1			       ================
2			       CIRCULAR BUFFERS
3			       ================
4
5By: David Howells <dhowells@redhat.com>
6    Paul E. McKenney <paulmck@linux.vnet.ibm.com>
7
8
9Linux provides a number of features that can be used to implement circular
10buffering.  There are two sets of such features:
11
12 (1) Convenience functions for determining information about power-of-2 sized
13     buffers.
14
15 (2) Memory barriers for when the producer and the consumer of objects in the
16     buffer don't want to share a lock.
17
18To use these facilities, as discussed below, there needs to be just one
19producer and just one consumer.  It is possible to handle multiple producers by
20serialising them, and to handle multiple consumers by serialising them.
21
22
23Contents:
24
25 (*) What is a circular buffer?
26
27 (*) Measuring power-of-2 buffers.
28
29 (*) Using memory barriers with circular buffers.
30     - The producer.
31     - The consumer.
32
33
34==========================
35WHAT IS A CIRCULAR BUFFER?
36==========================
37
38First of all, what is a circular buffer?  A circular buffer is a buffer of
39fixed, finite size into which there are two indices:
40
41 (1) A 'head' index - the point at which the producer inserts items into the
42     buffer.
43
44 (2) A 'tail' index - the point at which the consumer finds the next item in
45     the buffer.
46
47Typically when the tail pointer is equal to the head pointer, the buffer is
48empty; and the buffer is full when the head pointer is one less than the tail
49pointer.
50
51The head index is incremented when items are added, and the tail index when
52items are removed.  The tail index should never jump the head index, and both
53indices should be wrapped to 0 when they reach the end of the buffer, thus
54allowing an infinite amount of data to flow through the buffer.
55
56Typically, items will all be of the same unit size, but this isn't strictly
57required to use the techniques below.  The indices can be increased by more
58than 1 if multiple items or variable-sized items are to be included in the
59buffer, provided that neither index overtakes the other.  The implementer must
60be careful, however, as a region more than one unit in size may wrap the end of
61the buffer and be broken into two segments.
62
63
64============================
65MEASURING POWER-OF-2 BUFFERS
66============================
67
68Calculation of the occupancy or the remaining capacity of an arbitrarily sized
69circular buffer would normally be a slow operation, requiring the use of a
70modulus (divide) instruction.  However, if the buffer is of a power-of-2 size,
71then a much quicker bitwise-AND instruction can be used instead.
72
73Linux provides a set of macros for handling power-of-2 circular buffers.  These
74can be made use of by:
75
76	#include <linux/circ_buf.h>
77
78The macros are:
79
80 (*) Measure the remaining capacity of a buffer:
81
82	CIRC_SPACE(head_index, tail_index, buffer_size);
83
84     This returns the amount of space left in the buffer[1] into which items
85     can be inserted.
86
87
88 (*) Measure the maximum consecutive immediate space in a buffer:
89
90	CIRC_SPACE_TO_END(head_index, tail_index, buffer_size);
91
92     This returns the amount of consecutive space left in the buffer[1] into
93     which items can be immediately inserted without having to wrap back to the
94     beginning of the buffer.
95
96
97 (*) Measure the occupancy of a buffer:
98
99	CIRC_CNT(head_index, tail_index, buffer_size);
100
101     This returns the number of items currently occupying a buffer[2].
102
103
104 (*) Measure the non-wrapping occupancy of a buffer:
105
106	CIRC_CNT_TO_END(head_index, tail_index, buffer_size);
107
108     This returns the number of consecutive items[2] that can be extracted from
109     the buffer without having to wrap back to the beginning of the buffer.
110
111
112Each of these macros will nominally return a value between 0 and buffer_size-1,
113however:
114
115 [1] CIRC_SPACE*() are intended to be used in the producer.  To the producer
116     they will return a lower bound as the producer controls the head index,
117     but the consumer may still be depleting the buffer on another CPU and
118     moving the tail index.
119
120     To the consumer it will show an upper bound as the producer may be busy
121     depleting the space.
122
123 [2] CIRC_CNT*() are intended to be used in the consumer.  To the consumer they
124     will return a lower bound as the consumer controls the tail index, but the
125     producer may still be filling the buffer on another CPU and moving the
126     head index.
