xref: /illumos-gate/usr/src/uts/common/fs/zfs/txg.c (revision b9538c2103fde34bfe8a8c2b3e878ecd1586d182)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License (the "License").
6  * You may not use this file except in compliance with the License.
7  *
8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9  * or http://www.opensolaris.org/os/licensing.
10  * See the License for the specific language governing permissions
11  * and limitations under the License.
12  *
13  * When distributing Covered Code, include this CDDL HEADER in each
14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15  * If applicable, add the following below this CDDL HEADER, with the
16  * fields enclosed by brackets "[]" replaced with your own identifying
17  * information: Portions Copyright [yyyy] [name of copyright owner]
18  *
19  * CDDL HEADER END
20  */
21 /*
22  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
23  * Portions Copyright 2011 Martin Matuska
24  * Copyright (c) 2012, 2017 by Delphix. All rights reserved.
25  */
26 
27 #include <sys/zfs_context.h>
28 #include <sys/txg_impl.h>
29 #include <sys/dmu_impl.h>
30 #include <sys/dmu_tx.h>
31 #include <sys/dsl_pool.h>
32 #include <sys/dsl_scan.h>
33 #include <sys/zil.h>
34 #include <sys/callb.h>
35 
36 /*
37  * ZFS Transaction Groups
38  * ----------------------
39  *
40  * ZFS transaction groups are, as the name implies, groups of transactions
41  * that act on persistent state. ZFS asserts consistency at the granularity of
42  * these transaction groups. Each successive transaction group (txg) is
43  * assigned a 64-bit consecutive identifier. There are three active
44  * transaction group states: open, quiescing, or syncing. At any given time,
45  * there may be an active txg associated with each state; each active txg may
46  * either be processing, or blocked waiting to enter the next state. There may
47  * be up to three active txgs, and there is always a txg in the open state
48  * (though it may be blocked waiting to enter the quiescing state). In broad
49  * strokes, transactions -- operations that change in-memory structures -- are
50  * accepted into the txg in the open state, and are completed while the txg is
51  * in the open or quiescing states. The accumulated changes are written to
52  * disk in the syncing state.
53  *
54  * Open
55  *
56  * When a new txg becomes active, it first enters the open state. New
57  * transactions -- updates to in-memory structures -- are assigned to the
58  * currently open txg. There is always a txg in the open state so that ZFS can
59  * accept new changes (though the txg may refuse new changes if it has hit
60  * some limit). ZFS advances the open txg to the next state for a variety of
61  * reasons such as it hitting a time or size threshold, or the execution of an
62  * administrative action that must be completed in the syncing state.
63  *
64  * Quiescing
65  *
66  * After a txg exits the open state, it enters the quiescing state. The
67  * quiescing state is intended to provide a buffer between accepting new
68  * transactions in the open state and writing them out to stable storage in
69  * the syncing state. While quiescing, transactions can continue their
70  * operation without delaying either of the other states. Typically, a txg is
71  * in the quiescing state very briefly since the operations are bounded by
72  * software latencies rather than, say, slower I/O latencies. After all
73  * transactions complete, the txg is ready to enter the next state.
74  *
75  * Syncing
76  *
77  * In the syncing state, the in-memory state built up during the open and (to
78  * a lesser degree) the quiescing states is written to stable storage. The
79  * process of writing out modified data can, in turn modify more data. For
80  * example when we write new blocks, we need to allocate space for them; those
81  * allocations modify metadata (space maps)... which themselves must be
82  * written to stable storage. During the sync state, ZFS iterates, writing out
83  * data until it converges and all in-memory changes have been written out.
84  * The first such pass is the largest as it encompasses all the modified user
85  * data (as opposed to filesystem metadata). Subsequent passes typically have
86  * far less data to write as they consist exclusively of filesystem metadata.
87  *
88  * To ensure convergence, after a certain number of passes ZFS begins
89  * overwriting locations on stable storage that had been allocated earlier in
90  * the syncing state (and subsequently freed). ZFS usually allocates new
91  * blocks to optimize for large, continuous, writes. For the syncing state to
92  * converge however it must complete a pass where no new blocks are allocated
93  * since each allocation requires a modification of persistent metadata.
94  * Further, to hasten convergence, after a prescribed number of passes, ZFS
95  * also defers frees, and stops compressing.
96  *
97  * In addition to writing out user data, we must also execute synctasks during
98  * the syncing context. A synctask is the mechanism by which some
99  * administrative activities work such as creating and destroying snapshots or
100  * datasets. Note that when a synctask is initiated it enters the open txg,
101  * and ZFS then pushes that txg as quickly as possible to completion of the
102  * syncing state in order to reduce the latency of the administrative
103  * activity. To complete the syncing state, ZFS writes out a new uberblock,
104  * the root of the tree of blocks that comprise all state stored on the ZFS
105  * pool. Finally, if there is a quiesced txg waiting, we signal that it can
106  * now transition to the syncing state.
