ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/IO-AIO/AIO.pm
(Generate patch)

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.2 by root, Sun Jul 10 18:16:49 2005 UTC vs.
Revision 1.32 by root, Wed Aug 17 05:26:20 2005 UTC

3IO::AIO - Asynchronous Input/Output 3IO::AIO - Asynchronous Input/Output
4 4
5=head1 SYNOPSIS 5=head1 SYNOPSIS
6 6
7 use IO::AIO; 7 use IO::AIO;
8
9 aio_open "/etc/passwd", O_RDONLY, 0, sub {
10 my ($fh) = @_;
11 ...
12 };
13
14 aio_unlink "/tmp/file", sub { };
15
16 aio_read $fh, 30000, 1024, $buffer, 0, sub {
17 $_[0] > 0 or die "read error: $!";
18 };
19
20 # Event
21 Event->io (fd => IO::AIO::poll_fileno,
22 poll => 'r',
23 cb => \&IO::AIO::poll_cb);
24
25 # Glib/Gtk2
26 add_watch Glib::IO IO::AIO::poll_fileno,
27 in => sub { IO::AIO::poll_cb; 1 };
28
29 # Tk
30 Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
31 readable => \&IO::AIO::poll_cb);
32
33 # Danga::Socket
34 Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
35 \&IO::AIO::poll_cb);
36
8 37
9=head1 DESCRIPTION 38=head1 DESCRIPTION
10 39
11This module implements asynchronous I/O using whatever means your 40This module implements asynchronous I/O using whatever means your
12operating system supports. 41operating system supports.
19not well-supported (Linux doesn't allow them on normal files currently, 48not well-supported (Linux doesn't allow them on normal files currently,
20for example), and they would only support aio_read and aio_write, so the 49for example), and they would only support aio_read and aio_write, so the
21remaining functionality would have to be implemented using threads anyway. 50remaining functionality would have to be implemented using threads anyway.
22 51
23Although the module will work with in the presence of other threads, it is 52Although the module will work with in the presence of other threads, it is
24currently not reentrant, so use appropriate locking yourself. 53currently not reentrant, so use appropriate locking yourself, always call
25 54C<poll_cb> from within the same thread, or never call C<poll_cb> (or other
26=head2 API NOTES 55C<aio_> functions) recursively.
27
28All the C<aio_*> calls are more or less thin wrappers around the syscall
29with the same name (sans C<aio_>). The arguments are similar or identical,
30and they all accept an additional C<$callback> argument which must be
31a code reference. This code reference will get called with the syscall
32return code (e.g. most syscalls return C<-1> on error, unlike perl, which
33usually delivers "false") as it's sole argument when the given syscall has
34been executed asynchronously.
35
36All functions that expect a filehandle will also accept a file descriptor.
37
38The filenames you pass to these routines I<must> be absolute. The reason
39is that at the time the request is being executed, the current working
40directory could have changed. Alternatively, you can make sure that you
41never change the current working directory.
42
43=over 4
44 56
45=cut 57=cut
46 58
47package IO::AIO; 59package IO::AIO;
48 60
61no warnings;
62
49use base 'Exporter'; 63use base 'Exporter';
50 64
51use Fcntl (); 65use Fcntl ();
52 66
53BEGIN { 67BEGIN {
54 $VERSION = 0.1; 68 $VERSION = 1.3;
69
55 70
56 @EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink 71 @EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink
57 aio_fsync aio_fdatasync aio_readahead); 72 aio_rmdir aio_symlink aio_fsync aio_fdatasync aio_readahead);
58 @EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel nreqs); 73 @EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel max_outstanding nreqs);
59 74
60 require XSLoader; 75 require XSLoader;
61 XSLoader::load IO::AIO, $VERSION; 76 XSLoader::load IO::AIO, $VERSION;
62} 77}
63 78
64=item IO::AIO::min_parallel $nthreads 79=head1 FUNCTIONS
65 80
66Set the minimum number of AIO threads to C<$nthreads>. The default is 81=head2 AIO FUNCTIONS
67C<1>, which means a single asynchronous operation can be done at one time
68(the number of outstanding operations, however, is unlimited).
69 82
70It is recommended to keep the number of threads low, as some linux 83All the C<aio_*> calls are more or less thin wrappers around the syscall
71kernel versions will scale negatively with the number of threads (higher 84with the same name (sans C<aio_>). The arguments are similar or identical,
72parallelity => MUCH higher latency). 85and they all accept an additional (and optional) C<$callback> argument
86which must be a code reference. This code reference will get called with
87the syscall return code (e.g. most syscalls return C<-1> on error, unlike
88perl, which usually delivers "false") as it's sole argument when the given
89syscall has been executed asynchronously.
