1dispatch_io_create(3) BSD Library Functions Manual dispatch_io_create(3)
2
4 dispatch_io_create, dispatch_io_create_with_path, dispatch_io_close,
5 dispatch_io_set_high_water, dispatch_io_set_low_water,
6 dispatch_io_set_interval, dispatch_io_barrier — open, close and configure
7 dispatch I/O channels
8
10 #include <dispatch/dispatch.h>
11 dispatch_io_t
13 dispatch_io_create(dispatch_io_type_t type, int fd,
14 dispatch_queue_t queue, void (^cleanup_handler)(int error));
15 dispatch_io_t
17 dispatch_io_create_with_path(dispatch_io_type_t type, const char *path,
18 int oflag, mode_t mode, dispatch_queue_t queue,
19 void (^cleanup_handler)(int error));
20 void
22 dispatch_io_close(dispatch_io_t channel,
23 dispatch_io_close_flags_t flags);
24 void
26 dispatch_io_set_high_water(dispatch_io_t channel, size_t high_water);
27 void
29 dispatch_io_set_low_water(dispatch_io_t channel, size_t low_water);
30 void
32 dispatch_io_set_interval(dispatch_io_t channel, uint64_t interval,
33 dispatch_io_interval_flags_t flags);
34 void
36 dispatch_io_barrier(dispatch_io_t channel, void (^barrier)(void));
37
39 The dispatch I/O framework is an API for asynchronous read and write I/O
40 operations. It is an application of the ideas and idioms present in the
41 dispatch(3) framework to device I/O. Dispatch I/O enables an application
42 to more easily avoid blocking I/O operations and allows it to more
43 directly express its I/O requirements than by using the raw POSIX file
44 API. Dispatch I/O will make a best effort to optimize how and when asyn‐
45 chronous I/O operations are performed based on the capabilities of the
46 targeted device.
47 This page provides details on how to create and configure dispatch I/O
49 channels. Reading from and writing to these channels is covered in the
50 dispatch_io_read(3) page. The dispatch I/O framework also provides the
51 convenience functions dispatch_read(3) and dispatch_write(3) for uses
52 that do not require the full functionality provided by I/O channels.
53
55 A dispatch I/O channel represents the asynchronous I/O policy applied to
56 a file descriptor and encapsulates it for the purposes of ownership
57 tracking while I/O operations are ongoing.
58
60 Dispatch I/O channels can have one of the following types:
61 DISPATCH_IO_STREAM channels that represent a stream of bytes and
62 do not support reads and writes at arbitrary
63 offsets, such as pipes or sockets. Channels of
64 this type perform read and write operations
65 sequentially at the current file pointer posi‐
66 tion and ignore any offset specified. Depending
67 on the underlying file descriptor, read opera‐
68 tions may be performed simultaneously with
69 write operations.
70 DISPATCH_IO_RANDOM channels that represent random access files on
71 disk. Only supported for seekable file descrip‐
72 tors and paths. Channels of this type may per‐
73 form submitted read and write operations con‐
74 currently at the specified offset (interpreted
75 relative to the position of the file pointer
76 when the channel was created).
77
79 The dispatch_io_create() and dispatch_io_create_with_path() functions
80 create a dispatch I/O channel of provided type from a file descriptor fd
81 or an absolute pathname, respectively. They can be thought of as analo‐
82 gous to the fdopen(3) POSIX function and the fopen(3) function in the
83 standard C library. For a channel created from a pathname, the provided
84 path, oflag and mode parameters will be passed to open(2) when the first
85 I/O operation on the channel is ready to execute.
86 The provided cleanup_handler block will be submitted to the specified
88 queue when all I/O operations on the channel have completed and it is
89 closed or reaches the end of its lifecycle. If an error occurs during
90 channel creation, the cleanup_handler block will be submitted immediately
91 and passed an error parameter with the POSIX error encountered. If an
92 invalid type or a non-absolute path argument is specified, these func‐
93 tions will return NULL and the cleanup_handler will not be invoked. After
94 successfully creating a dispatch I/O channel from a file descriptor, the
95 application must take care not to modify that file descriptor until the
96 associated cleanup_handler is invoked, see FILEDESCRIPTOR OWNERSHIP for
97 details.