107  */
108 
109 static void txg_sync_thread(void *arg);
110 static void txg_quiesce_thread(void *arg);
111 
112 int zfs_txg_timeout = 5;	/* max seconds worth of delta per txg */
113 
114 /*
115  * Prepare the txg subsystem.
116  */
117 void
118 txg_init(dsl_pool_t *dp, uint64_t txg)
119 {
120 	tx_state_t *tx = &dp->dp_tx;
121 	int c;
122 	bzero(tx, sizeof (tx_state_t));
123 
124 	tx->tx_cpu = kmem_zalloc(max_ncpus * sizeof (tx_cpu_t), KM_SLEEP);
125 
126 	for (c = 0; c < max_ncpus; c++) {
127 		int i;
128 
129 		mutex_init(&tx->tx_cpu[c].tc_lock, NULL, MUTEX_DEFAULT, NULL);
130 		mutex_init(&tx->tx_cpu[c].tc_open_lock, NULL, MUTEX_DEFAULT,
131 		    NULL);
132 		for (i = 0; i < TXG_SIZE; i++) {
133 			cv_init(&tx->tx_cpu[c].tc_cv[i], NULL, CV_DEFAULT,
134 			    NULL);
135 			list_create(&tx->tx_cpu[c].tc_callbacks[i],
136 			    sizeof (dmu_tx_callback_t),
137 			    offsetof(dmu_tx_callback_t, dcb_node));
138 		}
139 	}
140 
141 	mutex_init(&tx->tx_sync_lock, NULL, MUTEX_DEFAULT, NULL);
142 
143 	cv_init(&tx->tx_sync_more_cv, NULL, CV_DEFAULT, NULL);
144 	cv_init(&tx->tx_sync_done_cv, NULL, CV_DEFAULT, NULL);
145 	cv_init(&tx->tx_quiesce_more_cv, NULL, CV_DEFAULT, NULL);
146 	cv_init(&tx->tx_quiesce_done_cv, NULL, CV_DEFAULT, NULL);
147 	cv_init(&tx->tx_exit_cv, NULL, CV_DEFAULT, NULL);
148 
149 	tx->tx_open_txg = txg;
150 }
151 
152 /*
153  * Close down the txg subsystem.
154  */
155 void
156 txg_fini(dsl_pool_t *dp)
157 {
158 	tx_state_t *tx = &dp->dp_tx;
159 	int c;
160 
161 	ASSERT0(tx->tx_threads);
162 
163 	mutex_destroy(&tx->tx_sync_lock);
164 
165 	cv_destroy(&tx->tx_sync_more_cv);
166 	cv_destroy(&tx->tx_sync_done_cv);
167 	cv_destroy(&tx->tx_quiesce_more_cv);
168 	cv_destroy(&tx->tx_quiesce_done_cv);
169 	cv_destroy(&tx->tx_exit_cv);
170 
171 	for (c = 0; c < max_ncpus; c++) {
172 		int i;
173 
174 		mutex_destroy(&tx->tx_cpu[c].tc_open_lock);
175 		mutex_destroy(&tx->tx_cpu[c].tc_lock);
176 		for (i = 0; i < TXG_SIZE; i++) {
177 			cv_destroy(&tx->tx_cpu[c].tc_cv[i]);
178 			list_destroy(&tx->tx_cpu[c].tc_callbacks[i]);
179 		}
180 	}
181 
182 	if (tx->tx_commit_cb_taskq != NULL)
183 		taskq_destroy(tx->tx_commit_cb_taskq);
184 
185 	kmem_free(tx->tx_cpu, max_ncpus * sizeof (tx_cpu_t));
186 
187 	bzero(tx, sizeof (tx_state_t));
188 }
189 
190 /*
191  * Start syncing transaction groups.
192  */
193 void
194 txg_sync_start(dsl_pool_t *dp)
195 {
196 	tx_state_t *tx = &dp->dp_tx;
197 
198 	mutex_enter(&tx->tx_sync_lock);
199 
200 	dprintf("pool %p\n", dp);
201 
202 	ASSERT0(tx->tx_threads);
203 
204 	tx->tx_threads = 2;
205 
206 	tx->tx_quiesce_thread = thread_create(NULL, 0, txg_quiesce_thread,
207 	    dp, 0, &p0, TS_RUN, minclsyspri);
208 
209 	/*
210 	 * The sync thread can need a larger-than-default stack size on
211 	 * 32-bit x86.  This is due in part to nested pools and
212 	 * scrub_visitbp() recursion.
