is non-zero.
 *  (3) <0 attempting to decrement semval to a value smaller than zero.
 *
627 628
 * Returns 0 if the operation was possible.
 * Returns 1 if the operation is impossible, the caller must sleep.
629
 * Returns <0 for error codes.
L
Linus Torvalds 已提交
630
 */
631
static int perform_atomic_semop_slow(struct sem_array *sma, struct sem_queue *q)
L
Linus Torvalds 已提交
632
{
633 634
	int result, sem_op, nsops;
	struct pid *pid;
L
Linus Torvalds 已提交
635
	struct sembuf *sop;
M
Manfred Spraul 已提交
636
	struct sem *curr;
637 638 639 640 641 642
	struct sembuf *sops;
	struct sem_undo *un;

	sops = q->sops;
	nsops = q->nsops;
	un = q->undo;
L
Linus Torvalds 已提交
643 644

	for (sop = sops; sop < sops + nsops; sop++) {
645 646
		int idx = array_index_nospec(sop->sem_num, sma->sem_nsems);
		curr = &sma->sems[idx];
L
Linus Torvalds 已提交
647 648
		sem_op = sop->sem_op;
		result = curr->semval;
649

L
Linus Torvalds 已提交
650 651 652 653 654 655 656 657
		if (!sem_op && result)
			goto would_block;

		result += sem_op;
		if (result < 0)
			goto would_block;
		if (result > SEMVMX)
			goto out_of_range;
658

L
Linus Torvalds 已提交
659 660
		if (sop->sem_flg & SEM_UNDO) {
			int undo = un->semadj[sop->sem_num] - sem_op;
661
			/* Exceeding the undo range is an error. */
L
Linus Torvalds 已提交
662 663
			if (undo < (-SEMAEM - 1) || undo > SEMAEM)
				goto out_of_range;
664
			un->semadj[sop->sem_num] = undo;
L
Linus Torvalds 已提交
665
		}
666

L
Linus Torvalds 已提交
667 668 669 670
		curr->semval = result;
	}

	sop--;
671
	pid = q->pid;
L
Linus Torvalds 已提交
672
	while (sop >= sops) {
673
		ipc_update_pid(&sma->sems[sop->sem_num].sempid, pid);
L
Linus Torvalds 已提交
674 675
		sop--;
	}
676

L
Linus Torvalds 已提交
677 678 679 680 681 682 683
	return 0;

out_of_range:
	result = -ERANGE;
	goto undo;

would_block:
684 685
	q->blocking = sop;

L
Linus Torvalds 已提交
686 687 688 689 690 691 692 693
	if (sop->sem_flg & IPC_NOWAIT)
		result = -EAGAIN;
	else
		result = 1;

undo:
	sop--;
	while (sop >= sops) {
694
		sem_op = sop->sem_op;
695
		sma->sems[sop->sem_num].semval -= sem_op;
696 697
		if (sop->sem_flg & SEM_UNDO)
			un->semadj[sop->sem_num] += sem_op;
L
Linus Torvalds 已提交
698 699 700 701 702 703
		sop--;
	}

	return result;
}

704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725
static int perform_atomic_semop(struct sem_array *sma, struct sem_queue *q)
{
	int result, sem_op, nsops;
	struct sembuf *sop;
	struct sem *curr;
	struct sembuf *sops;
	struct sem_undo *un;

	sops = q->sops;
	nsops = q->nsops;
	un = q->undo;

	if (unlikely(q->dupsop))
		return perform_atomic_semop_slow(sma, q);

	/*
	 * We scan the semaphore set twice, first to ensure that the entire
	 * operation can succeed, therefore avoiding any pointless writes
	 * to shared memory and having to undo such changes in order to block
	 * until the operations can go through.
	 */
	for (sop = sops; sop < sops + nsops; sop++) {
726 727 728
		int idx = array_index_nospec(sop->sem_num, sma->sem_nsems);

		curr = &sma->sems[idx];
729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751
		sem_op = sop->sem_op;
		result = curr->semval;

		if (!sem_op && result)
			goto would_block; /* wait-for-zero */

		result += sem_op;
		if (result < 0)
			goto would_block;

		if (result > SEMVMX)
			return -ERANGE;

		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)
				return -ERANGE;
		}
	}

	for (sop = sops; sop < sops + nsops; sop++) {
752
		curr = &sma->sems[sop->sem_num];
753 754 755 756 757 758 759 760 761
		sem_op = sop->sem_op;
		result = curr->semval;

		if (sop->sem_flg & SEM_UNDO) {
			int undo = un->semadj[sop->sem_num] - sem_op;

			un->semadj[sop->sem_num] = undo;
		}
		curr->semval += sem_op;
762
		ipc_update_pid(&curr->sempid, q->pid);
763 764 765 766 767 768 769 770 771
	}

	return 0;

would_block:
	q->blocking = sop;
	return sop->sem_flg & IPC_NOWAIT ? -EAGAIN : 1;
}

D
Davidlohr Bueso 已提交
772 773
static inline void wake_up_sem_queue_prepare(struct sem_queue *q, int error,
					     struct wake_q_head *wake_q)
774
{
D
Davidlohr Bueso 已提交
775 776 777 778 779 780 781 782 783
	wake_q_add(wake_q, q->sleeper);
	/*
	 * Rely on the above implicit barrier, such that we can
	 * ensure that we hold reference to the task before setting
	 * q->status. Otherwise we could race with do_exit if the
	 * task is awoken by an external event before calling
	 * wake_up_process().
	 */
	WRITE_ONCE(q->status, error);
N
Nick Piggin 已提交
784 785
}

786 787 788
static void unlink_queue(struct sem_array *sma, struct sem_queue *q)
{
	list_del(&q->list);
789
	if (q->nsops > 1)
790 791 792
		sma->complex_count--;
}

793 794 795 796 797 798 799
/** 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
800 801
 * modified the array.
 * Note that wait-for-zero operations are handled without restart.
802
 */
803
static inline int check_restart(struct sem_array *sma, struct sem_queue *q)
804
{
805 806
	/* pending complex alter operations are too difficult to analyse */
	if (!list_empty(&sma->pending_alter))
807 808 809 810 811 812
		return 1;

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

813 814 815 816 817 818 819 820 821 822 823 824 825
	/* It is impossible that someone waits for the new value:
	 * - complex operations always restart.
	 * - wait-for-zero are handled seperately.
	 * - q is a previously sleeping simple operation that
	 *   altered the array. It must be a decrement, because
	 *   simple increments never sleep.
	 * - 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.
	 */
	return 0;
}
826

827
/**
D
Davidlohr Bueso 已提交
828
 * wake_const_ops - wake up non-alter tasks
829 830
 * @sma: semaphore array.
 * @semnum: semaphore that was modified.
D
Davidlohr Bueso 已提交
831
 * @wake_q: lockless wake-queue head.
832 833 834 835 836
 *
 * wake_const_ops must be called after a semaphore in a semaphore array
 * was set to 0. If complex const operations are pending, wake_const_ops must
 * be called with semnum = -1, as well as with the number of each modified
 * semaphore.
D
Davidlohr Bueso 已提交
837
 * The tasks that must be woken up are added to @wake_q. The return code
838 839 840 841
 * is stored in q->pid.
 * The function returns 1 if at least one operation was completed successfully.
 */
static int wake_const_ops(struct sem_array *sma, int semnum,
D
Davidlohr Bueso 已提交
842
			  struct wake_q_head *wake_q)
843
{
844
	struct sem_queue *q, *tmp;
845 846 847 848 849 850
	struct list_head *pending_list;
	int semop_completed = 0;

	if (semnum == -1)
		pending_list = &sma->pending_const;
	else
851
		pending_list = &sma->sems[semnum].pending_const;
852

853 854
	list_for_each_entry_safe(q, tmp, pending_list, list) {
		int error = perform_atomic_semop(sma, q);
855

856 857 858 859
		if (error > 0)
			continue;
		/* operation completed, remove from queue & wakeup */
		unlink_queue(sma, q);
860

861 862 863
		wake_up_sem_queue_prepare(q, error, wake_q);
		if (error == 0)
			semop_completed = 1;
864
	}
865

866 867 868 869
	return semop_completed;
}

/**
D
Davidlohr Bueso 已提交
870
 * do_smart_wakeup_zero - wakeup all wait for zero tasks
871 872 873
 * @sma: semaphore array
 * @sops: operations that were performed
 * @nsops: number of operations
D
Davidlohr Bueso 已提交
874
 * @wake_q: lockless wake-queue head
875
 *
D
Davidlohr Bueso 已提交
876 877
 * Checks all required queue for wait-for-zero operations, based
 * on the actual changes that were performed on the semaphore array.
878 879 880
 * The function returns 1 if at least one operation was completed successfully.
 */
static int do_smart_wakeup_zero(struct sem_array *sma, struct sembuf *sops,
D
Davidlohr Bueso 已提交
881
				int nsops, struct wake_q_head *wake_q)
882 883 884 885 886 887 888 889 890 891
{
	int i;
	int semop_completed = 0;
	int got_zero = 0;

