1 | NAME |
1 | NAME |
2 | IO::AIO - Asynchronous Input/Output |
2 | IO::AIO - Asynchronous Input/Output |
3 | |
3 | |
4 | SYNOPSIS |
4 | SYNOPSIS |
5 | use IO::AIO; |
5 | use IO::AIO; |
|
|
6 | |
|
|
7 | aio_open "/etc/passwd", O_RDONLY, 0, sub { |
|
|
8 | my ($fh) = @_; |
|
|
9 | ... |
|
|
10 | }; |
|
|
11 | |
|
|
12 | aio_unlink "/tmp/file", sub { }; |
|
|
13 | |
|
|
14 | aio_read $fh, 30000, 1024, $buffer, 0, sub { |
|
|
15 | $_[0] > 0 or die "read error: $!"; |
|
|
16 | }; |
|
|
17 | |
|
|
18 | # Event |
|
|
19 | Event->io (fd => IO::AIO::poll_fileno, |
|
|
20 | poll => 'r', |
|
|
21 | cb => \&IO::AIO::poll_cb); |
|
|
22 | |
|
|
23 | # Glib/Gtk2 |
|
|
24 | add_watch Glib::IO IO::AIO::poll_fileno, |
|
|
25 | in => sub { IO::AIO::poll_cb; 1 }; |
|
|
26 | |
|
|
27 | # Tk |
|
|
28 | Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "", |
|
|
29 | readable => \&IO::AIO::poll_cb); |
|
|
30 | |
|
|
31 | # Danga::Socket |
|
|
32 | Danga::Socket->AddOtherFds (IO::AIO::poll_fileno => |
|
|
33 | \&IO::AIO::poll_cb); |
6 | |
34 | |
7 | DESCRIPTION |
35 | DESCRIPTION |
8 | This module implements asynchronous I/O using whatever means your |
36 | This module implements asynchronous I/O using whatever means your |
9 | operating system supports. |
37 | operating system supports. |
10 | |
38 | |
… | |
… | |
17 | currently, for example), and they would only support aio_read and |
45 | currently, for example), and they would only support aio_read and |
18 | aio_write, so the remaining functionality would have to be implemented |
46 | aio_write, so the remaining functionality would have to be implemented |
19 | using threads anyway. |
47 | using threads anyway. |
20 | |
48 | |
21 | Although the module will work with in the presence of other threads, it |
49 | Although the module will work with in the presence of other threads, it |
22 | is currently not reentrant, so use appropriate locking yourself. |
50 | is currently not reentrant, so use appropriate locking yourself, always |
|
|
51 | call "poll_cb" from within the same thread, or never call "poll_cb" (or |
|
|
52 | other "aio_" functions) recursively. |
23 | |
53 | |
24 | API NOTES |
54 | FUNCTIONS |
|
|
55 | AIO FUNCTIONS |
25 | All the "aio_*" calls are more or less thin wrappers around the syscall |
56 | All the "aio_*" calls are more or less thin wrappers around the syscall |
26 | with the same name (sans "aio_"). The arguments are similar or |
57 | with the same name (sans "aio_"). The arguments are similar or |
27 | identical, and they all accept an additional $callback argument which |
58 | identical, and they all accept an additional (and optional) $callback |
28 | must be a code reference. This code reference will get called with the |
59 | argument which must be a code reference. This code reference will get |
29 | syscall return code (e.g. most syscalls return -1 on error, unlike perl, |
60 | called with the syscall return code (e.g. most syscalls return -1 on |
30 | which usually delivers "false") as it's sole argument when the given |
61 | error, unlike perl, which usually delivers "false") as it's sole |
31 | syscall has been executed asynchronously. |
62 | argument when the given syscall has been executed asynchronously. |
32 | |
63 | |
33 | All functions that expect a filehandle will also accept a file |
64 | All functions expecting a filehandle keep a copy of the filehandle |
34 | descriptor. |
65 | internally until the request has finished. |
35 | |
66 | |
36 | The filenames you pass to these routines *must* be absolute. The reason |
67 | The pathnames you pass to these routines *must* be absolute and encoded |
37 | is that at the time the request is being executed, the current working |
68 | in byte form. The reason for the former is that at the time the request |
38 | directory could have changed. Alternatively, you can make sure that you |
69 | is being executed, the current working directory could have changed. |
