xref: /openbsd/sys/kern/vfs_lockf.c (revision 8e999735)
1 /*	$OpenBSD: vfs_lockf.c,v 1.47 2022/06/01 14:16:28 visa Exp $	*/
2 /*	$NetBSD: vfs_lockf.c,v 1.7 1996/02/04 02:18:21 christos Exp $	*/
3 
4 /*
5  * Copyright (c) 1982, 1986, 1989, 1993
6  *	The Regents of the University of California.  All rights reserved.
7  *
8  * This code is derived from software contributed to Berkeley by
9  * Scooter Morris at Genentech Inc.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
14  * 1. Redistributions of source code must retain the above copyright
15  *    notice, this list of conditions and the following disclaimer.
16  * 2. Redistributions in binary form must reproduce the above copyright
17  *    notice, this list of conditions and the following disclaimer in the
18  *    documentation and/or other materials provided with the distribution.
19  * 3. Neither the name of the University nor the names of its contributors
20  *    may be used to endorse or promote products derived from this software
21  *    without specific prior written permission.
22  *
23  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33  * SUCH DAMAGE.
34  *
35  *	@(#)ufs_lockf.c	8.3 (Berkeley) 1/6/94
36  */
37 
38 #include <sys/param.h>
39 #include <sys/systm.h>
40 #include <sys/kernel.h>
41 #include <sys/proc.h>
42 #include <sys/vnode.h>
43 #include <sys/pool.h>
44 #include <sys/fcntl.h>
45 #include <sys/lockf.h>
46 #include <sys/rwlock.h>
47 #include <sys/unistd.h>
48 
49 /*
50  * The lockf structure is a kernel structure which contains the information
51  * associated with a byte range lock.  The lockf structures are linked into
52  * the inode structure. Locks are sorted by the starting byte of the lock for
53  * efficiency.
54  */
55 TAILQ_HEAD(locklist, lockf);
56 
57 struct lockf {
58 	short	lf_flags;	 /* Lock semantics: F_POSIX, F_FLOCK, F_WAIT */
59 	short	lf_type;	 /* Lock type: F_RDLCK, F_WRLCK */
60 	off_t	lf_start;	 /* The byte # of the start of the lock */
61 	off_t	lf_end;		 /* The byte # of the end of the lock (-1=EOF)*/
62 	caddr_t	lf_id;		 /* The id of the resource holding the lock */
63 	struct	lockf_state *lf_state;	/* State associated with the lock */
64 	TAILQ_ENTRY(lockf) lf_entry;
65 	struct	lockf *lf_blk;	 /* The lock that blocks us */
66 	struct	locklist lf_blkhd;	/* The list of blocked locks */
67 	TAILQ_ENTRY(lockf) lf_block; /* A request waiting for a lock */
68 	uid_t	lf_uid;		/* User ID responsible */
69 	pid_t	lf_pid;		/* POSIX - owner pid */
70 };
71 
72 struct lockf_state {
73 	TAILQ_HEAD(, lockf)	  ls_locks;	/* list of active locks */
74 	TAILQ_HEAD(, lockf)	  ls_pending;	/* list of pending locks */
75 	struct lockf_state	**ls_owner;	/* owner */
76 	int		 	  ls_refs;	/* reference counter */
77 };
78 
79 struct pool lockf_state_pool;
80 struct pool lockf_pool;
81 
82 #define SELF	0x1
83 #define OTHERS	0x2
84 
85 #ifdef LOCKF_DEBUG
86 
87 #define	DEBUG_SETLOCK		0x01
88 #define	DEBUG_CLEARLOCK		0x02
89 #define	DEBUG_GETLOCK		0x04
90 #define	DEBUG_FINDOVR		0x08
91 #define	DEBUG_SPLIT		0x10
92 #define	DEBUG_WAKELOCK		0x20
93 #define	DEBUG_LINK		0x40
94 
95 int	lockf_debug = DEBUG_SETLOCK|DEBUG_CLEARLOCK|DEBUG_WAKELOCK;
96 
97 void	lf_print(const char *, struct lockf *);
98 void	lf_printlist(const char *, struct lockf *);
99 
100 #define	DPRINTF(args, level)	if (lockf_debug & (level)) printf args
101 #define	LFPRINT(args, level)	if (lockf_debug & (level)) lf_print args
102 #else
103 #define	DPRINTF(args, level)
104 #define	LFPRINT(args, level)
105 #endif
106 
107 struct lockf *lf_alloc(uid_t, int);
108 void lf_free(struct lockf *);
109 int lf_clearlock(struct lockf *);
110 int lf_findoverlap(struct lockf *, struct lockf *, int, struct lockf **);
111 struct lockf *lf_getblock(struct lockf *, struct lockf *);
112 int lf_getlock(struct lockf *, struct flock *);
113 int lf_setlock(struct lockf *);
114 void lf_split(struct lockf *, struct lockf *);
115 void lf_wakelock(struct lockf *, int);
116 int lf_deadlock(struct lockf *);
117 void ls_ref(struct lockf_state *);
118 void ls_rele(struct lockf_state *);
119 
120 /*
121  * Serializes access to each instance of struct lockf and struct lockf_state
122  * and each pointer from a vnode to struct lockf_state.