	/* first: the per-semaphore queues, if known */
	if (sops) {
		for (i = 0; i < nsops; i++) {
			int num = sops[i].sem_num;

892
			if (sma->sems[num].semval == 0) {
893
				got_zero = 1;
D
Davidlohr Bueso 已提交
894
				semop_completed |= wake_const_ops(sma, num, wake_q);
895 896 897 898 899 900
			}
		}
	} else {
		/*
		 * No sops means modified semaphores not known.
		 * Assume all were changed.
901
		 */
902
		for (i = 0; i < sma->sem_nsems; i++) {
903
			if (sma->sems[i].semval == 0) {
904
				got_zero = 1;
D
Davidlohr Bueso 已提交
905
				semop_completed |= wake_const_ops(sma, i, wake_q);
906 907
			}
		}
908 909
	}
	/*
910 911
	 * If one of the modified semaphores got 0,
	 * then check the global queue, too.
912
	 */
913
	if (got_zero)
D
Davidlohr Bueso 已提交
914
		semop_completed |= wake_const_ops(sma, -1, wake_q);
915

916
	return semop_completed;
917 918
}

919 920

/**
D
Davidlohr Bueso 已提交
921
 * update_queue - look for tasks that can be completed.
922 923
 * @sma: semaphore array.
 * @semnum: semaphore that was modified.
D
Davidlohr Bueso 已提交
924
 * @wake_q: lockless wake-queue head.
925 926
 *
 * update_queue must be called after a semaphore in a semaphore array
927 928 929
 * was modified. If multiple semaphores were modified, update_queue must
 * be called with semnum = -1, as well as with the number of each modified
 * semaphore.
D
Davidlohr Bueso 已提交
930
 * The tasks that must be woken up are added to @wake_q. The return code
931
 * is stored in q->pid.
932 933
 * The function internally checks if const operations can now succeed.
 *
934
 * The function return 1 if at least one semop was completed successfully.
L
Linus Torvalds 已提交
935
 */
D
Davidlohr Bueso 已提交
936
static int update_queue(struct sem_array *sma, int semnum, struct wake_q_head *wake_q)
L
Linus Torvalds 已提交
937
{
938
	struct sem_queue *q, *tmp;
939
	struct list_head *pending_list;
940
	int semop_completed = 0;
941

942
	if (semnum == -1)
943
		pending_list = &sma->pending_alter;
944
	else
945
		pending_list = &sma->sems[semnum].pending_alter;
N
Nick Piggin 已提交
946 947

again:
948
	list_for_each_entry_safe(q, tmp, pending_list, list) {
949
		int error, restart;
950

951 952
		/* If we are scanning the single sop, per-semaphore list of
		 * one semaphore and that semaphore is 0, then it is not
953
		 * necessary to scan further: simple increments
954 955 956 957
		 * 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.
		 */
958
		if (semnum != -1 && sma->sems[semnum].semval == 0)
959 960
			break;

961
		error = perform_atomic_semop(sma, q);
L
Linus Torvalds 已提交
962 963

		/* Does q->sleeper still need to sleep? */
N
Nick Piggin 已提交
964 965 966
		if (error > 0)
			continue;

967
		unlink_queue(sma, q);
N
Nick Piggin 已提交
968

969
		if (error) {
970
			restart = 0;
971 972
		} else {
			semop_completed = 1;
D
Davidlohr Bueso 已提交
973
			do_smart_wakeup_zero(sma, q->sops, q->nsops, wake_q);
974
			restart = check_restart(sma, q);
975
		}
976

D
Davidlohr Bueso 已提交
977
		wake_up_sem_queue_prepare(q, error, wake_q);
978
		if (restart)
N
Nick Piggin 已提交
979
			goto again;
L
Linus Torvalds 已提交
980
	}
981
	return semop_completed;
L
Linus Torvalds 已提交
982 983
}

984
/**
D
Davidlohr Bueso 已提交
985
 * set_semotime - set sem_otime
986 987 988 989 990 991 992 993 994
 * @sma: semaphore array
 * @sops: operations that modified the array, may be NULL
 *
 * sem_otime is replicated to avoid cache line trashing.
 * This function sets one instance to the current time.
 */
static void set_semotime(struct sem_array *sma, struct sembuf *sops)
{
	if (sops == NULL) {
995
		sma->sems[0].sem_otime = ktime_get_real_seconds();
996
	} else {
997
		sma->sems[sops[0].sem_num].sem_otime =
998
						ktime_get_real_seconds();
999 1000 1001
	}
}

1002
/**
D
Davidlohr Bueso 已提交
1003
 * do_smart_update - optimized update_queue
1004 1005 1006
 * @sma: semaphore array
 * @sops: operations that were performed
 * @nsops: number of operations
1007
 * @otime: force setting otime
D
Davidlohr Bueso 已提交
1008
 * @wake_q: lockless wake-queue head
1009
 *
1010 1011
 * do_smart_update() does the required calls to update_queue and wakeup_zero,
 * based on the actual changes that were performed on the semaphore array.
1012
 * Note that the function does not do the actual wake-up: the caller is
D
Davidlohr Bueso 已提交
1013
 * responsible for calling wake_up_q().
1014
 * It is safe to perform this call after dropping all locks.
1015
 */
1016
static void do_smart_update(struct sem_array *sma, struct sembuf *sops, int nsops,
D
Davidlohr Bueso 已提交
1017
			    int otime, struct wake_q_head *wake_q)
1018 1019 1020
{
	int i;

D
Davidlohr Bueso 已提交
1021
	otime |= do_smart_wakeup_zero(sma, sops, nsops, wake_q);
1022

1023 1024
	if (!list_empty(&sma->pending_alter)) {
		/* semaphore array uses the global queue - just process it. */
D
Davidlohr Bueso 已提交
1025
		otime |= update_queue(sma, -1, wake_q);
1026 1027 1028 1029 1030 1031 1032
	} else {
		if (!sops) {
			/*
			 * No sops, thus the modified semaphores are not
			 * known. Check all.
			 */
			for (i = 0; i < sma->sem_nsems; i++)
D
Davidlohr Bueso 已提交
1033
				otime |= update_queue(sma, i, wake_q);
1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046
		} else {
			/*
			 * Check the semaphores that were increased:
			 * - No complex ops, thus all sleeping ops are
			 *   decrease.
			 * - if we decreased the value, then any sleeping
			 *   semaphore ops wont be able to run: If the
			 *   previous value was too small, then the new
			 *   value will be too small, too.
			 */
			for (i = 0; i < nsops; i++) {
				if (sops[i].sem_op > 0) {
					otime |= update_queue(sma,
D
Davidlohr Bueso 已提交
1047
							      sops[i].sem_num, wake_q);
1048
				}
1049
			}
1050
		}
1051
	}
1052 1053
	if (otime)
		set_semotime(sma, sops);
1054 1055
}

1056
/*
1057
 * check_qop: Test if a queued operation sleeps on the semaphore semnum
1058 1059 1060 1061
 */
static int check_qop(struct sem_array *sma, int semnum, struct sem_queue *q,
			bool count_zero)
{
1062
	struct sembuf *sop = q->blocking;
1063

1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074
	/*
	 * Linux always (since 0.99.10) reported a task as sleeping on all
	 * semaphores. This violates SUS, therefore it was changed to the
	 * standard compliant behavior.
	 * Give the administrators a chance to notice that an application
	 * might misbehave because it relies on the Linux behavior.
	 */
	pr_info_once("semctl(GETNCNT/GETZCNT) is since 3.16 Single Unix Specification compliant.\n"
			"The task %s (%d) triggered the difference, watch for misbehavior.\n",
			current->comm, task_pid_nr(current));

1075 1076
	if (sop->sem_num != semnum)
		return 0;
1077

1078 1079 1080 1081 1082 1083
	if (count_zero && sop->sem_op == 0)
		return 1;
	if (!count_zero && sop->sem_op < 0)
		return 1;

	return 0;
1084 1085
}

L
Linus Torvalds 已提交
1086 1087 1088
/* 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
1089 1090 1091
 *
 * Per definition, a task waits only on the semaphore of the first semop
 * that cannot proceed, even if additional operation would block, too.
L
Linus Torvalds 已提交
1092
 */
1093 1094
static int count_semcnt(struct sem_array *sma, ushort semnum,
			bool count_zero)
L
Linus Torvalds 已提交
1095
{
1096
	struct list_head *l;
M
Manfred Spraul 已提交
1097
	struct sem_queue *q;
1098
	int semcnt;
L
Linus Torvalds 已提交
1099

1100 1101 1102
	semcnt = 0;
	/* First: check the simple operations. They are easy to evaluate */
	if (count_zero)
1103
		l = &sma->sems[semnum].pending_const;
1104
	else
1105
		l = &sma->sems[semnum].pending_alter;
L
Linus Torvalds 已提交
1106