98 The dispatch_io_close() function closes a dispatch I/O channel to new
100 submissions of I/O operations. If DISPATCH_IO_STOP is passed in the flags
101 parameter, the system will in addition not perform the I/O operations
102 already submitted to the channel that are still pending and will make a
103 best effort to interrupt any ongoing operations. Handlers for operations
104 so affected will be passed the ECANCELED error code, along with any par‐
105 tial results.
106
108 Dispatch I/O channels have high-water mark, low-water mark and interval
109 configuration settings that determine if and when partial results from
110 I/O operations are delivered via their associated I/O handlers.
111 The dispatch_io_set_high_water() and dispatch_io_set_low_water() func‐
113 tions configure the water mark settings of a channel. The system will
114 read or write at least the number of bytes specified by low_water before
115 submitting an I/O handler with partial results, and will make a best
116 effort to submit an I/O handler as soon as the number of bytes read or
117 written reaches high_water.
118 The dispatch_io_set_interval() function configures the time interval at
120 which I/O handlers are submitted (measured in nanoseconds). If
121 DISPATCH_IO_STRICT_INTERVAL is passed in the flags parameter, the inter‐
122 val will be strictly observed even if there is an insufficient amount of
123 data to deliver; otherwise delivery will be skipped for intervals where
124 the amount of available data is inferior to the channel's low-water mark.
125 Note that the system may defer enqueueing interval I/O handlers by a
126 small unspecified amount of leeway in order to align with other system
127 activity for improved system performance or power consumption.
128
130 The size of data objects passed to I/O handlers for a channel will never
131 be larger than the high-water mark set on the channel; it will also never
132 be smaller than the low-water mark, except in the following cases:
133 - the final handler invocation for an I/O operation
134 - EOF was encountered
135 - the channel has an interval with the
136 DISPATCH_IO_STRICT_INTERVAL flag set
137 Bear in mind that dispatch I/O channels will typically deliver amounts of
138 data significantly higher than the low-water mark. The default value for
139 the low-water mark is unspecified, but must be assumed to allow interme‐
140 diate handler invocations. The default value for the high-water mark is
141 unlimited (i.e. SIZE_MAX). Channels that require intermediate results of
142 fixed size should have both the low-water and the high-water mark set to
143 that size. Channels that do not wish to receive any intermediate results
144 should have the low-water mark set to SIZE_MAX.
145
147 When an application creates a dispatch I/O channel from a file descriptor
148 with the dispatch_io_create() function, the system takes control of that
149 file descriptor until the channel is closed, an error occurs on the file
150 descriptor or all references to the channel are released. At that time
151 the channel's cleanup handler will be enqueued and control over the file
152 descriptor relinquished, making it safe for the application to close(2)
153 the file descriptor. While a file descriptor is under the control of a
154 dispatch I/O channel, file descriptor flags such as O_NONBLOCK will be
155 modified by the system on behalf of the application. It is an error for
156 the application to modify a file descriptor directly while it is under
157 the control of a dispatch I/O channel, but it may create further I/O
158 channels from that file descriptor or use the dispatch_read(3) and
159 dispatch_write(3) convenience functions with that file descriptor. If
160 multiple I/O channels have been created from the same file descriptor,
161 all the associated cleanup handlers will be submitted together once the
162 last channel has been closed resp. all references to those channels have
163 been released. If convenience functions have also been used on that file
164 descriptor, submission of their handlers will be tied to the submission
165 of the channel cleanup handlers as well.
166
168 The dispatch_io_barrier() function schedules a barrier operation on an
169 I/O channel. The specified barrier block will be run once, after all cur‐
170 rent I/O operations (such as read(2) or write(2)) on the underlying file
171 descriptor have finished. No new I/O operations will start until the bar‐
172 rier block finishes.
173 The barrier block may operate on the underlying file descriptor with
175 functions like fsync(2) or lseek(2). As discussed in the FILEDESCRIPTOR
176 OWNERSHIP section, the barrier block must not close(2) the file descrip‐
177 tor, and if it changes any flags on the file descriptor, it must restore
178 them before finishing.
179 There is no synchronization between a barrier block and any
181 dispatch_io_read(3) or dispatch_io_write(3) handler blocks; they may be
182 running at the same time. The barrier block itself is responsible for any
183 required synchronization.
184
186 Dispatch I/O channel objects are retained and released via calls to
187 dispatch_retain() and dispatch_release().
188
190 dispatch(3), dispatch_io_read(3), dispatch_object(3), dispatch_read(3),
191 fopen(3), open(2)
192Darwin December 1, 2010 Darwin