123  */
124 struct rwlock lockf_lock = RWLOCK_INITIALIZER("lockflk");
125 
126 void
127 lf_init(void)
128 {
129 	pool_init(&lockf_state_pool, sizeof(struct lockf_state), 0, IPL_NONE,
130 	    PR_WAITOK | PR_RWLOCK, "lockfspl", NULL);
131 	pool_init(&lockf_pool, sizeof(struct lockf), 0, IPL_NONE,
132 	    PR_WAITOK | PR_RWLOCK, "lockfpl", NULL);
133 }
134 
135 void
136 ls_ref(struct lockf_state *ls)
137 {
138 	rw_assert_wrlock(&lockf_lock);
139 
140 	ls->ls_refs++;
141 }
142 
143 void
144 ls_rele(struct lockf_state *ls)
145 {
146 	rw_assert_wrlock(&lockf_lock);
147 
148 	if (--ls->ls_refs > 0)
149 		return;
150 
151 #ifdef LOCKF_DIAGNOSTIC
152 	KASSERT(TAILQ_EMPTY(&ls->ls_locks));
153 	KASSERT(TAILQ_EMPTY(&ls->ls_pending));
154 #endif
155 
156 	*ls->ls_owner = NULL;
157 	pool_put(&lockf_state_pool, ls);
158 }
159 
160 /*
161  * We enforce a limit on locks by uid, so that a single user cannot
162  * run the kernel out of memory.  For now, the limit is pretty coarse.
163  * There is no limit on root.
164  *
165  * Splitting a lock will always succeed, regardless of current allocations.
166  * If you're slightly above the limit, we still have to permit an allocation
167  * so that the unlock can succeed.  If the unlocking causes too many splits,
168  * however, you're totally cutoff.
169  */
170 int maxlocksperuid = 1024;
171 
172 /*
173  * 3 options for allowfail.
174  * 0 - always allocate.  1 - cutoff at limit.  2 - cutoff at double limit.
175  */
176 struct lockf *
177 lf_alloc(uid_t uid, int allowfail)
178 {
179 	struct uidinfo *uip;
180 	struct lockf *lock;
181 
182 	uip = uid_find(uid);
183 	if (uid && allowfail && uip->ui_lockcnt >
184 	    (allowfail == 1 ? maxlocksperuid : (maxlocksperuid * 2))) {
185 		uid_release(uip);
186 		return (NULL);
187 	}
188 	uip->ui_lockcnt++;
189 	uid_release(uip);
190 	lock = pool_get(&lockf_pool, PR_WAITOK);
191 	lock->lf_uid = uid;
192 	return (lock);
193 }
194 
195 void
196 lf_free(struct lockf *lock)
197 {
198 	struct uidinfo *uip;
199 
200 	rw_assert_wrlock(&lockf_lock);
201 
202 	LFPRINT(("lf_free", lock), DEBUG_LINK);
203 
204 #ifdef LOCKF_DIAGNOSTIC
205 	KASSERT(TAILQ_EMPTY(&lock->lf_blkhd));
206 #endif /* LOCKF_DIAGNOSTIC */
207 
208 	ls_rele(lock->lf_state);
209 
210 	uip = uid_find(lock->lf_uid);
211 	uip->ui_lockcnt--;
212 	uid_release(uip);
213 	pool_put(&lockf_pool, lock);
214 }
215 
216 
217 /*
218  * Do an advisory lock operation.