1107 1108 1109 1110 1111
	list_for_each_entry(q, l, list) {
		/* all task on a per-semaphore list sleep on exactly
		 * that semaphore
		 */
		semcnt++;
R
Rik van Riel 已提交
1112 1113
	}

1114
	/* Then: check the complex operations. */
1115
	list_for_each_entry(q, &sma->pending_alter, list) {
1116 1117 1118 1119 1120 1121
		semcnt += check_qop(sma, semnum, q, count_zero);
	}
	if (count_zero) {
		list_for_each_entry(q, &sma->pending_const, list) {
			semcnt += check_qop(sma, semnum, q, count_zero);
		}
1122
	}
1123
	return semcnt;
L
Linus Torvalds 已提交
1124 1125
}

D
Davidlohr Bueso 已提交
1126 1127
/* Free a semaphore set. freeary() is called with sem_ids.rwsem locked
 * as a writer and the spinlock for this semaphore set hold. sem_ids.rwsem
N
Nadia Derbey 已提交
1128
 * remains locked on exit.
L
Linus Torvalds 已提交
1129
 */
1130
static void freeary(struct ipc_namespace *ns, struct kern_ipc_perm *ipcp)
L
Linus Torvalds 已提交
1131
{
1132 1133
	struct sem_undo *un, *tu;
	struct sem_queue *q, *tq;
1134
	struct sem_array *sma = container_of(ipcp, struct sem_array, sem_perm);
1135
	int i;
D
Davidlohr Bueso 已提交
1136
	DEFINE_WAKE_Q(wake_q);
L
Linus Torvalds 已提交
1137

1138
	/* Free the existing undo structures for this semaphore set.  */
1139
	ipc_assert_locked_object(&sma->sem_perm);
1140 1141 1142
	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
1143
		un->semid = -1;
1144 1145
		list_del_rcu(&un->list_proc);
		spin_unlock(&un->ulp->lock);
1146
		kfree_rcu(un, rcu);
1147
	}
L
Linus Torvalds 已提交
1148 1149

	/* Wake up all pending processes and let them fail with EIDRM. */
1150 1151
	list_for_each_entry_safe(q, tq, &sma->pending_const, list) {
		unlink_queue(sma, q);
D
Davidlohr Bueso 已提交
1152
		wake_up_sem_queue_prepare(q, -EIDRM, &wake_q);
1153 1154 1155
	}

	list_for_each_entry_safe(q, tq, &sma->pending_alter, list) {
1156
		unlink_queue(sma, q);
D
Davidlohr Bueso 已提交
1157
		wake_up_sem_queue_prepare(q, -EIDRM, &wake_q);
L
Linus Torvalds 已提交
1158
	}
1159
	for (i = 0; i < sma->sem_nsems; i++) {
1160
		struct sem *sem = &sma->sems[i];
1161 1162
		list_for_each_entry_safe(q, tq, &sem->pending_const, list) {
			unlink_queue(sma, q);
D
Davidlohr Bueso 已提交
1163
			wake_up_sem_queue_prepare(q, -EIDRM, &wake_q);
1164 1165
		}
		list_for_each_entry_safe(q, tq, &sem->pending_alter, list) {
1166
			unlink_queue(sma, q);
D
Davidlohr Bueso 已提交
1167
			wake_up_sem_queue_prepare(q, -EIDRM, &wake_q);
1168
		}
1169
		ipc_update_pid(&sem->sempid, NULL);
1170
	}
L
Linus Torvalds 已提交
1171

N
Nadia Derbey 已提交
1172 1173
	/* Remove the semaphore set from the IDR */
	sem_rmid(ns, sma);
1174
	sem_unlock(sma, -1);
1175
	rcu_read_unlock();
L
Linus Torvalds 已提交
1176

D
Davidlohr Bueso 已提交
1177
	wake_up_q(&wake_q);
K
Kirill Korotaev 已提交
1178
	ns->used_sems -= sma->sem_nsems;
1179
	ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
L
Linus Torvalds 已提交
1180 1181 1182 1183
}

static unsigned long copy_semid_to_user(void __user *buf, struct semid64_ds *in, int version)
{
M
Manfred Spraul 已提交
1184
	switch (version) {
L
Linus Torvalds 已提交
1185 1186 1187 1188 1189 1190
	case IPC_64:
		return copy_to_user(buf, in, sizeof(*in));
	case IPC_OLD:
	    {
		struct semid_ds out;

1191 1192
		memset(&out, 0, sizeof(out));

L
Linus Torvalds 已提交
1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205
		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;
	}
}

1206
static time64_t get_semotime(struct sem_array *sma)
1207 1208
{
	int i;
1209
	time64_t res;
1210

1211
	res = sma->sems[0].sem_otime;
1212
	for (i = 1; i < sma->sem_nsems; i++) {
1213
		time64_t to = sma->sems[i].sem_otime;
1214 1215 1216 1217 1218 1219 1220

		if (to > res)
			res = to;
	}
	return res;
}

1221 1222
static int semctl_stat(struct ipc_namespace *ns, int semid,
			 int cmd, struct semid64_ds *semid64)
L
Linus Torvalds 已提交
1223 1224
{
	struct sem_array *sma;
1225
	time64_t semotime;
1226
	int err;
L
Linus Torvalds 已提交
1227

1228
	memset(semid64, 0, sizeof(*semid64));
P
Paul McQuade 已提交
1229

1230
	rcu_read_lock();
1231
	if (cmd == SEM_STAT || cmd == SEM_STAT_ANY) {
1232 1233 1234 1235 1236
		sma = sem_obtain_object(ns, semid);
		if (IS_ERR(sma)) {
			err = PTR_ERR(sma);
			goto out_unlock;
		}
1237
	} else { /* IPC_STAT */
1238 1239 1240 1241
		sma = sem_obtain_object_check(ns, semid);
		if (IS_ERR(sma)) {
			err = PTR_ERR(sma);
			goto out_unlock;
L
Linus Torvalds 已提交
1242 1243 1244
		}
	}

1245 1246 1247 1248 1249 1250 1251 1252
	/* see comment for SHM_STAT_ANY */
	if (cmd == SEM_STAT_ANY)
		audit_ipc_obj(&sma->sem_perm);
	else {
		err = -EACCES;
		if (ipcperms(ns, &sma->sem_perm, S_IRUGO))
			goto out_unlock;
	}
L
Linus Torvalds 已提交
1253

1254
	err = security_sem_semctl(&sma->sem_perm, cmd);
1255 1256
	if (err)
		goto out_unlock;
L
Linus Torvalds 已提交
1257

1258 1259 1260 1261 1262 1263 1264 1265
	ipc_lock_object(&sma->sem_perm);

	if (!ipc_valid_object(&sma->sem_perm)) {
		ipc_unlock_object(&sma->sem_perm);
		err = -EIDRM;
		goto out_unlock;
	}

1266
	kernel_to_ipc64_perm(&sma->sem_perm, &semid64->sem_perm);
1267 1268
	semotime = get_semotime(sma);
	semid64->sem_otime = semotime;
1269
	semid64->sem_ctime = sma->sem_ctime;
1270 1271 1272 1273
#ifndef CONFIG_64BIT
	semid64->sem_otime_high = semotime >> 32;
	semid64->sem_ctime_high = sma->sem_ctime >> 32;
#endif
1274
	semid64->sem_nsems = sma->sem_nsems;
1275

1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288
	if (cmd == IPC_STAT) {
		/*
		 * As defined in SUS:
		 * Return 0 on success
		 */
		err = 0;
	} else {
		/*
		 * SEM_STAT and SEM_STAT_ANY (both Linux specific)
		 * Return the full id, including the sequence number
		 */
		err = sma->sem_perm.id;
	}
1289
	ipc_unlock_object(&sma->sem_perm);
L
Linus Torvalds 已提交
1290
out_unlock:
1291
	rcu_read_unlock();
L
Linus Torvalds 已提交
1292 1293 1294
	return err;
}

1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329
static int semctl_info(struct ipc_namespace *ns, int semid,
			 int cmd, void __user *p)
{
	struct seminfo seminfo;
	int max_id;
	int err;

	err = security_sem_semctl(NULL, cmd);
	if (err)
		return err;

	memset(&seminfo, 0, sizeof(seminfo));
	seminfo.semmni = ns->sc_semmni;
	seminfo.semmns = ns->sc_semmns;
	seminfo.semmsl = ns->sc_semmsl;
	seminfo.semopm = ns->sc_semopm;
	seminfo.semvmx = SEMVMX;
	seminfo.semmnu = SEMMNU;
	seminfo.semmap = SEMMAP;
	seminfo.semume = SEMUME;
	down_read(&sem_ids(ns).rwsem);
	if (cmd == SEM_INFO) {
		seminfo.semusz = sem_ids(ns).in_use;
		seminfo.semaem = ns->used_sems;
	} else {
		seminfo.semusz = SEMUSZ;
		seminfo.semaem = SEMAEM;
	}
	max_id = ipc_get_maxid(&sem_ids(ns));
	up_read(&sem_ids(ns).rwsem);
	if (copy_to_user(p, &seminfo, sizeof(struct seminfo)))
		return -EFAULT;
	return (max_id < 0) ? 0 : max_id;
}