127
128     To the producer it will show an upper bound as the consumer may be busy
129     emptying the buffer.
130
131 [3] To a third party, the order in which the writes to the indices by the
132     producer and consumer become visible cannot be guaranteed as they are
133     independent and may be made on different CPUs - so the result in such a
134     situation will merely be a guess, and may even be negative.
135
136
137===========================================
138USING MEMORY BARRIERS WITH CIRCULAR BUFFERS
139===========================================
140
141By using memory barriers in conjunction with circular buffers, you can avoid
142the need to:
143
144 (1) use a single lock to govern access to both ends of the buffer, thus
145     allowing the buffer to be filled and emptied at the same time; and
146
147 (2) use atomic counter operations.
148
149There are two sides to this: the producer that fills the buffer, and the
150consumer that empties it.  Only one thing should be filling a buffer at any one
151time, and only one thing should be emptying a buffer at any one time, but the
152two sides can operate simultaneously.
153
154
155THE PRODUCER
156------------
157
158The producer will look something like this:
159
160	spin_lock(&producer_lock);
161
162	unsigned long head = buffer->head;
163	/* The spin_unlock() and next spin_lock() provide needed ordering. */
164	unsigned long tail = ACCESS_ONCE(buffer->tail);
165
166	if (CIRC_SPACE(head, tail, buffer->size) >= 1) {
167		/* insert one item into the buffer */
168		struct item *item = buffer[head];
169
170		produce_item(item);
171
172		smp_store_release(buffer->head,
173				  (head + 1) & (buffer->size - 1));
174
175		/* wake_up() will make sure that the head is committed before
176		 * waking anyone up */
177		wake_up(consumer);
178	}
179
180	spin_unlock(&producer_lock);
181
182This will instruct the CPU that the contents of the new item must be written
183before the head index makes it available to the consumer and then instructs the
184CPU that the revised head index must be written before the consumer is woken.
185
186Note that wake_up() does not guarantee any sort of barrier unless something
187is actually awakened.  We therefore cannot rely on it for ordering.  However,
188there is always one element of the array left empty.  Therefore, the
189producer must produce two elements before it could possibly corrupt the
190element currently being read by the consumer.  Therefore, the unlock-lock
191pair between consecutive invocations of the consumer provides the necessary
192ordering between the read of the index indicating that the consumer has
193vacated a given element and the write by the producer to that same element.
194
195
196THE CONSUMER
197------------
198
199The consumer will look something like this:
200
201	spin_lock(&consumer_lock);
202
203	/* Read index before reading contents at that index. */
204	unsigned long head = smp_load_acquire(buffer->head);
205	unsigned long tail = buffer->tail;
206
207	if (CIRC_CNT(head, tail, buffer->size) >= 1) {
208
209		/* extract one item from the buffer */
210		struct item *item = buffer[tail];
211
212		consume_item(item);
213
214		/* Finish reading descriptor before incrementing tail. */
215		smp_store_release(buffer->tail,
216				  (tail + 1) & (buffer->size - 1));
217	}
218
219	spin_unlock(&consumer_lock);
220
221This will instruct the CPU to make sure the index is up to date before reading
222the new item, and then it shall make sure the CPU has finished reading the item
223before it writes the new tail pointer, which will erase the item.
224
225Note the use of ACCESS_ONCE() and smp_load_acquire() to read the
226opposition index.  This prevents the compiler from discarding and
227reloading its cached value - which some compilers will do across
228smp_read_barrier_depends().  This isn't strictly needed if you can
229be sure that the opposition index will _only_ be used the once.
230The smp_load_acquire() additionally forces the CPU to order against
231subsequent memory references.  Similarly, smp_store_release() is used
232in both algorithms to write the thread's index.  This documents the
233fact that we are writing to something that can be read concurrently,
234prevents the compiler from tearing the store, and enforces ordering
235against previous accesses.
236
237
238===============
239FURTHER READING
240===============
241
242See also Documentation/memory-barriers.txt for a description of Linux's memory
243barrier facilities.
244