213 	 */
214 	tx->tx_sync_thread = thread_create(NULL, 32<<10, txg_sync_thread,
215 	    dp, 0, &p0, TS_RUN, minclsyspri);
216 
217 	mutex_exit(&tx->tx_sync_lock);
218 }
219 
220 static void
221 txg_thread_enter(tx_state_t *tx, callb_cpr_t *cpr)
222 {
223 	CALLB_CPR_INIT(cpr, &tx->tx_sync_lock, callb_generic_cpr, FTAG);
224 	mutex_enter(&tx->tx_sync_lock);
225 }
226 
227 static void
228 txg_thread_exit(tx_state_t *tx, callb_cpr_t *cpr, kthread_t **tpp)
229 {
230 	ASSERT(*tpp != NULL);
231 	*tpp = NULL;
232 	tx->tx_threads--;
233 	cv_broadcast(&tx->tx_exit_cv);
234 	CALLB_CPR_EXIT(cpr);		/* drops &tx->tx_sync_lock */
235 	thread_exit();
236 }
237 
238 static void
239 txg_thread_wait(tx_state_t *tx, callb_cpr_t *cpr, kcondvar_t *cv, clock_t time)
240 {
241 	CALLB_CPR_SAFE_BEGIN(cpr);
242 
243 	if (time)
244 		(void) cv_timedwait(cv, &tx->tx_sync_lock,
245 		    ddi_get_lbolt() + time);
246 	else
247 		cv_wait(cv, &tx->tx_sync_lock);
248 
249 	CALLB_CPR_SAFE_END(cpr, &tx->tx_sync_lock);
250 }
251 
252 /*
253  * Stop syncing transaction groups.
254  */
255 void
256 txg_sync_stop(dsl_pool_t *dp)
257 {
258 	tx_state_t *tx = &dp->dp_tx;
259 
260 	dprintf("pool %p\n", dp);
261 	/*
262 	 * Finish off any work in progress.
263 	 */
264 	ASSERT3U(tx->tx_threads, ==, 2);
265 
266 	/*
267 	 * We need to ensure that we've vacated the deferred space_maps.
268 	 */
269 	txg_wait_synced(dp, tx->tx_open_txg + TXG_DEFER_SIZE);
270 
271 	/*
272 	 * Wake all sync threads and wait for them to die.
273 	 */
274 	mutex_enter(&tx->tx_sync_lock);
275 
276 	ASSERT3U(tx->tx_threads, ==, 2);
277 
278 	tx->tx_exiting = 1;
279 
280 	cv_broadcast(&tx->tx_quiesce_more_cv);
281 	cv_broadcast(&tx->tx_quiesce_done_cv);
282 	cv_broadcast(&tx->tx_sync_more_cv);
283 
284 	while (tx->tx_threads != 0)
285 		cv_wait(&tx->tx_exit_cv, &tx->tx_sync_lock);
286 
287 	tx->tx_exiting = 0;
288 
289 	mutex_exit(&tx->tx_sync_lock);
290 }
291 
292 uint64_t
293 txg_hold_open(dsl_pool_t *dp, txg_handle_t *th)
294 {
295 	tx_state_t *tx = &dp->dp_tx;
296 	tx_cpu_t *tc = &tx->tx_cpu[CPU_SEQID];
297 	uint64_t txg;
298 
299 	mutex_enter(&tc->tc_open_lock);
300 	txg = tx->tx_open_txg;
301 
302 	mutex_enter(&tc->tc_lock);
303 	tc->tc_count[txg & TXG_MASK]++;
304 	mutex_exit(&tc->tc_lock);
305 
306 	th->th_cpu = tc;
307 	th->th_txg = txg;
308 
309 	return (txg);
310 }
311 
312 void
313 txg_rele_to_quiesce(txg_handle_t *th)
314 {
315 	tx_cpu_t *tc = th->th_cpu;
316 
317 	ASSERT(!MUTEX_HELD(&tc->tc_lock));
318 	mutex_exit(&tc->tc_open_lock);
319 }
320 
321 void
322 txg_register_callbacks(txg_handle_t *th, list_t *tx_callbacks)
323 {
324 	tx_cpu_t *tc = th->th_cpu;
325 	int g = th->th_txg & TXG_MASK;
326 
327 	mutex_enter(&tc->tc_lock);
328 	list_move_tail(&tc->tc_callbacks[g], tx_callbacks);
329 	mutex_exit(&tc->tc_lock);
330 }
331 
332 void
333 txg_rele_to_sync(txg_handle_t *th)
334 {
335 	tx_cpu_t *tc = th->th_cpu;
336 	int g = th->th_txg & TXG_MASK;
337 
338 	mutex_enter(&tc->tc_lock);
339 	ASSERT(tc->tc_count[g] != 0);
340 	if (--tc->tc_count[g] == 0)
341 		cv_broadcast(&tc->tc_cv[g]);
342 	mutex_exit(&tc->tc_lock);
343 
344 	th->th_cpu = NULL;	/* defensive */
345 }
346 
347 /*
348  * Blocks until all transactions in the group are committed.