1330
static int semctl_setval(struct ipc_namespace *ns, int semid, int semnum,
1331
		int val)
1332 1333 1334
{
	struct sem_undo *un;
	struct sem_array *sma;
M
Manfred Spraul 已提交
1335
	struct sem *curr;
1336
	int err;
D
Davidlohr Bueso 已提交
1337 1338
	DEFINE_WAKE_Q(wake_q);

1339 1340
	if (val > SEMVMX || val < 0)
		return -ERANGE;
1341

1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358
	rcu_read_lock();
	sma = sem_obtain_object_check(ns, semid);
	if (IS_ERR(sma)) {
		rcu_read_unlock();
		return PTR_ERR(sma);
	}

	if (semnum < 0 || semnum >= sma->sem_nsems) {
		rcu_read_unlock();
		return -EINVAL;
	}


	if (ipcperms(ns, &sma->sem_perm, S_IWUGO)) {
		rcu_read_unlock();
		return -EACCES;
	}
1359

1360
	err = security_sem_semctl(&sma->sem_perm, SETVAL);
1361 1362 1363 1364
	if (err) {
		rcu_read_unlock();
		return -EACCES;
	}
1365

1366
	sem_lock(sma, NULL, -1);
1367

1368
	if (!ipc_valid_object(&sma->sem_perm)) {
1369 1370 1371 1372 1373
		sem_unlock(sma, -1);
		rcu_read_unlock();
		return -EIDRM;
	}

1374
	semnum = array_index_nospec(semnum, sma->sem_nsems);
1375
	curr = &sma->sems[semnum];
1376

1377
	ipc_assert_locked_object(&sma->sem_perm);
1378 1379 1380 1381
	list_for_each_entry(un, &sma->list_id, list_id)
		un->semadj[semnum] = 0;

	curr->semval = val;
1382
	ipc_update_pid(&curr->sempid, task_tgid(current));
1383
	sma->sem_ctime = ktime_get_real_seconds();
1384
	/* maybe some queued-up processes were waiting for this */
D
Davidlohr Bueso 已提交
1385
	do_smart_update(sma, NULL, 0, 0, &wake_q);
1386
	sem_unlock(sma, -1);
1387
	rcu_read_unlock();
D
Davidlohr Bueso 已提交
1388
	wake_up_q(&wake_q);
1389
	return 0;
1390 1391
}

K
Kirill Korotaev 已提交
1392
static int semctl_main(struct ipc_namespace *ns, int semid, int semnum,
1393
		int cmd, void __user *p)
L
Linus Torvalds 已提交
1394 1395
{
	struct sem_array *sma;
M
Manfred Spraul 已提交
1396
	struct sem *curr;
1397
	int err, nsems;
L
Linus Torvalds 已提交
1398
	ushort fast_sem_io[SEMMSL_FAST];
M
Manfred Spraul 已提交
1399
	ushort *sem_io = fast_sem_io;
D
Davidlohr Bueso 已提交
1400
	DEFINE_WAKE_Q(wake_q);
1401 1402 1403 1404 1405

	rcu_read_lock();
	sma = sem_obtain_object_check(ns, semid);
	if (IS_ERR(sma)) {
		rcu_read_unlock();
1406
		return PTR_ERR(sma);
1407
	}
L
Linus Torvalds 已提交
1408 1409 1410 1411

	nsems = sma->sem_nsems;

	err = -EACCES;
1412 1413
	if (ipcperms(ns, &sma->sem_perm, cmd == SETALL ? S_IWUGO : S_IRUGO))
		goto out_rcu_wakeup;
L
Linus Torvalds 已提交
1414

1415
	err = security_sem_semctl(&sma->sem_perm, cmd);
1416 1417
	if (err)
		goto out_rcu_wakeup;
L
Linus Torvalds 已提交
1418 1419 1420 1421 1422

	err = -EACCES;
	switch (cmd) {
	case GETALL:
	{
1423
		ushort __user *array = p;
L
Linus Torvalds 已提交
1424 1425
		int i;

1426
		sem_lock(sma, NULL, -1);
1427
		if (!ipc_valid_object(&sma->sem_perm)) {
1428 1429 1430
			err = -EIDRM;
			goto out_unlock;
		}
M
Manfred Spraul 已提交
1431
		if (nsems > SEMMSL_FAST) {
1432
			if (!ipc_rcu_getref(&sma->sem_perm)) {
1433
				err = -EIDRM;
1434
				goto out_unlock;
1435 1436
			}
			sem_unlock(sma, -1);
1437
			rcu_read_unlock();
K
Kees Cook 已提交
1438 1439
			sem_io = kvmalloc_array(nsems, sizeof(ushort),
						GFP_KERNEL);
M
Manfred Spraul 已提交
1440
			if (sem_io == NULL) {
1441
				ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
L
Linus Torvalds 已提交
1442 1443 1444
				return -ENOMEM;
			}

1445
			rcu_read_lock();
1446
			sem_lock_and_putref(sma);
1447
			if (!ipc_valid_object(&sma->sem_perm)) {
L
Linus Torvalds 已提交
1448
				err = -EIDRM;
1449
				goto out_unlock;
L
Linus Torvalds 已提交
1450
			}
1451
		}
L
Linus Torvalds 已提交
1452
		for (i = 0; i < sma->sem_nsems; i++)
1453
			sem_io[i] = sma->sems[i].semval;
1454
		sem_unlock(sma, -1);
1455
		rcu_read_unlock();
L
Linus Torvalds 已提交
1456
		err = 0;
M
Manfred Spraul 已提交
1457
		if (copy_to_user(array, sem_io, nsems*sizeof(ushort)))
L
Linus Torvalds 已提交
1458 1459 1460 1461 1462 1463 1464 1465
			err = -EFAULT;
		goto out_free;
	}
	case SETALL:
	{
		int i;
		struct sem_undo *un;

1466
		if (!ipc_rcu_getref(&sma->sem_perm)) {
1467 1468
			err = -EIDRM;
			goto out_rcu_wakeup;
1469
		}
1470
		rcu_read_unlock();
L
Linus Torvalds 已提交
1471

M
Manfred Spraul 已提交
1472
		if (nsems > SEMMSL_FAST) {
K
Kees Cook 已提交
1473 1474
			sem_io = kvmalloc_array(nsems, sizeof(ushort),
						GFP_KERNEL);
M
Manfred Spraul 已提交
1475
			if (sem_io == NULL) {
1476
				ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
L
Linus Torvalds 已提交
1477 1478 1479 1480
				return -ENOMEM;
			}
		}

M
Manfred Spraul 已提交
1481
		if (copy_from_user(sem_io, p, nsems*sizeof(ushort))) {
1482
			ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
L
Linus Torvalds 已提交
1483 1484 1485 1486 1487 1488
			err = -EFAULT;
			goto out_free;
		}

		for (i = 0; i < nsems; i++) {
			if (sem_io[i] > SEMVMX) {
1489
				ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
L
Linus Torvalds 已提交
1490 1491 1492 1493
				err = -ERANGE;
				goto out_free;
			}
		}
1494
		rcu_read_lock();
1495
		sem_lock_and_putref(sma);
1496
		if (!ipc_valid_object(&sma->sem_perm)) {
L
Linus Torvalds 已提交
1497
			err = -EIDRM;
1498
			goto out_unlock;
L
Linus Torvalds 已提交
1499 1500
		}

1501
		for (i = 0; i < nsems; i++) {
1502
			sma->sems[i].semval = sem_io[i];
1503
			ipc_update_pid(&sma->sems[i].sempid, task_tgid(current));
1504
		}
1505

1506
		ipc_assert_locked_object(&sma->sem_perm);
1507
		list_for_each_entry(un, &sma->list_id, list_id) {
L
Linus Torvalds 已提交
1508 1509
			for (i = 0; i < nsems; i++)
				un->semadj[i] = 0;
1510
		}
1511
		sma->sem_ctime = ktime_get_real_seconds();
L
Linus Torvalds 已提交
1512
		/* maybe some queued-up processes were waiting for this */
D
Davidlohr Bueso 已提交
1513
		do_smart_update(sma, NULL, 0, 0, &wake_q);
L
Linus Torvalds 已提交
1514 1515 1516
		err = 0;
		goto out_unlock;
	}
1517
	/* GETVAL, GETPID, GETNCTN, GETZCNT: fall-through */
L
Linus Torvalds 已提交
1518 1519
	}
	err = -EINVAL;
1520 1521
	if (semnum < 0 || semnum >= nsems)
		goto out_rcu_wakeup;
L
Linus Torvalds 已提交
1522