39 | never change the current working directory. |
70 | Alternatively, you can make sure that you never change the current |
|
|
71 | working directory. |
40 | |
72 | |
41 | IO::AIO::min_parallel $nthreads |
73 | To encode pathnames to byte form, either make sure you either: a) always |
42 | Set the minimum number of AIO threads to $nthreads. The default is |
74 | pass in filenames you got from outside (command line, readdir etc.), b) |
43 | 1, which means a single asynchronous operation can be done at one |
75 | are ASCII or ISO 8859-1, c) use the Encode module and encode your |
44 | time (the number of outstanding operations, however, is unlimited). |
76 | pathnames to the locale (or other) encoding in effect in the user |
45 | |
77 | environment, d) use Glib::filename_from_unicode on unicode filenames or |
46 | It is recommended to keep the number of threads low, as some linux |
78 | e) use something else. |
47 | kernel versions will scale negatively with the number of threads |
|
|
48 | (higher parallelity => MUCH higher latency). |
|
|
49 | |
|
|
50 | Under normal circumstances you don't need to call this function, as |
|
|
51 | this module automatically starts a single async thread. |
|
|
52 | |
|
|
53 | IO::AIO::max_parallel $nthreads |
|
|
54 | Sets the maximum number of AIO threads to $nthreads. If more than |
|
|
55 | the specified number of threads are currently running, kill them. |
|
|
56 | This function blocks until the limit is reached. |
|
|
57 | |
|
|
58 | This module automatically runs "max_parallel 0" at program end, to |
|
|
59 | ensure that all threads are killed and that there are no outstanding |
|
|
60 | requests. |
|
|
61 | |
|
|
62 | Under normal circumstances you don't need to call this function. |
|
|
63 | |
|
|
64 | $fileno = IO::AIO::poll_fileno |
|
|
65 | Return the *request result pipe filehandle*. This filehandle must be |
|
|
66 | polled for reading by some mechanism outside this module (e.g. Event |
|
|
67 | or select, see below). If the pipe becomes readable you have to call |
|
|
68 | "poll_cb" to check the results. |
|
|
69 | |
|
|
70 | See "poll_cb" for an example. |
|
|
71 | |
|
|
72 | IO::AIO::poll_cb |
|
|
73 | Process all outstanding events on the result pipe. You have to call |
|
|
74 | this regularly. Returns the number of events processed. Returns |
|
|
75 | immediately when no events are outstanding. |
|
|
76 | |
|
|
77 | You can use Event to multiplex, e.g.: |
|
|
78 | |
|
|
79 | Event->io (fd => IO::AIO::poll_fileno, |
|
|
80 | poll => 'r', async => 1, |
|
|
81 | cb => \&IO::AIO::poll_cb); |
|
|
82 | |
|
|
83 | IO::AIO::poll_wait |
|
|
84 | Wait till the result filehandle becomes ready for reading (simply |
|
|
85 | does a select on the filehandle. This is useful if you want to |
|
|
86 | synchronously wait for some requests to finish). |
|
|
87 | |
|
|
88 | See "nreqs" for an example. |
|
|
89 | |
|
|
90 | IO::AIO::nreqs |
|
|
91 | Returns the number of requests currently outstanding. |
|
|
92 | |
|
|
93 | Example: wait till there are no outstanding requests anymore: |
|
|
94 | |
|
|
95 | IO::AIO::poll_wait, IO::AIO::poll_cb |
|
|
96 | while IO::AIO::nreqs; |
|
|
97 | |
79 | |
98 | aio_open $pathname, $flags, $mode, $callback |
80 | aio_open $pathname, $flags, $mode, $callback |
99 | Asynchronously open or create a file and call the callback with a |
81 | Asynchronously open or create a file and call the callback with a |
100 | newly created filehandle for the file. |
82 | newly created filehandle for the file. |
101 | |
83 | |
102 | The pathname passed to "aio_open" must be absolute. See API NOTES, |
84 | The pathname passed to "aio_open" must be absolute. See API NOTES, |
103 | above, for an explanation. |
85 | above, for an explanation. |
104 | |
86 | |
105 | The $mode argument is a bitmask. See the "Fcntl" module for a list. |
87 | The $flags argument is a bitmask. See the "Fcntl" module for a list. |
106 | They are the same as used in "sysopen". |