349  *
350  * On return, the transaction group has reached a stable state in which it can
351  * then be passed off to the syncing context.
352  */
353 static void
354 txg_quiesce(dsl_pool_t *dp, uint64_t txg)
355 {
356 	tx_state_t *tx = &dp->dp_tx;
357 	int g = txg & TXG_MASK;
358 	int c;
359 
360 	/*
361 	 * Grab all tc_open_locks so nobody else can get into this txg.
362 	 */
363 	for (c = 0; c < max_ncpus; c++)
364 		mutex_enter(&tx->tx_cpu[c].tc_open_lock);
365 
366 	ASSERT(txg == tx->tx_open_txg);
367 	tx->tx_open_txg++;
368 	tx->tx_open_time = gethrtime();
369 
370 	DTRACE_PROBE2(txg__quiescing, dsl_pool_t *, dp, uint64_t, txg);
371 	DTRACE_PROBE2(txg__opened, dsl_pool_t *, dp, uint64_t, tx->tx_open_txg);
372 
373 	/*
374 	 * Now that we've incremented tx_open_txg, we can let threads
375 	 * enter the next transaction group.
376 	 */
377 	for (c = 0; c < max_ncpus; c++)
378 		mutex_exit(&tx->tx_cpu[c].tc_open_lock);
379 
380 	/*
381 	 * Quiesce the transaction group by waiting for everyone to txg_exit().
382 	 */
383 	for (c = 0; c < max_ncpus; c++) {
384 		tx_cpu_t *tc = &tx->tx_cpu[c];
385 		mutex_enter(&tc->tc_lock);
386 		while (tc->tc_count[g] != 0)
387 			cv_wait(&tc->tc_cv[g], &tc->tc_lock);
388 		mutex_exit(&tc->tc_lock);
389 	}
390 }
391 
392 static void
393 txg_do_callbacks(list_t *cb_list)
394 {
395 	dmu_tx_do_callbacks(cb_list, 0);
396 
397 	list_destroy(cb_list);
398 
399 	kmem_free(cb_list, sizeof (list_t));
400 }
401 
402 /*
403  * Dispatch the commit callbacks registered on this txg to worker threads.
404  *
405  * If no callbacks are registered for a given TXG, nothing happens.
406  * This function creates a taskq for the associated pool, if needed.
407  */
408 static void
409 txg_dispatch_callbacks(dsl_pool_t *dp, uint64_t txg)
410 {
411 	int c;
412 	tx_state_t *tx = &dp->dp_tx;
413 	list_t *cb_list;
414 
415 	for (c = 0; c < max_ncpus; c++) {
416 		tx_cpu_t *tc = &tx->tx_cpu[c];
417 		/*
418 		 * No need to lock tx_cpu_t at this point, since this can
419 		 * only be called once a txg has been synced.
420 		 */
421 
422 		int g = txg & TXG_MASK;
423 
424 		if (list_is_empty(&tc->tc_callbacks[g]))
425 			continue;
426 
427 		if (tx->tx_commit_cb_taskq == NULL) {
428 			/*
429 			 * Commit callback taskq hasn't been created yet.
430 			 */
431 			tx->tx_commit_cb_taskq = taskq_create("tx_commit_cb",
432 			    max_ncpus, minclsyspri, max_ncpus, max_ncpus * 2,
433 			    TASKQ_PREPOPULATE);
434 		}
435 
436 		cb_list = kmem_alloc(sizeof (list_t), KM_SLEEP);
437 		list_create(cb_list, sizeof (dmu_tx_callback_t),
438 		    offsetof(dmu_tx_callback_t, dcb_node));
439 
440 		list_move_tail(cb_list, &tc->tc_callbacks[g]);
441 
442 		(void) taskq_dispatch(tx->tx_commit_cb_taskq, (task_func_t *)
443 		    txg_do_callbacks, cb_list, TQ_SLEEP);
444 	}
445 }
446 
447 static void
448 txg_sync_thread(void *arg)
449 {
450 	dsl_pool_t *dp = arg;
451 	spa_t *spa = dp->dp_spa;
452 	tx_state_t *tx = &dp->dp_tx;
453 	callb_cpr_t cpr;
454 	uint64_t start, delta;
455 
456 	txg_thread_enter(tx, &cpr);
457 
458 	start = delta = 0;
459 	for (;;) {
460 		uint64_t timeout = zfs_txg_timeout * hz;
461 		uint64_t timer;
462 		uint64_t txg;
463 
464 		/*
465 		 * We sync when we're scanning, there's someone waiting
466 		 * on us, or the quiesce thread has handed off a txg to
467 		 * us, or we have reached our timeout.