1523
	sem_lock(sma, NULL, -1);
1524
	if (!ipc_valid_object(&sma->sem_perm)) {
1525 1526 1527
		err = -EIDRM;
		goto out_unlock;
	}
1528 1529

	semnum = array_index_nospec(semnum, nsems);
1530
	curr = &sma->sems[semnum];
L
Linus Torvalds 已提交
1531 1532 1533 1534 1535 1536

	switch (cmd) {
	case GETVAL:
		err = curr->semval;
		goto out_unlock;
	case GETPID:
1537
		err = pid_vnr(curr->sempid);
L
Linus Torvalds 已提交
1538 1539
		goto out_unlock;
	case GETNCNT:
1540
		err = count_semcnt(sma, semnum, 0);
L
Linus Torvalds 已提交
1541 1542
		goto out_unlock;
	case GETZCNT:
1543
		err = count_semcnt(sma, semnum, 1);
L
Linus Torvalds 已提交
1544 1545
		goto out_unlock;
	}
1546

L
Linus Torvalds 已提交
1547
out_unlock:
1548
	sem_unlock(sma, -1);
1549
out_rcu_wakeup:
1550
	rcu_read_unlock();
D
Davidlohr Bueso 已提交
1551
	wake_up_q(&wake_q);
L
Linus Torvalds 已提交
1552
out_free:
M
Manfred Spraul 已提交
1553
	if (sem_io != fast_sem_io)
K
Kees Cook 已提交
1554
		kvfree(sem_io);
L
Linus Torvalds 已提交
1555 1556 1557
	return err;
}

1558 1559
static inline unsigned long
copy_semid_from_user(struct semid64_ds *out, void __user *buf, int version)
L
Linus Torvalds 已提交
1560
{
M
Manfred Spraul 已提交
1561
	switch (version) {
L
Linus Torvalds 已提交
1562
	case IPC_64:
1563
		if (copy_from_user(out, buf, sizeof(*out)))
L
Linus Torvalds 已提交
1564 1565 1566 1567 1568 1569
			return -EFAULT;
		return 0;
	case IPC_OLD:
	    {
		struct semid_ds tbuf_old;

M
Manfred Spraul 已提交
1570
		if (copy_from_user(&tbuf_old, buf, sizeof(tbuf_old)))
L
Linus Torvalds 已提交
1571 1572
			return -EFAULT;

1573 1574 1575
		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
1576 1577 1578 1579 1580 1581 1582 1583

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

1584
/*
D
Davidlohr Bueso 已提交
1585
 * This function handles some semctl commands which require the rwsem
1586
 * to be held in write mode.
D
Davidlohr Bueso 已提交
1587
 * NOTE: no locks must be held, the rwsem is taken inside this function.
1588
 */
1589
static int semctl_down(struct ipc_namespace *ns, int semid,
1590
		       int cmd, struct semid64_ds *semid64)
L
Linus Torvalds 已提交
1591 1592 1593 1594 1595
{
	struct sem_array *sma;
	int err;
	struct kern_ipc_perm *ipcp;

D
Davidlohr Bueso 已提交
1596
	down_write(&sem_ids(ns).rwsem);
1597 1598
	rcu_read_lock();

1599
	ipcp = ipcctl_pre_down_nolock(ns, &sem_ids(ns), semid, cmd,
1600
				      &semid64->sem_perm, 0);
1601 1602 1603 1604
	if (IS_ERR(ipcp)) {
		err = PTR_ERR(ipcp);
		goto out_unlock1;
	}
S
Steve Grubb 已提交
1605

1606
	sma = container_of(ipcp, struct sem_array, sem_perm);
L
Linus Torvalds 已提交
1607

1608
	err = security_sem_semctl(&sma->sem_perm, cmd);
1609 1610
	if (err)
		goto out_unlock1;
L
Linus Torvalds 已提交
1611

1612
	switch (cmd) {
L
Linus Torvalds 已提交
1613
	case IPC_RMID:
1614
		sem_lock(sma, NULL, -1);
1615
		/* freeary unlocks the ipc object and rcu */
1616
		freeary(ns, ipcp);
1617
		goto out_up;
L
Linus Torvalds 已提交
1618
	case IPC_SET:
1619
		sem_lock(sma, NULL, -1);
1620
		err = ipc_update_perm(&semid64->sem_perm, ipcp);
1621
		if (err)
1622
			goto out_unlock0;
1623
		sma->sem_ctime = ktime_get_real_seconds();
L
Linus Torvalds 已提交
1624 1625 1626
		break;
	default:
		err = -EINVAL;
1627
		goto out_unlock1;
L
Linus Torvalds 已提交
1628 1629
	}

1630
out_unlock0:
1631
	sem_unlock(sma, -1);
1632
out_unlock1:
1633
	rcu_read_unlock();
1634
out_up:
D
Davidlohr Bueso 已提交
1635
	up_write(&sem_ids(ns).rwsem);
L
Linus Torvalds 已提交
1636 1637 1638
	return err;
}

1639
long ksys_semctl(int semid, int semnum, int cmd, unsigned long arg)
L
Linus Torvalds 已提交
1640 1641
{
	int version;
K
Kirill Korotaev 已提交
1642
	struct ipc_namespace *ns;
1643
	void __user *p = (void __user *)arg;
1644 1645
	struct semid64_ds semid64;
	int err;
L
Linus Torvalds 已提交
1646 1647 1648 1649 1650

	if (semid < 0)
		return -EINVAL;

	version = ipc_parse_version(&cmd);
K
Kirill Korotaev 已提交
1651
	ns = current->nsproxy->ipc_ns;
L
Linus Torvalds 已提交
1652

M
Manfred Spraul 已提交
1653
	switch (cmd) {
L
Linus Torvalds 已提交
1654 1655
	case IPC_INFO:
	case SEM_INFO:
1656
		return semctl_info(ns, semid, cmd, p);
1657
	case IPC_STAT:
L
Linus Torvalds 已提交
1658
	case SEM_STAT:
1659
	case SEM_STAT_ANY:
1660 1661 1662 1663 1664 1665
		err = semctl_stat(ns, semid, cmd, &semid64);
		if (err < 0)
			return err;
		if (copy_semid_to_user(p, &semid64, version))
			err = -EFAULT;
		return err;
L
Linus Torvalds 已提交
1666 1667 1668 1669 1670 1671
	case GETALL:
	case GETVAL:
	case GETPID:
	case GETNCNT:
	case GETZCNT:
	case SETALL:
1672
		return semctl_main(ns, semid, semnum, cmd, p);
1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683
	case SETVAL: {
		int val;
#if defined(CONFIG_64BIT) && defined(__BIG_ENDIAN)
		/* big-endian 64bit */
		val = arg >> 32;
#else
		/* 32bit or little-endian 64bit */
		val = arg;
#endif
		return semctl_setval(ns, semid, semnum, val);
	}
L
Linus Torvalds 已提交
1684
	case IPC_SET:
1685 1686 1687 1688
		if (copy_semid_from_user(&semid64, p, version))
			return -EFAULT;
	case IPC_RMID:
		return semctl_down(ns, semid, cmd, &semid64);
L
Linus Torvalds 已提交
1689 1690 1691 1692 1693
	default:
		return -EINVAL;
	}
}

1694 1695 1696 1697 1698
SYSCALL_DEFINE4(semctl, int, semid, int, semnum, int, cmd, unsigned long, arg)
{
	return ksys_semctl(semid, semnum, cmd, arg);
}

A
Al Viro 已提交
1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716
#ifdef CONFIG_COMPAT

struct compat_semid_ds {
	struct compat_ipc_perm sem_perm;
	compat_time_t sem_otime;
	compat_time_t sem_ctime;
	compat_uptr_t sem_base;
	compat_uptr_t sem_pending;
	compat_uptr_t sem_pending_last;
	compat_uptr_t undo;
	unsigned short sem_nsems;
};

static int copy_compat_semid_from_user(struct semid64_ds *out, void __user *buf,
					int version)
{
	memset(out, 0, sizeof(*out));
	if (version == IPC_64) {
1717
		struct compat_semid64_ds __user *p = buf;
A
Al Viro 已提交
1718 1719
		return get_compat_ipc64_perm(&out->sem_perm, &p->sem_perm);
	} else {
1720
		struct compat_semid_ds __user *p = buf;
A
Al Viro 已提交
1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731
		return get_compat_ipc_perm(&out->sem_perm, &p->sem_perm);
	}
}

static int copy_compat_semid_to_user(void __user *buf, struct semid64_ds *in,
					int version)
{
	if (version == IPC_64) {
		struct compat_semid64_ds v;
		memset(&v, 0, sizeof(v));
		to_compat_ipc64_perm(&v.sem_perm, &in->sem_perm);
1732 1733 1734 1735
		v.sem_otime	 = lower_32_bits(in->sem_otime);
		v.sem_otime_high = upper_32_bits(in->sem_otime);
		v.sem_ctime	 = lower_32_bits(in->sem_ctime);
		v.sem_ctime_high = upper_32_bits(in->sem_ctime);
A
Al Viro 已提交
1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748
		v.sem_nsems = in->sem_nsems;
		return copy_to_user(buf, &v, sizeof(v));
	} else {
		struct compat_semid_ds v;
		memset(&v, 0, sizeof(v));
		to_compat_ipc_perm(&v.sem_perm, &in->sem_perm);
		v.sem_otime = in->sem_otime;
		v.sem_ctime = in->sem_ctime;
		v.sem_nsems = in->sem_nsems;
		return copy_to_user(buf, &v, sizeof(v));
	}
}

