]> git.proxmox.com Git - mirror_zfs.git/blob - module/zfs/bqueue.c
Implement Redacted Send/Receive
[mirror_zfs.git] / module / zfs / bqueue.c
1 /*
2 * CDDL HEADER START
3 *
4 * This file and its contents are supplied under the terms of the
5 * Common Development and Distribution License ("CDDL"), version 1.0.
6 * You may only use this file in accordance with the terms of version
7 * 1.0 of the CDDL.
8 *
9 * A full copy of the text of the CDDL should have accompanied this
10 * source. A copy of the CDDL is also available via the Internet at
11 * http://www.illumos.org/license/CDDL.
12 *
13 * CDDL HEADER END
14 */
15 /*
16 * Copyright (c) 2014, 2018 by Delphix. All rights reserved.
17 */
18
19 #include <sys/bqueue.h>
20 #include <sys/zfs_context.h>
21
22 static inline bqueue_node_t *
23 obj2node(bqueue_t *q, void *data)
24 {
25 return ((bqueue_node_t *)((char *)data + q->bq_node_offset));
26 }
27
28 /*
29 * Initialize a blocking queue The maximum capacity of the queue is set to
30 * size. Types that are stored in a bqueue must contain a bqueue_node_t,
31 * and node_offset must be its offset from the start of the struct.
32 * fill_fraction is a performance tuning value; when the queue is full, any
33 * threads attempting to enqueue records will block. They will block until
34 * they're signaled, which will occur when the queue is at least 1/fill_fraction
35 * empty. Similar behavior occurs on dequeue; if the queue is empty, threads
36 * block. They will be signalled when the queue has 1/fill_fraction full, or
37 * when bqueue_flush is called. As a result, you must call bqueue_flush when
38 * you enqueue your final record on a thread, in case the dequeueing threads are
39 * currently blocked and that enqueue does not cause them to be awoken.
40 * Alternatively, this behavior can be disabled (causing signaling to happen
41 * immediately) by setting fill_fraction to any value larger than size.
42 * Return 0 on success, or -1 on failure.
43 */
44 int
45 bqueue_init(bqueue_t *q, uint64_t fill_fraction, uint64_t size,
46 size_t node_offset)
47 {
48 if (fill_fraction == 0) {
49 return (-1);
50 }
51 list_create(&q->bq_list, node_offset + sizeof (bqueue_node_t),
52 node_offset + offsetof(bqueue_node_t, bqn_node));
53 cv_init(&q->bq_add_cv, NULL, CV_DEFAULT, NULL);
54 cv_init(&q->bq_pop_cv, NULL, CV_DEFAULT, NULL);
55 mutex_init(&q->bq_lock, NULL, MUTEX_DEFAULT, NULL);
56 q->bq_node_offset = node_offset;
57 q->bq_size = 0;
58 q->bq_maxsize = size;
59 q->bq_fill_fraction = fill_fraction;
60 return (0);
61 }
62
63 /*
64 * Destroy a blocking queue. This function asserts that there are no
65 * elements in the queue, and no one is blocked on the condition
66 * variables.
67 */
68 void
69 bqueue_destroy(bqueue_t *q)
70 {
71 mutex_enter(&q->bq_lock);
72 ASSERT0(q->bq_size);
73 cv_destroy(&q->bq_add_cv);
74 cv_destroy(&q->bq_pop_cv);
75 list_destroy(&q->bq_list);
76 mutex_exit(&q->bq_lock);
77 mutex_destroy(&q->bq_lock);
78 }
79
80 static void
81 bqueue_enqueue_impl(bqueue_t *q, void *data, uint64_t item_size,
82 boolean_t flush)
83 {
84 ASSERT3U(item_size, >, 0);
85 ASSERT3U(item_size, <=, q->bq_maxsize);
86 mutex_enter(&q->bq_lock);
87 obj2node(q, data)->bqn_size = item_size;
88 while (q->bq_size + item_size > q->bq_maxsize) {
89 cv_wait_sig(&q->bq_add_cv, &q->bq_lock);
90 }
91 q->bq_size += item_size;
92 list_insert_tail(&q->bq_list, data);
93 if (q->bq_size >= q->bq_maxsize / q->bq_fill_fraction)
94 cv_signal(&q->bq_pop_cv);
95 if (flush)
96 cv_broadcast(&q->bq_pop_cv);
97 mutex_exit(&q->bq_lock);
98 }
99
100 /*
101 * Add data to q, consuming size units of capacity. If there is insufficient
102 * capacity to consume size units, block until capacity exists. Asserts size is
103 * > 0.
104 */
105 void
106 bqueue_enqueue(bqueue_t *q, void *data, uint64_t item_size)
107 {
108 bqueue_enqueue_impl(q, data, item_size, B_FALSE);
109 }
110
111 /*
112 * Enqueue an entry, and then flush the queue. This forces the popping threads
113 * to wake up, even if we're below the fill fraction. We have this in a single
114 * function, rather than having a separate call, because it prevents race
115 * conditions between the enqueuing thread and the dequeueing thread, where the
116 * enqueueing thread will wake up the dequeueing thread, that thread will
117 * destroy the condvar before the enqueuing thread is done.
118 */
119 void
120 bqueue_enqueue_flush(bqueue_t *q, void *data, uint64_t item_size)
121 {
122 bqueue_enqueue_impl(q, data, item_size, B_TRUE);
123 }
124
125 /*
126 * Take the first element off of q. If there are no elements on the queue, wait
127 * until one is put there. Return the removed element.
128 */
129 void *
130 bqueue_dequeue(bqueue_t *q)
131 {
132 void *ret = NULL;
133 uint64_t item_size;
134 mutex_enter(&q->bq_lock);
135 while (q->bq_size == 0) {
136 cv_wait_sig(&q->bq_pop_cv, &q->bq_lock);
137 }
138 ret = list_remove_head(&q->bq_list);
139 ASSERT3P(ret, !=, NULL);
140 item_size = obj2node(q, ret)->bqn_size;
141 q->bq_size -= item_size;
142 if (q->bq_size <= q->bq_maxsize - (q->bq_maxsize / q->bq_fill_fraction))
143 cv_signal(&q->bq_add_cv);
144 mutex_exit(&q->bq_lock);
145 return (ret);
146 }
147
148 /*
149 * Returns true if the space used is 0.
150 */
151 boolean_t
152 bqueue_empty(bqueue_t *q)
153 {
154 return (q->bq_size == 0);
155 }