219  */
220 int
221 lf_advlock(struct lockf_state **state, off_t size, caddr_t id, int op,
222     struct flock *fl, int flags)
223 {
224 	struct proc *p = curproc;
225 	struct lockf_state *ls;
226 	struct lockf *lock;
227 	off_t start, end;
228 	int error = 0;
229 
230 	/*
231 	 * Convert the flock structure into a start and end.
232 	 */
233 	switch (fl->l_whence) {
234 	case SEEK_SET:
235 	case SEEK_CUR:
236 		/*
237 		 * Caller is responsible for adding any necessary offset
238 		 * when SEEK_CUR is used.
239 		 */
240 		start = fl->l_start;
241 		break;
242 	case SEEK_END:
243 		start = size + fl->l_start;
244 		break;
245 	default:
246 		return (EINVAL);
247 	}
248 	if (start < 0)
249 		return (EINVAL);
250 	if (fl->l_len > 0) {
251 		if (fl->l_len - 1 > LLONG_MAX - start)
252 			return (EOVERFLOW);
253 		end = start + (fl->l_len - 1);
254 	} else if (fl->l_len < 0) {
255 		if (start + fl->l_len < 0)
256 			return (EINVAL);
257 		end = start - 1;
258 		start += fl->l_len;
259 	} else {
260 		end = -1;
261 	}
262 
263 	rw_enter_write(&lockf_lock);
264 	ls = *state;
265 
266 	/*
267 	 * Avoid the common case of unlocking when inode has no locks.
268 	 */
269 	if (ls == NULL && op != F_SETLK) {
270 		fl->l_type = F_UNLCK;
271 		goto out;
272 	}
273 
274 	if (ls == NULL) {
275 		ls = pool_get(&lockf_state_pool, PR_WAITOK | PR_ZERO);
276 		ls->ls_owner = state;
277 		TAILQ_INIT(&ls->ls_locks);
278 		TAILQ_INIT(&ls->ls_pending);
279 		*state = ls;
280 	}
281 	ls_ref(ls);
282 
283 	lock = lf_alloc(p->p_ucred->cr_uid, op == F_SETLK ? 1 : 2);
284 	if (!lock) {
285 		ls_rele(ls);
286 		error = ENOLCK;
287 		goto out;
288 	}
289 	lock->lf_flags = flags;
290 	lock->lf_type = fl->l_type;
291 	lock->lf_start = start;
292 	lock->lf_end = end;
293 	lock->lf_id = id;
294 	lock->lf_state = ls;
295 	lock->lf_blk = NULL;
296 	lock->lf_pid = (flags & F_POSIX) ? p->p_p->ps_pid : -1;
297 	TAILQ_INIT(&lock->lf_blkhd);
298 
299 	switch (op) {
300 	case F_SETLK:
301 		error = lf_setlock(lock);
302 		break;
303 	case F_UNLCK:
304 		error = lf_clearlock(lock);
305 		lf_free(lock);
306 		break;
307 	case F_GETLK:
308 		error = lf_getlock(lock, fl);
309 		lf_free(lock);
310 		break;
311 	default:
312 		lf_free(lock);
313 		error = EINVAL;
314 		break;
315 	}
316 
317 out:
318 	rw_exit_write(&lockf_lock);
319 	return (error);
320 }
321 
322 /*
323  * Set a byte-range lock.
324  */
325 int
326 lf_setlock(struct lockf *lock)
327 {
328 	struct lockf *block;
329 	struct lockf *overlap, *ltmp;
330 	int ovcase, priority, needtolink, error;
331 
332 	rw_assert_wrlock(&lockf_lock);
333 
334 	LFPRINT(("lf_setlock", lock), DEBUG_SETLOCK);
335 
336 	priority = PLOCK;
337 	if (lock->lf_type == F_WRLCK)
338 		priority += 4;
339 	priority |= PCATCH;
340 	/*
341 	 * Scan lock list for this file looking for locks that would block us.
342 	 */
343 	for (;;) {
344 		block = lf_getblock(TAILQ_FIRST(&lock->lf_state->ls_locks),
345 		    lock);
346 		if (block == NULL)
347 			break;
348 
349 		if ((lock->lf_flags & F_WAIT) == 0) {
350 			lf_free(lock);
351 			return (EAGAIN);
352 		}
353 
354 		/*
355 		 * Lock is blocked, check for deadlock before proceeding.