1749
long compat_ksys_semctl(int semid, int semnum, int cmd, int arg)
A
Al Viro 已提交
1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767
{
	void __user *p = compat_ptr(arg);
	struct ipc_namespace *ns;
	struct semid64_ds semid64;
	int version = compat_ipc_parse_version(&cmd);
	int err;

	ns = current->nsproxy->ipc_ns;

	if (semid < 0)
		return -EINVAL;

	switch (cmd & (~IPC_64)) {
	case IPC_INFO:
	case SEM_INFO:
		return semctl_info(ns, semid, cmd, p);
	case IPC_STAT:
	case SEM_STAT:
1768
	case SEM_STAT_ANY:
A
Al Viro 已提交
1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779
		err = semctl_stat(ns, semid, cmd, &semid64);
		if (err < 0)
			return err;
		if (copy_compat_semid_to_user(p, &semid64, version))
			err = -EFAULT;
		return err;
	case GETVAL:
	case GETPID:
	case GETNCNT:
	case GETZCNT:
	case GETALL:
L
Linus Torvalds 已提交
1780
	case SETALL:
1781 1782 1783
		return semctl_main(ns, semid, semnum, cmd, p);
	case SETVAL:
		return semctl_setval(ns, semid, semnum, arg);
L
Linus Torvalds 已提交
1784
	case IPC_SET:
A
Al Viro 已提交
1785 1786 1787 1788 1789
		if (copy_compat_semid_from_user(&semid64, p, version))
			return -EFAULT;
		/* fallthru */
	case IPC_RMID:
		return semctl_down(ns, semid, cmd, &semid64);
L
Linus Torvalds 已提交
1790 1791 1792 1793
	default:
		return -EINVAL;
	}
}
1794 1795 1796 1797 1798

COMPAT_SYSCALL_DEFINE4(semctl, int, semid, int, semnum, int, cmd, int, arg)
{
	return compat_ksys_semctl(semid, semnum, cmd, arg);
}
A
Al Viro 已提交
1799
#endif
L
Linus Torvalds 已提交
1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817

/* 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) {
1818
		undo_list = kzalloc(sizeof(*undo_list), GFP_KERNEL);
L
Linus Torvalds 已提交
1819 1820
		if (undo_list == NULL)
			return -ENOMEM;
I
Ingo Molnar 已提交
1821
		spin_lock_init(&undo_list->lock);
1822
		refcount_set(&undo_list->refcnt, 1);
1823 1824
		INIT_LIST_HEAD(&undo_list->list_proc);

L
Linus Torvalds 已提交
1825 1826 1827 1828 1829 1830
		current->sysvsem.undo_list = undo_list;
	}
	*undo_listp = undo_list;
	return 0;
}

1831
static struct sem_undo *__lookup_undo(struct sem_undo_list *ulp, int semid)
L
Linus Torvalds 已提交
1832
{
1833
	struct sem_undo *un;
1834

1835 1836 1837
	list_for_each_entry_rcu(un, &ulp->list_proc, list_proc) {
		if (un->semid == semid)
			return un;
L
Linus Torvalds 已提交
1838
	}
1839
	return NULL;
L
Linus Torvalds 已提交
1840 1841
}

1842 1843 1844 1845
static struct sem_undo *lookup_undo(struct sem_undo_list *ulp, int semid)
{
	struct sem_undo *un;

M
Manfred Spraul 已提交
1846
	assert_spin_locked(&ulp->lock);
1847 1848 1849 1850 1851 1852 1853 1854 1855

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

1856
/**
D
Davidlohr Bueso 已提交
1857
 * find_alloc_undo - lookup (and if not present create) undo array
1858 1859 1860 1861 1862 1863
 * @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.
1864 1865
 * Lifetime-rules: sem_undo is rcu-protected, on success, the function
 * performs a rcu_read_lock().
1866 1867
 */
static struct sem_undo *find_alloc_undo(struct ipc_namespace *ns, int semid)
L
Linus Torvalds 已提交
1868 1869 1870 1871
{
	struct sem_array *sma;
	struct sem_undo_list *ulp;
	struct sem_undo *un, *new;
1872
	int nsems, error;
L
Linus Torvalds 已提交
1873 1874 1875 1876 1877

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

1878
	rcu_read_lock();
1879
	spin_lock(&ulp->lock);
L
Linus Torvalds 已提交
1880
	un = lookup_undo(ulp, semid);
1881
	spin_unlock(&ulp->lock);
M
Manfred Spraul 已提交
1882
	if (likely(un != NULL))
L
Linus Torvalds 已提交
1883 1884 1885
		goto out;

	/* no undo structure around - allocate one. */
1886
	/* step 1: figure out the size of the semaphore array */
1887 1888 1889
	sma = sem_obtain_object_check(ns, semid);
	if (IS_ERR(sma)) {
		rcu_read_unlock();
J
Julia Lawall 已提交
1890
		return ERR_CAST(sma);
1891
	}
1892

L
Linus Torvalds 已提交
1893
	nsems = sma->sem_nsems;
1894
	if (!ipc_rcu_getref(&sma->sem_perm)) {
1895 1896 1897 1898
		rcu_read_unlock();
		un = ERR_PTR(-EIDRM);
		goto out;
	}
1899
	rcu_read_unlock();
L
Linus Torvalds 已提交
1900

1901
	/* step 2: allocate new undo structure */
1902
	new = kzalloc(sizeof(struct sem_undo) + sizeof(short)*nsems, GFP_KERNEL);
L
Linus Torvalds 已提交
1903
	if (!new) {
1904
		ipc_rcu_putref(&sma->sem_perm, sem_rcu_free);
L
Linus Torvalds 已提交
1905 1906 1907
		return ERR_PTR(-ENOMEM);
	}

1908
	/* step 3: Acquire the lock on semaphore array */
1909
	rcu_read_lock();
1910
	sem_lock_and_putref(sma);
1911
	if (!ipc_valid_object(&sma->sem_perm)) {
1912
		sem_unlock(sma, -1);
1913
		rcu_read_unlock();
L
Linus Torvalds 已提交
1914 1915 1916 1917
		kfree(new);
		un = ERR_PTR(-EIDRM);
		goto out;
	}
1918 1919 1920 1921 1922 1923 1924 1925 1926 1927
	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;
	}
1928 1929
	/* step 5: initialize & link new undo structure */
	new->semadj = (short *) &new[1];
1930
	new->ulp = ulp;
1931 1932
	new->semid = semid;
	assert_spin_locked(&ulp->lock);
1933
	list_add_rcu(&new->list_proc, &ulp->list_proc);
1934
	ipc_assert_locked_object(&sma->sem_perm);
1935
	list_add(&new->list_id, &sma->list_id);
1936
	un = new;
1937

1938
success:
1939
	spin_unlock(&ulp->lock);
1940
	sem_unlock(sma, -1);
L
Linus Torvalds 已提交
1941 1942 1943 1944
out:
	return un;
}

A
Al Viro 已提交
1945
static long do_semtimedop(int semid, struct sembuf __user *tsops,
1946
		unsigned nsops, const struct timespec64 *timeout)
L
Linus Torvalds 已提交
1947 1948 1949 1950
{
	int error = -EINVAL;
	struct sem_array *sma;
	struct sembuf fast_sops[SEMOPM_FAST];
M
Manfred Spraul 已提交
1951
	struct sembuf *sops = fast_sops, *sop;
L
Linus Torvalds 已提交
1952
	struct sem_undo *un;
1953 1954
	int max, locknum;
	bool undos = false, alter = false, dupsop = false;
L
Linus Torvalds 已提交
1955
	struct sem_queue queue;
1956
	unsigned long dup = 0, jiffies_left = 0;
K
Kirill Korotaev 已提交
1957 1958 1959
	struct ipc_namespace *ns;

	ns = current->nsproxy->ipc_ns;
L
Linus Torvalds 已提交
1960 1961 1962

	if (nsops < 1 || semid < 0)
		return -EINVAL;
K
Kirill Korotaev 已提交
1963
	if (nsops > ns->sc_semopm)
L
Linus Torvalds 已提交
1964
		return -E2BIG;
M
Manfred Spraul 已提交
1965
	if (nsops > SEMOPM_FAST) {
1966
		sops = kvmalloc_array(nsops, sizeof(*sops), GFP_KERNEL);
M
Manfred Spraul 已提交
1967
		if (sops == NULL)
L
Linus Torvalds 已提交
1968 1969
			return -ENOMEM;
	}
1970

M
Manfred Spraul 已提交
1971 1972
	if (copy_from_user(sops, tsops, nsops * sizeof(*tsops))) {
		error =  -EFAULT;
L
Linus Torvalds 已提交
1973 1974
		goto out_free;
	}
1975