468 		 */
469 		timer = (delta >= timeout ? 0 : timeout - delta);
470 		while (!dsl_scan_active(dp->dp_scan) &&
471 		    !tx->tx_exiting && timer > 0 &&
472 		    tx->tx_synced_txg >= tx->tx_sync_txg_waiting &&
473 		    tx->tx_quiesced_txg == 0 &&
474 		    dp->dp_dirty_total < zfs_dirty_data_sync) {
475 			dprintf("waiting; tx_synced=%llu waiting=%llu dp=%p\n",
476 			    tx->tx_synced_txg, tx->tx_sync_txg_waiting, dp);
477 			txg_thread_wait(tx, &cpr, &tx->tx_sync_more_cv, timer);
478 			delta = ddi_get_lbolt() - start;
479 			timer = (delta > timeout ? 0 : timeout - delta);
480 		}
481 
482 		/*
483 		 * Wait until the quiesce thread hands off a txg to us,
484 		 * prompting it to do so if necessary.
485 		 */
486 		while (!tx->tx_exiting && tx->tx_quiesced_txg == 0) {
487 			if (tx->tx_quiesce_txg_waiting < tx->tx_open_txg+1)
488 				tx->tx_quiesce_txg_waiting = tx->tx_open_txg+1;
489 			cv_broadcast(&tx->tx_quiesce_more_cv);
490 			txg_thread_wait(tx, &cpr, &tx->tx_quiesce_done_cv, 0);
491 		}
492 
493 		if (tx->tx_exiting)
494 			txg_thread_exit(tx, &cpr, &tx->tx_sync_thread);
495 
496 		/*
497 		 * Consume the quiesced txg which has been handed off to
498 		 * us.  This may cause the quiescing thread to now be
499 		 * able to quiesce another txg, so we must signal it.
500 		 */
501 		txg = tx->tx_quiesced_txg;
502 		tx->tx_quiesced_txg = 0;
503 		tx->tx_syncing_txg = txg;
504 		DTRACE_PROBE2(txg__syncing, dsl_pool_t *, dp, uint64_t, txg);
505 		cv_broadcast(&tx->tx_quiesce_more_cv);
506 
507 		dprintf("txg=%llu quiesce_txg=%llu sync_txg=%llu\n",
508 		    txg, tx->tx_quiesce_txg_waiting, tx->tx_sync_txg_waiting);
509 		mutex_exit(&tx->tx_sync_lock);
510 
511 		start = ddi_get_lbolt();
512 		spa_sync(spa, txg);
513 		delta = ddi_get_lbolt() - start;
514 
515 		mutex_enter(&tx->tx_sync_lock);
516 		tx->tx_synced_txg = txg;
517 		tx->tx_syncing_txg = 0;
518 		DTRACE_PROBE2(txg__synced, dsl_pool_t *, dp, uint64_t, txg);
519 		cv_broadcast(&tx->tx_sync_done_cv);
520 
521 		/*
522 		 * Dispatch commit callbacks to worker threads.
523 		 */
524 		txg_dispatch_callbacks(dp, txg);
525 	}
526 }
527 
528 static void
529 txg_quiesce_thread(void *arg)
530 {
531 	dsl_pool_t *dp = arg;
532 	tx_state_t *tx = &dp->dp_tx;
533 	callb_cpr_t cpr;
534 
535 	txg_thread_enter(tx, &cpr);
536 
537 	for (;;) {
538 		uint64_t txg;
539 
540 		/*
541 		 * We quiesce when there's someone waiting on us.
542 		 * However, we can only have one txg in "quiescing" or
543 		 * "quiesced, waiting to sync" state.  So we wait until
544 		 * the "quiesced, waiting to sync" txg has been consumed
545 		 * by the sync thread.