356 		 * Note: flock style locks cover the whole file, there is no
357 		 * chance for deadlock.
358 		 */
359 		if ((lock->lf_flags & F_POSIX) && lf_deadlock(lock)) {
360 			lf_free(lock);
361 			return (EDEADLK);
362 		}
363 
364 		/*
365 		 * For flock type locks, we must first remove
366 		 * any shared locks that we hold before we sleep
367 		 * waiting for an exclusive lock.
368 		 */
369 		if ((lock->lf_flags & F_FLOCK) && lock->lf_type == F_WRLCK) {
370 			lock->lf_type = F_UNLCK;
371 			(void)lf_clearlock(lock);
372 			lock->lf_type = F_WRLCK;
373 		}
374 		/*
375 		 * Add our lock to the blocked list and sleep until we're free.
376 		 * Remember who blocked us (for deadlock detection).
377 		 */
378 		lock->lf_blk = block;
379 		LFPRINT(("lf_setlock", lock), DEBUG_SETLOCK);
380 		LFPRINT(("lf_setlock: blocking on", block), DEBUG_SETLOCK);
381 		TAILQ_INSERT_TAIL(&block->lf_blkhd, lock, lf_block);
382 		TAILQ_INSERT_TAIL(&lock->lf_state->ls_pending, lock, lf_entry);
383 		error = rwsleep_nsec(lock, &lockf_lock, priority, "lockf",
384 		    INFSLP);
385 		TAILQ_REMOVE(&lock->lf_state->ls_pending, lock, lf_entry);
386 		wakeup_one(lock->lf_state);
387 		if (lock->lf_blk != NULL) {
388 			TAILQ_REMOVE(&lock->lf_blk->lf_blkhd, lock, lf_block);
389 			lock->lf_blk = NULL;
390 		}
391 		if (error) {
392 			lf_free(lock);
393 			return (error);
394 		}
395 		if (lock->lf_flags & F_INTR) {
396 			lf_free(lock);
397 			return (EINTR);
398 		}
399 	}
400 	/*
401 	 * No blocks!!  Add the lock.  Note that we will
402 	 * downgrade or upgrade any overlapping locks this
403 	 * process already owns.
404 	 *
405 	 * Skip over locks owned by other processes.
406 	 * Handle any locks that overlap and are owned by ourselves.
407 	 */
408 	block = TAILQ_FIRST(&lock->lf_state->ls_locks);
409 	overlap = NULL;
410 	needtolink = 1;
411 	for (;;) {
412 		ovcase = lf_findoverlap(block, lock, SELF, &overlap);
413 		if (ovcase)
414 			block = TAILQ_NEXT(overlap, lf_entry);
415 		/*
416 		 * Six cases:
417 		 *	0) no overlap
418 		 *	1) overlap == lock
419 		 *	2) overlap contains lock
420 		 *	3) lock contains overlap
421 		 *	4) overlap starts before lock
422 		 *	5) overlap ends after lock
423 		 */
424 		switch (ovcase) {
425 		case 0: /* no overlap */
426 			if (needtolink) {
427 				if (overlap)	/* insert before overlap */
428 					TAILQ_INSERT_BEFORE(overlap, lock,
429 					    lf_entry);
430 				else		/* first or last lock in list */
431 					TAILQ_INSERT_TAIL(&lock->lf_state->ls_locks,
432 					    lock, lf_entry);
433 			}
434 			break;
435 		case 1: /* overlap == lock */
436 			/*
437 			 * If downgrading lock, others may be
438 			 * able to acquire it.
439 			 */
440 			if (lock->lf_type == F_RDLCK &&
441 			    overlap->lf_type == F_WRLCK)
442 				lf_wakelock(overlap, 0);
443 			overlap->lf_type = lock->lf_type;
444 			lf_free(lock);
445 			lock = overlap; /* for debug output below */
446 			break;
447 		case 2: /* overlap contains lock */
448 			/*
449 			 * Check for common starting point and different types.