L
Linus Torvalds 已提交
1976
	if (timeout) {
A
Al Viro 已提交
1977 1978
		if (timeout->tv_sec < 0 || timeout->tv_nsec < 0 ||
			timeout->tv_nsec >= 1000000000L) {
L
Linus Torvalds 已提交
1979 1980 1981
			error = -EINVAL;
			goto out_free;
		}
1982
		jiffies_left = timespec64_to_jiffies(timeout);
L
Linus Torvalds 已提交
1983
	}
1984

L
Linus Torvalds 已提交
1985 1986
	max = 0;
	for (sop = sops; sop < sops + nsops; sop++) {
1987 1988
		unsigned long mask = 1ULL << ((sop->sem_num) % BITS_PER_LONG);

L
Linus Torvalds 已提交
1989 1990 1991
		if (sop->sem_num >= max)
			max = sop->sem_num;
		if (sop->sem_flg & SEM_UNDO)
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
			undos = true;
		if (dup & mask) {
			/*
			 * There was a previous alter access that appears
			 * to have accessed the same semaphore, thus use
			 * the dupsop logic. "appears", because the detection
			 * can only check % BITS_PER_LONG.
			 */
			dupsop = true;
		}
		if (sop->sem_op != 0) {
			alter = true;
			dup |= mask;
		}
L
Linus Torvalds 已提交
2006 2007 2008
	}

	if (undos) {
2009
		/* On success, find_alloc_undo takes the rcu_read_lock */
2010
		un = find_alloc_undo(ns, semid);
L
Linus Torvalds 已提交
2011 2012 2013 2014
		if (IS_ERR(un)) {
			error = PTR_ERR(un);
			goto out_free;
		}
2015
	} else {
L
Linus Torvalds 已提交
2016
		un = NULL;
2017 2018
		rcu_read_lock();
	}
L
Linus Torvalds 已提交
2019

2020
	sma = sem_obtain_object_check(ns, semid);
2021
	if (IS_ERR(sma)) {
2022
		rcu_read_unlock();
2023
		error = PTR_ERR(sma);
L
Linus Torvalds 已提交
2024
		goto out_free;
2025 2026
	}

2027
	error = -EFBIG;
2028 2029 2030 2031
	if (max >= sma->sem_nsems) {
		rcu_read_unlock();
		goto out_free;
	}
2032 2033

	error = -EACCES;
2034 2035 2036 2037
	if (ipcperms(ns, &sma->sem_perm, alter ? S_IWUGO : S_IRUGO)) {
		rcu_read_unlock();
		goto out_free;
	}
2038

2039
	error = security_sem_semop(&sma->sem_perm, sops, nsops, alter);
2040 2041 2042 2043
	if (error) {
		rcu_read_unlock();
		goto out_free;
	}
2044

2045 2046
	error = -EIDRM;
	locknum = sem_lock(sma, sops, nsops);
2047 2048 2049 2050 2051 2052 2053 2054 2055
	/*
	 * We eventually might perform the following check in a lockless
	 * fashion, considering ipc_valid_object() locking constraints.
	 * If nsops == 1 and there is no contention for sem_perm.lock, then
	 * only a per-semaphore lock is held and it's OK to proceed with the
	 * check below. More details on the fine grained locking scheme
	 * entangled here and why it's RMID race safe on comments at sem_lock()
	 */
	if (!ipc_valid_object(&sma->sem_perm))
2056
		goto out_unlock_free;
L
Linus Torvalds 已提交
2057
	/*
2058
	 * semid identifiers are not unique - find_alloc_undo may have
L
Linus Torvalds 已提交
2059
	 * allocated an undo structure, it was invalidated by an RMID
2060
	 * and now a new array with received the same id. Check and fail.
L
Lucas De Marchi 已提交
2061
	 * This case can be detected checking un->semid. The existence of
2062
	 * "un" itself is guaranteed by rcu.
L
Linus Torvalds 已提交
2063
	 */
2064 2065
	if (un && un->semid == -1)
		goto out_unlock_free;
2066

2067 2068 2069
	queue.sops = sops;
	queue.nsops = nsops;
	queue.undo = un;
2070
	queue.pid = task_tgid(current);
2071
	queue.alter = alter;
2072
	queue.dupsop = dupsop;
2073 2074

	error = perform_atomic_semop(sma, &queue);
D
Davidlohr Bueso 已提交
2075 2076 2077 2078 2079
	if (error == 0) { /* non-blocking succesfull path */
		DEFINE_WAKE_Q(wake_q);

		/*
		 * If the operation was successful, then do
2080 2081 2082
		 * the required updates.
		 */
		if (alter)
D
Davidlohr Bueso 已提交
2083
			do_smart_update(sma, sops, nsops, 1, &wake_q);
2084 2085
		else
			set_semotime(sma, sops);
D
Davidlohr Bueso 已提交
2086 2087 2088 2089 2090 2091

		sem_unlock(sma, locknum);
		rcu_read_unlock();
		wake_up_q(&wake_q);

		goto out_free;
L
Linus Torvalds 已提交
2092
	}
D
Davidlohr Bueso 已提交
2093
	if (error < 0) /* non-blocking error path */
2094
		goto out_unlock_free;
L
Linus Torvalds 已提交
2095

D
Davidlohr Bueso 已提交
2096 2097
	/*
	 * We need to sleep on this operation, so we put the current
L
Linus Torvalds 已提交
2098 2099
	 * task into the pending queue and go to sleep.
	 */
2100 2101
	if (nsops == 1) {
		struct sem *curr;
2102 2103
		int idx = array_index_nospec(sops->sem_num, sma->sem_nsems);
		curr = &sma->sems[idx];
2104

2105 2106 2107 2108 2109 2110 2111 2112 2113 2114
		if (alter) {
			if (sma->complex_count) {
				list_add_tail(&queue.list,
						&sma->pending_alter);
			} else {

				list_add_tail(&queue.list,
						&curr->pending_alter);
			}
		} else {
2115
			list_add_tail(&queue.list, &curr->pending_const);
2116
		}
2117
	} else {
2118 2119 2120
		if (!sma->complex_count)
			merge_queues(sma);

2121
		if (alter)
2122
			list_add_tail(&queue.list, &sma->pending_alter);
2123
		else
2124 2125
			list_add_tail(&queue.list, &sma->pending_const);

2126 2127 2128
		sma->complex_count++;
	}

D
Davidlohr Bueso 已提交
2129
	do {
2130
		WRITE_ONCE(queue.status, -EINTR);
D
Davidlohr Bueso 已提交
2131
		queue.sleeper = current;
2132

D
Davidlohr Bueso 已提交
2133 2134 2135
		__set_current_state(TASK_INTERRUPTIBLE);
		sem_unlock(sma, locknum);
		rcu_read_unlock();
L
Linus Torvalds 已提交
2136

D
Davidlohr Bueso 已提交
2137 2138 2139 2140
		if (timeout)
			jiffies_left = schedule_timeout(jiffies_left);
		else
			schedule();
L
Linus Torvalds 已提交
2141

D
Davidlohr Bueso 已提交
2142
		/*
D
Davidlohr Bueso 已提交
2143 2144 2145 2146 2147 2148 2149 2150 2151
		 * fastpath: the semop has completed, either successfully or
		 * not, from the syscall pov, is quite irrelevant to us at this
		 * point; we're done.
		 *
		 * We _do_ care, nonetheless, about being awoken by a signal or
		 * spuriously.  The queue.status is checked again in the
		 * slowpath (aka after taking sem_lock), such that we can detect
		 * scenarios where we were awakened externally, during the
		 * window between wake_q_add() and wake_up_q().
2152
		 */
D
Davidlohr Bueso 已提交
2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163
		error = READ_ONCE(queue.status);
		if (error != -EINTR) {
			/*
			 * User space could assume that semop() is a memory
			 * barrier: Without the mb(), the cpu could
			 * speculatively read in userspace stale data that was
			 * overwritten by the previous owner of the semaphore.
			 */
			smp_mb();
			goto out_free;
		}
2164

D
Davidlohr Bueso 已提交
2165
		rcu_read_lock();
2166
		locknum = sem_lock(sma, sops, nsops);
L
Linus Torvalds 已提交
2167

2168 2169 2170 2171
		if (!ipc_valid_object(&sma->sem_perm))
			goto out_unlock_free;

		error = READ_ONCE(queue.status);
L
Linus Torvalds 已提交
2172

D
Davidlohr Bueso 已提交
2173 2174 2175 2176 2177 2178
		/*
		 * If queue.status != -EINTR we are woken up by another process.
		 * Leave without unlink_queue(), but with sem_unlock().
		 */
		if (error != -EINTR)
			goto out_unlock_free;
2179

D
Davidlohr Bueso 已提交
2180 2181 2182 2183 2184 2185
		/*
		 * If an interrupt occurred we have to clean up the queue.
		 */
		if (timeout && jiffies_left == 0)
			error = -EAGAIN;
	} while (error == -EINTR && !signal_pending(current)); /* spurious */
2186