546 		 */
547 		while (!tx->tx_exiting &&
548 		    (tx->tx_open_txg >= tx->tx_quiesce_txg_waiting ||
549 		    tx->tx_quiesced_txg != 0))
550 			txg_thread_wait(tx, &cpr, &tx->tx_quiesce_more_cv, 0);
551 
552 		if (tx->tx_exiting)
553 			txg_thread_exit(tx, &cpr, &tx->tx_quiesce_thread);
554 
555 		txg = tx->tx_open_txg;
556 		dprintf("txg=%llu quiesce_txg=%llu sync_txg=%llu\n",
557 		    txg, tx->tx_quiesce_txg_waiting,
558 		    tx->tx_sync_txg_waiting);
559 		mutex_exit(&tx->tx_sync_lock);
560 		txg_quiesce(dp, txg);
561 		mutex_enter(&tx->tx_sync_lock);
562 
563 		/*
564 		 * Hand this txg off to the sync thread.
565 		 */
566 		dprintf("quiesce done, handing off txg %llu\n", txg);
567 		tx->tx_quiesced_txg = txg;
568 		DTRACE_PROBE2(txg__quiesced, dsl_pool_t *, dp, uint64_t, txg);
569 		cv_broadcast(&tx->tx_sync_more_cv);
570 		cv_broadcast(&tx->tx_quiesce_done_cv);
571 	}
572 }
573 
574 /*
575  * Delay this thread by delay nanoseconds if we are still in the open
576  * transaction group and there is already a waiting txg quiescing or quiesced.
577  * Abort the delay if this txg stalls or enters the quiescing state.
578  */
579 void
580 txg_delay(dsl_pool_t *dp, uint64_t txg, hrtime_t delay, hrtime_t resolution)
581 {
582 	tx_state_t *tx = &dp->dp_tx;
583 	hrtime_t start = gethrtime();
584 
585 	/* don't delay if this txg could transition to quiescing immediately */
586 	if (tx->tx_open_txg > txg ||
587 	    tx->tx_syncing_txg == txg-1 || tx->tx_synced_txg == txg-1)
588 		return;
589 
590 	mutex_enter(&tx->tx_sync_lock);
591 	if (tx->tx_open_txg > txg || tx->tx_synced_txg == txg-1) {
592 		mutex_exit(&tx->tx_sync_lock);
593 		return;
594 	}
595 
596 	while (gethrtime() - start < delay &&
597 	    tx->tx_syncing_txg < txg-1 && !txg_stalled(dp)) {
598 		(void) cv_timedwait_hires(&tx->tx_quiesce_more_cv,
599 		    &tx->tx_sync_lock, delay, resolution, 0);
600 	}
601 
602 	mutex_exit(&tx->tx_sync_lock);
603 }
604 
605 void
606 txg_wait_synced(dsl_pool_t *dp, uint64_t txg)
607 {
608 	tx_state_t *tx = &dp->dp_tx;
609 
610 	ASSERT(!dsl_pool_config_held(dp));
611 
612 	mutex_enter(&tx->tx_sync_lock);
613 	ASSERT3U(tx->tx_threads, ==, 2);
614 	if (txg == 0)
615 		txg = tx->tx_open_txg + TXG_DEFER_SIZE;
616 	if (tx->tx_sync_txg_waiting < txg)
617 		tx->tx_sync_txg_waiting = txg;
618 	dprintf("txg=%llu quiesce_txg=%llu sync_txg=%llu\n",
619 	    txg, tx->tx_quiesce_txg_waiting, tx->tx_sync_txg_waiting);
620 	while (tx->tx_synced_txg < txg) {
621 		dprintf("broadcasting sync more "
622 		    "tx_synced=%llu waiting=%llu dp=%p\n",
623 		    tx->tx_synced_txg, tx->tx_sync_txg_waiting, dp);
624 		cv_broadcast(&tx->tx_sync_more_cv);
625 		cv_wait(&tx->tx_sync_done_cv, &tx->tx_sync_lock);
626 	}
627 	mutex_exit(&tx->tx_sync_lock);
628 }
629 
630 void
631 txg_wait_open(dsl_pool_t *dp, uint64_t txg)
632 {
633 	tx_state_t *tx = &dp->dp_tx;
634 
635 	ASSERT(!dsl_pool_config_held(dp));
636 
637 	mutex_enter(&tx->tx_sync_lock);
638 	ASSERT3U(tx->tx_threads, ==, 2);
639 	if (txg == 0)
640 		txg = tx->tx_open_txg + 1;
641 	if (tx->tx_quiesce_txg_waiting < txg)
642 		tx->tx_quiesce_txg_waiting = txg;
643 	dprintf("txg=%llu quiesce_txg=%llu sync_txg=%llu\n",
644 	    txg, tx->tx_quiesce_txg_waiting, tx->tx_sync_txg_waiting);
645 	while (tx->tx_open_txg < txg) {
646 		cv_broadcast(&tx->tx_quiesce_more_cv);
647 		cv_wait(&tx->tx_quiesce_done_cv, &tx->tx_sync_lock);
648 	}
649 	mutex_exit(&tx->tx_sync_lock);
650 }
651 
652 /*
653  * If there isn't a txg syncing or in the pipeline, push another txg through
654  * the pipeline by queiscing the open txg.