88 | They are the same as used by "sysopen". |
|
|
89 | |
|
|
90 | Likewise, $mode specifies the mode of the newly created file, if it |
|
|
91 | didn't exist and "O_CREAT" has been given, just like perl's |
|
|
92 | "sysopen", except that it is mandatory (i.e. use 0 if you don't |
|
|
93 | create new files, and 0666 or 0777 if you do). |
107 | |
94 | |
108 | Example: |
95 | Example: |
109 | |
96 | |
110 | aio_open "/etc/passwd", O_RDONLY, 0, sub { |
97 | aio_open "/etc/passwd", O_RDONLY, 0, sub { |
111 | if ($_[0]) { |
98 | if ($_[0]) { |
… | |
… | |
118 | |
105 | |
119 | aio_close $fh, $callback |
106 | aio_close $fh, $callback |
120 | Asynchronously close a file and call the callback with the result |
107 | Asynchronously close a file and call the callback with the result |
121 | code. *WARNING:* although accepted, you should not pass in a perl |
108 | code. *WARNING:* although accepted, you should not pass in a perl |
122 | filehandle here, as perl will likely close the file descriptor |
109 | filehandle here, as perl will likely close the file descriptor |
123 | itself when the filehandle is destroyed. Normally, you can safely |
110 | another time when the filehandle is destroyed. Normally, you can |
124 | call perls "close" or just let filehandles go out of scope. |
111 | safely call perls "close" or just let filehandles go out of scope. |
|
|
112 | |
|
|
113 | This is supposed to be a bug in the API, so that might change. It's |
|
|
114 | therefore best to avoid this function. |
125 | |
115 | |
126 | aio_read $fh,$offset,$length, $data,$dataoffset,$callback |
116 | aio_read $fh,$offset,$length, $data,$dataoffset,$callback |
127 | aio_write $fh,$offset,$length, $data,$dataoffset,$callback |
117 | aio_write $fh,$offset,$length, $data,$dataoffset,$callback |
128 | Reads or writes "length" bytes from the specified "fh" and "offset" |
118 | Reads or writes "length" bytes from the specified "fh" and "offset" |
129 | into the scalar given by "data" and offset "dataoffset" and calls |
119 | into the scalar given by "data" and offset "dataoffset" and calls |
130 | the callback without the actual number of bytes read (or -1 on |
120 | the callback without the actual number of bytes read (or -1 on |
131 | error, just like the syscall). |
121 | error, just like the syscall). |
132 | |
122 | |
|
|
123 | The $data scalar *MUST NOT* be modified in any way while the request |
|
|
124 | is outstanding. Modifying it can result in segfaults or WW3 (if the |
|
|
125 | necessary/optional hardware is installed). |
|
|
126 | |
133 | Example: Read 15 bytes at offset 7 into scalar $buffer, strating at |
127 | Example: Read 15 bytes at offset 7 into scalar $buffer, starting at |
134 | offset 0 within the scalar: |
128 | offset 0 within the scalar: |
135 | |
129 | |
136 | aio_read $fh, 7, 15, $buffer, 0, sub { |
130 | aio_read $fh, 7, 15, $buffer, 0, sub { |
137 | $_[0] >= 0 or die "read error: $!"; |
131 | $_[0] > 0 or die "read error: $!"; |
138 | print "read <$buffer>\n"; |
132 | print "read $_[0] bytes: <$buffer>\n"; |
139 | }; |
133 | }; |
140 | |
134 | |
141 | aio_readahead $fh,$offset,$length, $callback |
135 | aio_readahead $fh,$offset,$length, $callback |
142 | Asynchronously reads the specified byte range into the page cache, |
|
|
143 | using the "readahead" syscall. If that syscall doesn't exist the |
|
|
144 | status will be -1 and $! is set to ENOSYS. |
|
|
145 | |
|
|
146 | readahead() populates the page cache with data from a file so that |
136 | "aio_readahead" populates the page cache with data from a file so |
147 | subsequent reads from that file will not block on disk I/O. The |
137 | that subsequent reads from that file will not block on disk I/O. The |
148 | $offset argument specifies the starting point from which data is to |
138 | $offset argument specifies the starting point from which data is to |
149 | be read and $length specifies the number of bytes to be read. I/O is |