2187
	unlink_queue(sma, &queue);
L
Linus Torvalds 已提交
2188 2189

out_unlock_free:
2190
	sem_unlock(sma, locknum);
2191
	rcu_read_unlock();
L
Linus Torvalds 已提交
2192
out_free:
M
Manfred Spraul 已提交
2193
	if (sops != fast_sops)
2194
		kvfree(sops);
L
Linus Torvalds 已提交
2195 2196 2197
	return error;
}

2198
long ksys_semtimedop(int semid, struct sembuf __user *tsops,
2199
		     unsigned int nsops, const struct __kernel_timespec __user *timeout)
A
Al Viro 已提交
2200 2201
{
	if (timeout) {
2202 2203
		struct timespec64 ts;
		if (get_timespec64(&ts, timeout))
A
Al Viro 已提交
2204 2205 2206 2207 2208 2209
			return -EFAULT;
		return do_semtimedop(semid, tsops, nsops, &ts);
	}
	return do_semtimedop(semid, tsops, nsops, NULL);
}

2210
SYSCALL_DEFINE4(semtimedop, int, semid, struct sembuf __user *, tsops,
2211
		unsigned int, nsops, const struct __kernel_timespec __user *, timeout)
2212 2213 2214 2215
{
	return ksys_semtimedop(semid, tsops, nsops, timeout);
}

2216
#ifdef CONFIG_COMPAT_32BIT_TIME
2217 2218 2219
long compat_ksys_semtimedop(int semid, struct sembuf __user *tsems,
			    unsigned int nsops,
			    const struct compat_timespec __user *timeout)
A
Al Viro 已提交
2220 2221
{
	if (timeout) {
2222 2223
		struct timespec64 ts;
		if (compat_get_timespec64(&ts, timeout))
A
Al Viro 已提交
2224 2225 2226 2227 2228
			return -EFAULT;
		return do_semtimedop(semid, tsems, nsops, &ts);
	}
	return do_semtimedop(semid, tsems, nsops, NULL);
}
2229 2230 2231 2232 2233 2234 2235

COMPAT_SYSCALL_DEFINE4(semtimedop, int, semid, struct sembuf __user *, tsems,
		       unsigned int, nsops,
		       const struct compat_timespec __user *, timeout)
{
	return compat_ksys_semtimedop(semid, tsems, nsops, timeout);
}
A
Al Viro 已提交
2236 2237
#endif

2238 2239
SYSCALL_DEFINE3(semop, int, semid, struct sembuf __user *, tsops,
		unsigned, nsops)
L
Linus Torvalds 已提交
2240
{
A
Al Viro 已提交
2241
	return do_semtimedop(semid, tsops, nsops, NULL);
L
Linus Torvalds 已提交
2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256
}

/* 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;
2257
		refcount_inc(&undo_list->refcnt);
L
Linus Torvalds 已提交
2258
		tsk->sysvsem.undo_list = undo_list;
P
Paul McQuade 已提交
2259
	} else
L
Linus Torvalds 已提交
2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278
		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)
{
2279
	struct sem_undo_list *ulp;
L
Linus Torvalds 已提交
2280

2281 2282
	ulp = tsk->sysvsem.undo_list;
	if (!ulp)
L
Linus Torvalds 已提交
2283
		return;
2284
	tsk->sysvsem.undo_list = NULL;
L
Linus Torvalds 已提交
2285

2286
	if (!refcount_dec_and_test(&ulp->refcnt))
L
Linus Torvalds 已提交
2287 2288
		return;

2289
	for (;;) {
L
Linus Torvalds 已提交
2290
		struct sem_array *sma;
2291
		struct sem_undo *un;
2292
		int semid, i;
D
Davidlohr Bueso 已提交
2293
		DEFINE_WAKE_Q(wake_q);
2294

2295 2296
		cond_resched();

2297
		rcu_read_lock();
2298 2299
		un = list_entry_rcu(ulp->list_proc.next,
				    struct sem_undo, list_proc);
2300 2301 2302 2303 2304 2305 2306
		if (&un->list_proc == &ulp->list_proc) {
			/*
			 * We must wait for freeary() before freeing this ulp,
			 * in case we raced with last sem_undo. There is a small
			 * possibility where we exit while freeary() didn't
			 * finish unlocking sem_undo_list.
			 */
2307 2308
			spin_lock(&ulp->lock);
			spin_unlock(&ulp->lock);
2309 2310 2311 2312 2313 2314
			rcu_read_unlock();
			break;
		}
		spin_lock(&ulp->lock);
		semid = un->semid;
		spin_unlock(&ulp->lock);
2315

2316
		/* exit_sem raced with IPC_RMID, nothing to do */
2317 2318
		if (semid == -1) {
			rcu_read_unlock();
2319
			continue;
2320
		}
L
Linus Torvalds 已提交
2321

2322
		sma = sem_obtain_object_check(tsk->nsproxy->ipc_ns, semid);
2323
		/* exit_sem raced with IPC_RMID, nothing to do */
2324 2325
		if (IS_ERR(sma)) {
			rcu_read_unlock();
2326
			continue;
2327
		}
L
Linus Torvalds 已提交
2328

2329
		sem_lock(sma, NULL, -1);
2330
		/* exit_sem raced with IPC_RMID, nothing to do */
2331
		if (!ipc_valid_object(&sma->sem_perm)) {
2332 2333 2334 2335
			sem_unlock(sma, -1);
			rcu_read_unlock();
			continue;
		}
2336
		un = __lookup_undo(ulp, semid);
2337 2338 2339 2340
		if (un == NULL) {
			/* exit_sem raced with IPC_RMID+semget() that created
			 * exactly the same semid. Nothing to do.
			 */
2341
			sem_unlock(sma, -1);
2342
			rcu_read_unlock();
2343 2344 2345 2346
			continue;
		}

		/* remove un from the linked lists */
2347
		ipc_assert_locked_object(&sma->sem_perm);
2348 2349
		list_del(&un->list_id);

2350 2351 2352 2353
		/* we are the last process using this ulp, acquiring ulp->lock
		 * isn't required. Besides that, we are also protected against
		 * IPC_RMID as we hold sma->sem_perm lock now
		 */
2354 2355
		list_del_rcu(&un->list_proc);

2356 2357
		/* perform adjustments registered in un */
		for (i = 0; i < sma->sem_nsems; i++) {
2358
			struct sem *semaphore = &sma->sems[i];
2359 2360
			if (un->semadj[i]) {
				semaphore->semval += un->semadj[i];
L
Linus Torvalds 已提交
2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371
				/*
				 * 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.
				 *
M
Manfred Spraul 已提交
2372
				 *	Manfred <manfred@colorfullife.com>
L
Linus Torvalds 已提交
2373
				 */
I
Ingo Molnar 已提交
2374 2375 2376 2377
				if (semaphore->semval < 0)
					semaphore->semval = 0;
				if (semaphore->semval > SEMVMX)
					semaphore->semval = SEMVMX;
2378
				ipc_update_pid(&semaphore->sempid, task_tgid(current));
L
Linus Torvalds 已提交
2379 2380 2381
			}
		}
		/* maybe some queued-up processes were waiting for this */
D
Davidlohr Bueso 已提交
2382
		do_smart_update(sma, NULL, 0, 1, &wake_q);
2383
		sem_unlock(sma, -1);
2384
		rcu_read_unlock();
D
Davidlohr Bueso 已提交
2385
		wake_up_q(&wake_q);
2386

2387
		kfree_rcu(un, rcu);
L
Linus Torvalds 已提交
2388
	}
2389
	kfree(ulp);
L
Linus Torvalds 已提交
2390 2391 2392
}

#ifdef CONFIG_PROC_FS
2393
static int sysvipc_sem_proc_show(struct seq_file *s, void *it)
L
Linus Torvalds 已提交
2394
{
2395
	struct user_namespace *user_ns = seq_user_ns(s);
2396 2397
	struct kern_ipc_perm *ipcp = it;
	struct sem_array *sma = container_of(ipcp, struct sem_array, sem_perm);
2398
	time64_t sem_otime;
2399

2400 2401 2402
	/*
	 * The proc interface isn't aware of sem_lock(), it calls
	 * ipc_lock_object() directly (in sysvipc_find_ipc).
2403 2404
	 * In order to stay compatible with sem_lock(), we must
	 * enter / leave complex_mode.
2405
	 */
2406
	complexmode_enter(sma);
2407

2408
	sem_otime = get_semotime(sma);
2409

2410
	seq_printf(s,
2411
		   "%10d %10d  %4o %10u %5u %5u %5u %5u %10llu %10llu\n",
2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422
		   sma->sem_perm.key,
		   sma->sem_perm.id,
		   sma->sem_perm.mode,
		   sma->sem_nsems,
		   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),
		   sem_otime,
		   sma->sem_ctime);

2423 2424
	complexmode_tryleave(sma);

2425
	return 0;
L
Linus Torvalds 已提交
2426 2427
}
#endif