655  */
656 void
657 txg_kick(dsl_pool_t *dp)
658 {
659 	tx_state_t *tx = &dp->dp_tx;
660 
661 	ASSERT(!dsl_pool_config_held(dp));
662 
663 	mutex_enter(&tx->tx_sync_lock);
664 	if (tx->tx_syncing_txg == 0 &&
665 	    tx->tx_quiesce_txg_waiting <= tx->tx_open_txg &&
666 	    tx->tx_sync_txg_waiting <= tx->tx_synced_txg &&
667 	    tx->tx_quiesced_txg <= tx->tx_synced_txg) {
668 		tx->tx_quiesce_txg_waiting = tx->tx_open_txg + 1;
669 		cv_broadcast(&tx->tx_quiesce_more_cv);
670 	}
671 	mutex_exit(&tx->tx_sync_lock);
672 }
673 
674 boolean_t
675 txg_stalled(dsl_pool_t *dp)
676 {
677 	tx_state_t *tx = &dp->dp_tx;
678 	return (tx->tx_quiesce_txg_waiting > tx->tx_open_txg);
679 }
680 
681 boolean_t
682 txg_sync_waiting(dsl_pool_t *dp)
683 {
684 	tx_state_t *tx = &dp->dp_tx;
685 
686 	return (tx->tx_syncing_txg <= tx->tx_sync_txg_waiting ||
687 	    tx->tx_quiesced_txg != 0);
688 }
689 
690 /*
691  * Verify that this txg is active (open, quiescing, syncing).  Non-active
692  * txg's should not be manipulated.
693  */
694 void
695 txg_verify(spa_t *spa, uint64_t txg)
696 {
697 	dsl_pool_t *dp = spa_get_dsl(spa);
698 	if (txg <= TXG_INITIAL || txg == ZILTEST_TXG)
699 		return;
700 	ASSERT3U(txg, <=, dp->dp_tx.tx_open_txg);
701 	ASSERT3U(txg, >=, dp->dp_tx.tx_synced_txg);
702 	ASSERT3U(txg, >=, dp->dp_tx.tx_open_txg - TXG_CONCURRENT_STATES);
703 }
704 
705 /*
706  * Per-txg object lists.
707  */
708 void
709 txg_list_create(txg_list_t *tl, spa_t *spa, size_t offset)
710 {
711 	int t;
712 
713 	mutex_init(&tl->tl_lock, NULL, MUTEX_DEFAULT, NULL);
714 
715 	tl->tl_offset = offset;
716 	tl->tl_spa = spa;
717 
718 	for (t = 0; t < TXG_SIZE; t++)
719 		tl->tl_head[t] = NULL;
720 }
721 
722 void
723 txg_list_destroy(txg_list_t *tl)
724 {
725 	int t;
726 
727 	for (t = 0; t < TXG_SIZE; t++)
728 		ASSERT(txg_list_empty(tl, t));
729 
730 	mutex_destroy(&tl->tl_lock);
731 }
732 
733 boolean_t
734 txg_list_empty(txg_list_t *tl, uint64_t txg)
735 {
736 	txg_verify(tl->tl_spa, txg);
737 	return (tl->tl_head[txg & TXG_MASK] == NULL);
738 }
739 
740 /*
741  * Returns true if all txg lists are empty.
742  *
743  * Warning: this is inherently racy (an item could be added immediately
744  * after this function returns). We don't bother with the lock because
745  * it wouldn't change the semantics.
746  */
747 boolean_t
748 txg_all_lists_empty(txg_list_t *tl)
749 {
750 	for (int i = 0; i < TXG_SIZE; i++) {
751 		if (!txg_list_empty(tl, i)) {
752 			return (B_FALSE);
753 		}
754 	}
755 	return (B_TRUE);
756 }
757 
758 /*
759  * Add an entry to the list (unless it's already on the list).
760  * Returns B_TRUE if it was actually added.