139 | be read and $length specifies the number of bytes to be read. I/O is |
150 | performed in whole pages, so that offset is effectively rounded down |
140 | performed in whole pages, so that offset is effectively rounded down |
151 | to a page boundary and bytes are read up to the next page boundary |
141 | to a page boundary and bytes are read up to the next page boundary |
152 | greater than or equal to (off-set+length). aio_readahead() does not |
142 | greater than or equal to (off-set+length). "aio_readahead" does not |
153 | read beyond the end of the file. The current file offset of the file |
143 | read beyond the end of the file. The current file offset of the file |
154 | is left unchanged. |
144 | is left unchanged. |
|
|
145 | |
|
|
146 | If that syscall doesn't exist (likely if your OS isn't Linux) it |
|
|
147 | will be emulated by simply reading the data, which would have a |
|
|
148 | similar effect. |
155 | |
149 | |
156 | aio_stat $fh_or_path, $callback |
150 | aio_stat $fh_or_path, $callback |
157 | aio_lstat $fh, $callback |
151 | aio_lstat $fh, $callback |
158 | Works like perl's "stat" or "lstat" in void context. The callback |
152 | Works like perl's "stat" or "lstat" in void context. The callback |
159 | will be called after the stat and the results will be available |
153 | will be called after the stat and the results will be available |
… | |
… | |
176 | |
170 | |
177 | aio_unlink $pathname, $callback |
171 | aio_unlink $pathname, $callback |
178 | Asynchronously unlink (delete) a file and call the callback with the |
172 | Asynchronously unlink (delete) a file and call the callback with the |
179 | result code. |
173 | result code. |
180 | |
174 | |
|
|
175 | aio_rmdir $pathname, $callback |
|
|
176 | Asynchronously rmdir (delete) a directory and call the callback with |
|
|
177 | the result code. |
|
|
178 | |
181 | aio_fsync $fh, $callback |
179 | aio_fsync $fh, $callback |
182 | Asynchronously call fsync on the given filehandle and call the |
180 | Asynchronously call fsync on the given filehandle and call the |
183 | callback with the fsync result code. |
181 | callback with the fsync result code. |
184 | |
182 | |
185 | aio_fdatasync $fh, $callback |
183 | aio_fdatasync $fh, $callback |
186 | Asynchronously call fdatasync on the given filehandle and call the |
184 | Asynchronously call fdatasync on the given filehandle and call the |
187 | callback with the fdatasync result code. |
185 | callback with the fdatasync result code. |
188 | |
186 | |
189 | BUGS |
187 | If this call isn't available because your OS lacks it or it couldn't |
190 | - could be optimized to use more semaphores instead of filehandles. |
188 | be detected, it will be emulated by calling "fsync" instead. |
|
|
189 | |
|
|
190 | SUPPORT FUNCTIONS |
|
|
191 | $fileno = IO::AIO::poll_fileno |
|
|
192 | Return the *request result pipe file descriptor*. This filehandle |
|
|
193 | must be polled for reading by some mechanism outside this module |
|
|
194 | (e.g. Event or select, see below or the SYNOPSIS). If the pipe |
|
|
195 | becomes readable you have to call "poll_cb" to check the results. |
|
|
196 | |
|
|
197 | See "poll_cb" for an example. |
|
|
198 | |
|
|
199 | IO::AIO::poll_cb |
|
|
200 | Process all outstanding events on the result pipe. You have to call |
|
|
201 | this regularly. Returns the number of events processed. Returns |
|
|
202 | immediately when no events are outstanding. |
|
|
203 | |
|
|
204 | Example: Install an Event watcher that automatically calls |
|
|
205 | IO::AIO::poll_cb with high priority: |
|
|
206 | |
|
|
207 | Event->io (fd => IO::AIO::poll_fileno, |
|
|
208 | poll => 'r', async => 1, |
|
|
209 | cb => \&IO::AIO::poll_cb); |
|
|
210 | |
|
|
211 | IO::AIO::poll_wait |
|
|
212 | Wait till the result filehandle becomes ready for reading (simply |
|
|
213 | does a "select" on the filehandle. This is useful if you want to |