73 90
74Under normal circumstances you don't need to call this function, as this 91All functions expecting a filehandle keep a copy of the filehandle
75module automatically starts a single async thread. 92internally until the request has finished.
76 93
77=item IO::AIO::max_parallel $nthreads 94The pathnames you pass to these routines I<must> be absolute and
95encoded in byte form. The reason for the former is that at the time the
96request is being executed, the current working directory could have
97changed. Alternatively, you can make sure that you never change the
98current working directory.
78 99
79Sets the maximum number of AIO threads to C<$nthreads>. If more than 100To encode pathnames to byte form, either make sure you either: a)
80the specified number of threads are currently running, kill them. This 101always pass in filenames you got from outside (command line, readdir
81function blocks until the limit is reached. 102etc.), b) are ASCII or ISO 8859-1, c) use the Encode module and encode
103your pathnames to the locale (or other) encoding in effect in the user
104environment, d) use Glib::filename_from_unicode on unicode filenames or e)
105use something else.
82 106
83This module automatically runs C<max_parallel 0> at program end, to ensure 107=over 4
84that all threads are killed and that there are no outstanding requests.
85
86Under normal circumstances you don't need to call this function.
87
88=item $fileno = IO::AIO::poll_fileno
89
90Return the I<request result pipe filehandle>. This filehandle must be
91polled for reading by some mechanism outside this module (e.g. Event
92or select, see below). If the pipe becomes readable you have to call
93C<poll_cb> to check the results.
94
95See C<poll_cb> for an example.
96
97=item IO::AIO::poll_cb
98
99Process all outstanding events on the result pipe. You have to call this
100regularly. Returns the number of events processed. Returns immediately
101when no events are outstanding.
102
103You can use Event to multiplex, e.g.:
104
105 Event->io (fd => IO::AIO::poll_fileno,
106 poll => 'r', async => 1,
107 cb => \&IO::AIO::poll_cb);
108
109=item IO::AIO::poll_wait
110
111Wait till the result filehandle becomes ready for reading (simply does a
112select on the filehandle. This is useful if you want to synchronously wait
113for some requests to finish).
114
115See C<nreqs> for an example.
116
117=item IO::AIO::nreqs
118
119Returns the number of requests currently outstanding.
120
121Example: wait till there are no outstanding requests anymore:
122
123 IO::AIO::poll_wait, IO::AIO::poll_cb
124 while IO::AIO::nreqs;
125 108
126=item aio_open $pathname, $flags, $mode, $callback 109=item aio_open $pathname, $flags, $mode, $callback
127 110
128Asynchronously open or create a file and call the callback with a newly 111Asynchronously open or create a file and call the callback with a newly
129created filehandle for the file. 112created filehandle for the file.
130 113
131The pathname passed to C<aio_open> must be absolute. See API NOTES, above, 114The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
132for an explanation. 115for an explanation.
133 116
134The C<$mode> argument is a bitmask. See the C<Fcntl> module for a 117The C<$flags> argument is a bitmask. See the C<Fcntl> module for a
135list. They are the same as used in C<sysopen>. 118list. They are the same as used by C<sysopen>.
119
120Likewise, C<$mode> specifies the mode of the newly created file, if it
121didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
122except that it is mandatory (i.e. use C<0> if you don't create new files,
123and C<0666> or C<0777> if you do).
136 124
137Example: 125Example:
138 126
139 aio_open "/etc/passwd", O_RDONLY, 0, sub { 127 aio_open "/etc/passwd", O_RDONLY, 0, sub {
140 if ($_[0]) { 128 if ($_[0]) {
147 135
148=item aio_close $fh, $callback 136=item aio_close $fh, $callback
149 137
150Asynchronously close a file and call the callback with the result 138Asynchronously close a file and call the callback with the result
151code. I<WARNING:> although accepted, you should not pass in a perl 139code. I<WARNING:> although accepted, you should not pass in a perl
152filehandle here, as perl will likely close the file descriptor itself when 140filehandle here, as perl will likely close the file descriptor another
153the filehandle is destroyed. Normally, you can safely call perls C<close> 141time when the filehandle is destroyed. Normally, you can safely call perls
154or just let filehandles go out of scope. 142C<close> or just let filehandles go out of scope.
143
144This is supposed to be a bug in the API, so that might change. It's
145therefore best to avoid this function.
155 146
156=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback 147=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback
157 148
158=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback 149=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback
159 150
160Reads or writes C<length> bytes from the specified C<fh> and C<offset> 151Reads or writes C<length> bytes from the specified C<fh> and C<offset>
161into the scalar given by C<data> and offset C<dataoffset> and calls the 152into the scalar given by C<data> and offset C<dataoffset> and calls the
162callback without the actual number of bytes read (or -1 on error, just 153callback without the actual number of bytes read (or -1 on error, just
163like the syscall). 154like the syscall).