450 			 */
451 			if (overlap->lf_type == lock->lf_type) {
452 				if (!needtolink)
453 					TAILQ_REMOVE(&lock->lf_state->ls_locks,
454 					    lock, lf_entry);
455 				lf_free(lock);
456 				lock = overlap; /* for debug output below */
457 				break;
458 			}
459 			if (overlap->lf_start == lock->lf_start) {
460 				if (!needtolink)
461 					TAILQ_REMOVE(&lock->lf_state->ls_locks,
462 					    lock, lf_entry);
463 				TAILQ_INSERT_BEFORE(overlap, lock, lf_entry);
464 				overlap->lf_start = lock->lf_end + 1;
465 			} else
466 				lf_split(overlap, lock);
467 			lf_wakelock(overlap, 0);
468 			break;
469 		case 3: /* lock contains overlap */
470 			/*
471 			 * If downgrading lock, others may be able to
472 			 * acquire it, otherwise take the list.
473 			 */
474 			if (lock->lf_type == F_RDLCK &&
475 			    overlap->lf_type == F_WRLCK) {
476 				lf_wakelock(overlap, 0);
477 			} else {
478 				while ((ltmp =
479 				    TAILQ_FIRST(&overlap->lf_blkhd))) {
480 					TAILQ_REMOVE(&overlap->lf_blkhd, ltmp,
481 					    lf_block);
482 					ltmp->lf_blk = lock;
483 					TAILQ_INSERT_TAIL(&lock->lf_blkhd,
484 					    ltmp, lf_block);
485 				}
486 			}
487 			/*
488 			 * Add the new lock if necessary and delete the overlap.
489 			 */
490 			if (needtolink) {
491 				TAILQ_INSERT_BEFORE(overlap, lock, lf_entry);
492 				needtolink = 0;
493 			}
494 			TAILQ_REMOVE(&lock->lf_state->ls_locks, overlap, lf_entry);
495 			lf_free(overlap);
496 			continue;
497 		case 4: /* overlap starts before lock */
498 			/*
499 			 * Add lock after overlap on the list.
500 			 */
501 			if (!needtolink)
502 				TAILQ_REMOVE(&lock->lf_state->ls_locks, lock,
503 				    lf_entry);
504 			TAILQ_INSERT_AFTER(&lock->lf_state->ls_locks, overlap,
505 			    lock, lf_entry);
506 			overlap->lf_end = lock->lf_start - 1;
507 			lf_wakelock(overlap, 0);
508 			needtolink = 0;
509 			continue;
510 		case 5: /* overlap ends after lock */
511 			/*
512 			 * Add the new lock before overlap.
513 			 */
514 			if (needtolink)
515 				TAILQ_INSERT_BEFORE(overlap, lock, lf_entry);
516 			overlap->lf_start = lock->lf_end + 1;
517 			lf_wakelock(overlap, 0);
518 			break;
519 		}
520 		break;
521 	}
522 	LFPRINT(("lf_setlock: got the lock", lock), DEBUG_SETLOCK);
523 	return (0);
524 }
525 
526 /*
527  * Remove a byte-range lock on an inode.
528  *
529  * Generally, find the lock (or an overlap to that lock)
530  * and remove it (or shrink it), then wakeup anyone we can.
531  */
532 int
533 lf_clearlock(struct lockf *lock)
534 {
535 	struct lockf *lf, *overlap;
536 	int ovcase;
537 
538 	rw_assert_wrlock(&lockf_lock);
539 
540 	lf = TAILQ_FIRST(&lock->lf_state->ls_locks);
541 	if (lf == NULL)
542 		return (0);
543 
544 	LFPRINT(("lf_clearlock", lock), DEBUG_CLEARLOCK);
545 	while ((ovcase = lf_findoverlap(lf, lock, SELF, &overlap))) {
546 		lf_wakelock(overlap, 0);
547 
548 		switch (ovcase) {
549 		case 1: /* overlap == lock */
550 			TAILQ_REMOVE(&lock->lf_state->ls_locks, overlap,
551 			    lf_entry);
552 			lf_free(overlap);
553 			break;
554 		case 2: /* overlap contains lock: split it */
555 			if (overlap->lf_start == lock->lf_start) {
556 				overlap->lf_start = lock->lf_end + 1;
557 				break;
558 			}
559 			lf_split(overlap, lock);
560 			/*
561 			 * The lock is now part of the list, lf_clearlock() must
562 			 * ensure that the lock remains detached from the list.