761  */
762 boolean_t
763 txg_list_add(txg_list_t *tl, void *p, uint64_t txg)
764 {
765 	int t = txg & TXG_MASK;
766 	txg_node_t *tn = (txg_node_t *)((char *)p + tl->tl_offset);
767 	boolean_t add;
768 
769 	txg_verify(tl->tl_spa, txg);
770 	mutex_enter(&tl->tl_lock);
771 	add = (tn->tn_member[t] == 0);
772 	if (add) {
773 		tn->tn_member[t] = 1;
774 		tn->tn_next[t] = tl->tl_head[t];
775 		tl->tl_head[t] = tn;
776 	}
777 	mutex_exit(&tl->tl_lock);
778 
779 	return (add);
780 }
781 
782 /*
783  * Add an entry to the end of the list, unless it's already on the list.
784  * (walks list to find end)
785  * Returns B_TRUE if it was actually added.
786  */
787 boolean_t
788 txg_list_add_tail(txg_list_t *tl, void *p, uint64_t txg)
789 {
790 	int t = txg & TXG_MASK;
791 	txg_node_t *tn = (txg_node_t *)((char *)p + tl->tl_offset);
792 	boolean_t add;
793 
794 	txg_verify(tl->tl_spa, txg);
795 	mutex_enter(&tl->tl_lock);
796 	add = (tn->tn_member[t] == 0);
797 	if (add) {
798 		txg_node_t **tp;
799 
800 		for (tp = &tl->tl_head[t]; *tp != NULL; tp = &(*tp)->tn_next[t])
801 			continue;
802 
803 		tn->tn_member[t] = 1;
804 		tn->tn_next[t] = NULL;
805 		*tp = tn;
806 	}
807 	mutex_exit(&tl->tl_lock);
808 
809 	return (add);
810 }
811 
812 /*
813  * Remove the head of the list and return it.
814  */
815 void *
816 txg_list_remove(txg_list_t *tl, uint64_t txg)
817 {
818 	int t = txg & TXG_MASK;
819 	txg_node_t *tn;
820 	void *p = NULL;
821 
822 	txg_verify(tl->tl_spa, txg);
823 	mutex_enter(&tl->tl_lock);
824 	if ((tn = tl->tl_head[t]) != NULL) {
825 		ASSERT(tn->tn_member[t]);
826 		ASSERT(tn->tn_next[t] == NULL || tn->tn_next[t]->tn_member[t]);
827 		p = (char *)tn - tl->tl_offset;
828 		tl->tl_head[t] = tn->tn_next[t];
829 		tn->tn_next[t] = NULL;
830 		tn->tn_member[t] = 0;
831 	}
832 	mutex_exit(&tl->tl_lock);
833 
834 	return (p);
835 }
836 
837 /*
838  * Remove a specific item from the list and return it.
839  */
840 void *
841 txg_list_remove_this(txg_list_t *tl, void *p, uint64_t txg)
842 {
843 	int t = txg & TXG_MASK;
844 	txg_node_t *tn, **tp;
845 
846 	txg_verify(tl->tl_spa, txg);
847 	mutex_enter(&tl->tl_lock);
848 
849 	for (tp = &tl->tl_head[t]; (tn = *tp) != NULL; tp = &tn->tn_next[t]) {
850 		if ((char *)tn - tl->tl_offset == p) {
851 			*tp = tn->tn_next[t];
852 			tn->tn_next[t] = NULL;
853 			tn->tn_member[t] = 0;
854 			mutex_exit(&tl->tl_lock);
855 			return (p);
856 		}
857 	}
858 
859 	mutex_exit(&tl->tl_lock);
860 
861 	return (NULL);
862 }
863 
864 boolean_t
865 txg_list_member(txg_list_t *tl, void *p, uint64_t txg)
866 {
867 	int t = txg & TXG_MASK;
868 	txg_node_t *tn = (txg_node_t *)((char *)p + tl->tl_offset);
869 
870 	txg_verify(tl->tl_spa, txg);
871 	return (tn->tn_member[t] != 0);
872 }
873 
874 /*
875  * Walk a txg list -- only safe if you know it's not changing.
876  */
877 void *
878 txg_list_head(txg_list_t *tl, uint64_t txg)
879 {
880 	int t = txg & TXG_MASK;
881 	txg_node_t *tn = tl->tl_head[t];
882 
883 	txg_verify(tl->tl_spa, txg);
884 	return (tn == NULL ? NULL : (char *)tn - tl->tl_offset);
885 }
886 
887 void *
888 txg_list_next(txg_list_t *tl, void *p, uint64_t txg)
889 {
890 	int t = txg & TXG_MASK;
891 	txg_node_t *tn = (txg_node_t *)((char *)p + tl->tl_offset);
892 
893 	txg_verify(tl->tl_spa, txg);
894 	tn = tn->tn_next[t];
895 
896 	return (tn == NULL ? NULL : (char *)tn - tl->tl_offset);
897 }
898