|
|
214 | synchronously wait for some requests to finish). |
|
|
215 | |
|
|
216 | See "nreqs" for an example. |
|
|
217 | |
|
|
218 | IO::AIO::nreqs |
|
|
219 | Returns the number of requests currently outstanding (i.e. for which |
|
|
220 | their callback has not been invoked yet). |
|
|
221 | |
|
|
222 | Example: wait till there are no outstanding requests anymore: |
|
|
223 | |
|
|
224 | IO::AIO::poll_wait, IO::AIO::poll_cb |
|
|
225 | while IO::AIO::nreqs; |
|
|
226 | |
|
|
227 | IO::AIO::flush |
|
|
228 | Wait till all outstanding AIO requests have been handled. |
|
|
229 | |
|
|
230 | Strictly equivalent to: |
|
|
231 | |
|
|
232 | IO::AIO::poll_wait, IO::AIO::poll_cb |
|
|
233 | while IO::AIO::nreqs; |
|
|
234 | |
|
|
235 | IO::AIO::poll |
|
|
236 | Waits until some requests have been handled. |
|
|
237 | |
|
|
238 | Strictly equivalent to: |
|
|
239 | |
|
|
240 | IO::AIO::poll_wait, IO::AIO::poll_cb |
|
|
241 | if IO::AIO::nreqs; |
|
|
242 | |
|
|
243 | IO::AIO::min_parallel $nthreads |
|
|
244 | Set the minimum number of AIO threads to $nthreads. The current |
|
|
245 | default is 4, which means four asynchronous operations can be done |
|
|
246 | at one time (the number of outstanding operations, however, is |
|
|
247 | unlimited). |
|
|
248 | |
|
|
249 | IO::AIO starts threads only on demand, when an AIO request is queued |
|
|
250 | and no free thread exists. |
|
|
251 | |
|
|
252 | It is recommended to keep the number of threads low, as some Linux |
|
|
253 | kernel versions will scale negatively with the number of threads |
|
|
254 | (higher parallelity => MUCH higher latency). With current Linux 2.6 |
|
|
255 | versions, 4-32 threads should be fine. |
|
|
256 | |
|
|
257 | Under most circumstances you don't need to call this function, as |
|
|
258 | the module selects a default that is suitable for low to moderate |
|
|
259 | load. |
|
|
260 | |
|
|
261 | IO::AIO::max_parallel $nthreads |
|
|
262 | Sets the maximum number of AIO threads to $nthreads. If more than |
|
|
263 | the specified number of threads are currently running, this function |
|
|
264 | kills them. This function blocks until the limit is reached. |
|
|
265 | |
|
|
266 | While $nthreads are zero, aio requests get queued but not executed |
|
|
267 | until the number of threads has been increased again. |
|
|
268 | |
|
|
269 | This module automatically runs "max_parallel 0" at program end, to |
|
|
270 | ensure that all threads are killed and that there are no outstanding |
|
|
271 | requests. |
|
|
272 | |
|
|
273 | Under normal circumstances you don't need to call this function. |
|
|
274 | |
|
|
275 | $oldnreqs = IO::AIO::max_outstanding $nreqs |
|
|
276 | Sets the maximum number of outstanding requests to $nreqs. If you |
|
|
277 | try to queue up more than this number of requests, the caller will |
|
|
278 | block until some requests have been handled. |
|
|
279 | |
|
|
280 | The default is very large, so normally there is no practical limit. |
|
|
281 | If you queue up many requests in a loop it often improves speed if |
|
|
282 | you set this to a relatively low number, such as 100. |
|
|
283 | |
|
|
284 | Under normal circumstances you don't need to call this function. |
|
|
285 | |
|
|
286 | FORK BEHAVIOUR |
|
|
287 | Before the fork, IO::AIO enters a quiescent state where no requests can |
|
|
288 | be added in other threads and no results will be processed. After the |
|
|
289 | fork the parent simply leaves the quiescent state and continues |
|
|
290 | request/result processing, while the child clears the request/result |
|
|
291 | queue (so the requests started before the fork will only be handled in |
|
|
292 | the parent). Threats will be started on demand until the limit ste in |
|
|
293 | the parent process has been reached again. |
191 | |
294 | |
192 | SEE ALSO |
295 | SEE ALSO |
193 | Coro, Linux::AIO. |
296 | Coro, Linux::AIO. |
194 | |
297 | |
195 | AUTHOR |
298 | AUTHOR |