563 			 */
564 			TAILQ_REMOVE(&lock->lf_state->ls_locks, lock, lf_entry);
565 			break;
566 		case 3: /* lock contains overlap */
567 			lf = TAILQ_NEXT(overlap, lf_entry);
568 			TAILQ_REMOVE(&lock->lf_state->ls_locks, overlap,
569 			    lf_entry);
570 			lf_free(overlap);
571 			continue;
572 		case 4: /* overlap starts before lock */
573 			overlap->lf_end = lock->lf_start - 1;
574 			lf = TAILQ_NEXT(overlap, lf_entry);
575 			continue;
576 		case 5: /* overlap ends after lock */
577 			overlap->lf_start = lock->lf_end + 1;
578 			break;
579 		}
580 		break;
581 	}
582 	return (0);
583 }
584 
585 /*
586  * Check whether there is a blocking lock,
587  * and if so return its process identifier.
588  */
589 int
590 lf_getlock(struct lockf *lock, struct flock *fl)
591 {
592 	struct lockf *block, *lf;
593 
594 	rw_assert_wrlock(&lockf_lock);
595 
596 	LFPRINT(("lf_getlock", lock), DEBUG_CLEARLOCK);
597 
598 	lf = TAILQ_FIRST(&lock->lf_state->ls_locks);
599 	if ((block = lf_getblock(lf, lock)) != NULL) {
600 		fl->l_type = block->lf_type;
601 		fl->l_whence = SEEK_SET;
602 		fl->l_start = block->lf_start;
603 		if (block->lf_end == -1)
604 			fl->l_len = 0;
605 		else
606 			fl->l_len = block->lf_end - block->lf_start + 1;
607 		fl->l_pid = block->lf_pid;
608 	} else {
609 		fl->l_type = F_UNLCK;
610 	}
611 	return (0);
612 }
613 
614 /*
615  * Walk the list of locks for an inode and
616  * return the first blocking lock.
617  */
618 struct lockf *
619 lf_getblock(struct lockf *lf, struct lockf *lock)
620 {
621 	struct lockf *overlap;
622 
623 	rw_assert_wrlock(&lockf_lock);
624 
625 	while (lf_findoverlap(lf, lock, OTHERS, &overlap) != 0) {
626 		/*
627 		 * We've found an overlap, see if it blocks us
628 		 */
629 		if ((lock->lf_type == F_WRLCK || overlap->lf_type == F_WRLCK))
630 			return (overlap);
631 		/*
632 		 * Nope, point to the next one on the list and
633 		 * see if it blocks us
634 		 */
635 		lf = TAILQ_NEXT(overlap, lf_entry);
636 	}
637 	return (NULL);
638 }
639 
640 /*
641  * Walk the list of locks for an inode to
642  * find an overlapping lock (if any).
643  *
644  * NOTE: this returns only the FIRST overlapping lock.  There
645  *	 may be more than one.
646  */
647 int
648 lf_findoverlap(struct lockf *lf, struct lockf *lock, int type,
649     struct lockf **overlap)
650 {
651 	off_t start, end;
652 
653 	rw_assert_wrlock(&lockf_lock);
654 
655 	LFPRINT(("lf_findoverlap: looking for overlap in", lock), DEBUG_FINDOVR);
656 
657 	*overlap = lf;
658 	start = lock->lf_start;
659 	end = lock->lf_end;
660 	while (lf != NULL) {
661 		if (((type & SELF) && lf->lf_id != lock->lf_id) ||
662 		    ((type & OTHERS) && lf->lf_id == lock->lf_id)) {
663 			*overlap = lf = TAILQ_NEXT(lf, lf_entry);
664 			continue;
665 		}
666 		LFPRINT(("\tchecking", lf), DEBUG_FINDOVR);
667 		/*
668 		 * OK, check for overlap
669 		 *
670 		 * Six cases:
671 		 *	0) no overlap
672 		 *	1) overlap == lock
673 		 *	2) overlap contains lock
674 		 *	3) lock contains overlap
675 		 *	4) overlap starts before lock
676 		 *	5) overlap ends after lock
677 		 */
678 
679 		/* Case 0 */
680 		if ((lf->lf_end != -1 && start > lf->lf_end) ||
681 		    (end != -1 && lf->lf_start > end)) {
682 			DPRINTF(("no overlap\n"), DEBUG_FINDOVR);
683 			if ((type & SELF) && end != -1 && lf->lf_start > end)
684 				return (0);
685 			*overlap = lf = TAILQ_NEXT(lf, lf_entry);
686 			continue;
687 		}
688 		/* Case 1 */
689 		if ((lf->lf_start == start) && (lf->lf_end == end)) {
690 			DPRINTF(("overlap == lock\n"), DEBUG_FINDOVR);
691 			return (1);
692 		}
693 		/* Case 2 */
694 		if ((lf->lf_start <= start) &&
695 		    (lf->lf_end == -1 || (end != -1 && lf->lf_end >= end))) {
696 			DPRINTF(("overlap contains lock\n"), DEBUG_FINDOVR);
697 			return (2);
698 		}
699 		/* Case 3 */
700 		if (start <= lf->lf_start &&
701 		    (end == -1 || (lf->lf_end != -1 && end >= lf->lf_end))) {
702 			DPRINTF(("lock contains overlap\n"), DEBUG_FINDOVR);
703 			return (3);
704 		}
705 		/* Case 4 */
706 		if ((lf->lf_start < start) &&
707 		    ((lf->lf_end >= start) || (lf->lf_end == -1))) {
708 			DPRINTF(("overlap starts before lock\n"),
709 			    DEBUG_FINDOVR);
710 			return (4);
711 		}
712 		/* Case 5 */
713 		if ((lf->lf_start > start) && (end != -1) &&
714 		    ((lf->lf_end > end) || (lf->lf_end == -1))) {
715 			DPRINTF(("overlap ends after lock\n"), DEBUG_FINDOVR);
716 			return (5);
717 		}
718 		panic("lf_findoverlap: default");
719 	}
720 	return (0);
721 }
722 
723 /*
724  * Purge all locks associated with the given lock state.
725  */
726 void
727 lf_purgelocks(struct lockf_state **state)
728 {
729 	struct lockf_state *ls;
730 	struct lockf *lock;
731 
732 	rw_enter_write(&lockf_lock);
733 
734 	ls = *state;
735 	if (ls == NULL)
736 		goto out;
737 
738 	ls_ref(ls);
739 
740 	/* Interrupt blocked locks and wait for all of them to finish. */
741 	TAILQ_FOREACH(lock, &ls->ls_locks, lf_entry) {
742 		LFPRINT(("lf_purgelocks: wakeup", lock), DEBUG_SETLOCK);
743 		lf_wakelock(lock, F_INTR);
744 	}
745 	while (!TAILQ_EMPTY(&ls->ls_pending))
746 		rwsleep_nsec(ls, &lockf_lock, PLOCK, "lockfp", INFSLP);
747 
748 	/*
749 	 * Any remaining locks cannot block other locks at this point and can
750 	 * safely be removed.
751 	 */
752 	while ((lock = TAILQ_FIRST(&ls->ls_locks))) {
753 		TAILQ_REMOVE(&ls->ls_locks, lock, lf_entry);
754 		lf_free(lock);
755 	}
756 
757 	/* This is the last expected thread to hold a lock state reference. */
758 #ifdef LOCKF_DIAGNOSTIC
759 	KASSERT(ls->ls_refs == 1);
760 #endif
761 	ls_rele(ls);
762 
763 out:
764 	rw_exit_write(&lockf_lock);
765 }
766 
767 /*
768  * Split a lock and a contained region into
769  * two or three locks as necessary.
770  */
771 void
772 lf_split(struct lockf *lock1, struct lockf *lock2)
773 {
774 	struct lockf *splitlock;
775 
776 	rw_assert_wrlock(&lockf_lock);
777 
778 	LFPRINT(("lf_split", lock1), DEBUG_SPLIT);
779 	LFPRINT(("splitting from", lock2), DEBUG_SPLIT);
780 
781 	/*
782 	 * Check to see if splitting into only two pieces.
783 	 */
784 	if (lock1->lf_start == lock2->lf_start) {
785 		lock1->lf_start = lock2->lf_end + 1;
786 		TAILQ_INSERT_BEFORE(lock1, lock2, lf_entry);
787 		return;
788 	}
789 	if (lock1->lf_end == lock2->lf_end) {
790 		lock1->lf_end = lock2->lf_start - 1;
791 		TAILQ_INSERT_AFTER(&lock1->lf_state->ls_locks, lock1, lock2,
792 		    lf_entry);
793 		return;
794 	}
795 	/*
796 	 * Make a new lock consisting of the last part of
797 	 * the encompassing lock
798 	 */
799 	splitlock = lf_alloc(lock1->lf_uid, 0);
800 	splitlock->lf_flags = lock1->lf_flags;
801 	splitlock->lf_type = lock1->lf_type;
802 	splitlock->lf_start = lock2->lf_end + 1;
803 	splitlock->lf_end = lock1->lf_end;
804 	splitlock->lf_id = lock1->lf_id;
805 	splitlock->lf_state = lock1->lf_state;
806 	splitlock->lf_blk = NULL;
807 	splitlock->lf_pid = lock1->lf_pid;
808 	TAILQ_INIT(&splitlock->lf_blkhd);
809 	ls_ref(splitlock->lf_state);
810 	lock1->lf_end = lock2->lf_start - 1;
811 
812 	TAILQ_INSERT_AFTER(&lock1->lf_state->ls_locks, lock1, lock2, lf_entry);
813 	TAILQ_INSERT_AFTER(&lock1->lf_state->ls_locks, lock2, splitlock,
814 	    lf_entry);
815 }
816 
817 /*
818  * Wakeup a blocklist
819  */
820 void
821 lf_wakelock(struct lockf *lock, int flags)
822 {
823 	struct lockf *wakelock;
824 
825 	rw_assert_wrlock(&lockf_lock);
826 
827 	while ((wakelock = TAILQ_FIRST(&lock->lf_blkhd))) {
828 		TAILQ_REMOVE(&lock->lf_blkhd, wakelock, lf_block);
829 		wakelock->lf_blk = NULL;
830 		wakelock->lf_flags |= flags;
831 		wakeup_one(wakelock);
832 	}
833 }
834 
835 /*
836  * Returns non-zero if the given lock would cause a deadlock.
837  */
838 int
839 lf_deadlock(struct lockf *lock)
840 {
841 	struct lockf *block, *lf, *pending;
842 
843 	lf = TAILQ_FIRST(&lock->lf_state->ls_locks);
844 	for (; (block = lf_getblock(lf, lock)) != NULL;
845 	    lf = TAILQ_NEXT(block, lf_entry)) {
846 		if ((block->lf_flags & F_POSIX) == 0)
847 			continue;
848 
849 		TAILQ_FOREACH(pending, &lock->lf_state->ls_pending, lf_entry) {
850 			if (pending->lf_blk == NULL)
851 				continue; /* lock already unblocked */
852 
853 			if (pending->lf_pid == block->lf_pid &&
854 			    pending->lf_blk->lf_pid == lock->lf_pid)
855 				return (1);
856 		}
857 	}
858 
859 	return (0);
860 }
861 
862 #ifdef LOCKF_DEBUG
863 /*
864  * Print out a lock.
865  */
866 void
867 lf_print(const char *tag, struct lockf *lock)
868 {
869 	struct lockf	*block;
870 
871 	if (tag)
872 		printf("%s: ", tag);
873 	printf("lock %p", lock);
874 	if (lock == NULL) {
875 		printf("\n");
876 		return;
877 	}
878 	printf(", %s %p %s, start %lld, end %lld",
879 		lock->lf_flags & F_POSIX ? "posix" : "flock",
880 		lock->lf_id,
881 		lock->lf_type == F_RDLCK ? "shared" :
882 		lock->lf_type == F_WRLCK ? "exclusive" :
883 		lock->lf_type == F_UNLCK ? "unlock" :
884 		"unknown", lock->lf_start, lock->lf_end);
885 	printf(", next %p, state %p",
886 	    TAILQ_NEXT(lock, lf_entry), lock->lf_state);
887 	block = TAILQ_FIRST(&lock->lf_blkhd);
888 	if (block)
889 		printf(", block");
890 	TAILQ_FOREACH(block, &lock->lf_blkhd, lf_block)
891 		printf(" %p,", block);
892 	printf("\n");
893 }
894 
895 void
896 lf_printlist(const char *tag, struct lockf *lock)
897 {
898 	struct lockf *lf;
899 
900 	printf("%s: Lock list:\n", tag);
901 	TAILQ_FOREACH(lf, &lock->lf_state->ls_locks, lf_entry) {
902 		if (lock == lf)
903 			printf(" * ");
904 		else
905 			printf("   ");
906 		lf_print(NULL, lf);
907 	}
908 }
909 #endif /* LOCKF_DEBUG */
910