164 155
156The C<$data> scalar I<MUST NOT> be modified in any way while the request
157is outstanding. Modifying it can result in segfaults or WW3 (if the
158necessary/optional hardware is installed).
159
165Example: Read 15 bytes at offset 7 into scalar C<$buffer>, strating at 160Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
166offset C<0> within the scalar: 161offset C<0> within the scalar:
167 162
168 aio_read $fh, 7, 15, $buffer, 0, sub { 163 aio_read $fh, 7, 15, $buffer, 0, sub {
169 $_[0] >= 0 or die "read error: $!"; 164 $_[0] > 0 or die "read error: $!";
170 print "read <$buffer>\n"; 165 print "read $_[0] bytes: <$buffer>\n";
171 }; 166 };
172 167
173=item aio_readahead $fh,$offset,$length, $callback 168=item aio_readahead $fh,$offset,$length, $callback
174 169
175Asynchronously reads the specified byte range into the page cache, using
176the C<readahead> syscall. If that syscall doesn't exist the status will be
177C<-1> and C<$!> is set to ENOSYS.
178
179readahead() populates the page cache with data from a file so that 170C<aio_readahead> populates the page cache with data from a file so that
180subsequent reads from that file will not block on disk I/O. The C<$offset> 171subsequent reads from that file will not block on disk I/O. The C<$offset>
181argument specifies the starting point from which data is to be read and 172argument specifies the starting point from which data is to be read and
182C<$length> specifies the number of bytes to be read. I/O is performed in 173C<$length> specifies the number of bytes to be read. I/O is performed in
183whole pages, so that offset is effectively rounded down to a page boundary 174whole pages, so that offset is effectively rounded down to a page boundary
184and bytes are read up to the next page boundary greater than or equal to 175and bytes are read up to the next page boundary greater than or equal to
185(off-set+length). aio_readahead() does not read beyond the end of the 176(off-set+length). C<aio_readahead> does not read beyond the end of the
186file. The current file offset of the file is left unchanged. 177file. The current file offset of the file is left unchanged.
178
179If that syscall doesn't exist (likely if your OS isn't Linux) it will be
180emulated by simply reading the data, which would have a similar effect.
187 181
188=item aio_stat $fh_or_path, $callback 182=item aio_stat $fh_or_path, $callback
189 183
190=item aio_lstat $fh, $callback 184=item aio_lstat $fh, $callback
191 185
210=item aio_unlink $pathname, $callback 204=item aio_unlink $pathname, $callback
211 205
212Asynchronously unlink (delete) a file and call the callback with the 206Asynchronously unlink (delete) a file and call the callback with the
213result code. 207result code.
214 208
209=item aio_rmdir $pathname, $callback
210
211Asynchronously rmdir (delete) a directory and call the callback with the
212result code.
213
215=item aio_fsync $fh, $callback 214=item aio_fsync $fh, $callback
216 215
217Asynchronously call fsync on the given filehandle and call the callback 216Asynchronously call fsync on the given filehandle and call the callback
218with the fsync result code. 217with the fsync result code.
219 218
220=item aio_fdatasync $fh, $callback 219=item aio_fdatasync $fh, $callback
221 220
222Asynchronously call fdatasync on the given filehandle and call the 221Asynchronously call fdatasync on the given filehandle and call the
223callback with the fdatasync result code. 222callback with the fdatasync result code.
223
224If this call isn't available because your OS lacks it or it couldn't be
225detected, it will be emulated by calling C<fsync> instead.
226
227=back
228
229=head2 SUPPORT FUNCTIONS
230
231=over 4
232
233=item $fileno = IO::AIO::poll_fileno
234
235Return the I<request result pipe file descriptor>. This filehandle must be
236polled for reading by some mechanism outside this module (e.g. Event or
237select, see below or the SYNOPSIS). If the pipe becomes readable you have
238to call C<poll_cb> to check the results.
239
240See C<poll_cb> for an example.
241
242=item IO::AIO::poll_cb
243
244Process all outstanding events on the result pipe. You have to call this
245regularly. Returns the number of events processed. Returns immediately
246when no events are outstanding.
247
248Example: Install an Event watcher that automatically calls
249IO::AIO::poll_cb with high priority:
250
251 Event->io (fd => IO::AIO::poll_fileno,
252 poll => 'r', async => 1,
253 cb => \&IO::AIO::poll_cb);
254
255=item IO::AIO::poll_wait
256
257Wait till the result filehandle becomes ready for reading (simply does a
258C<select> on the filehandle. This is useful if you want to synchronously wait
259for some requests to finish).
260
261See C<nreqs> for an example.
262
263=item IO::AIO::nreqs
264
265Returns the number of requests currently outstanding (i.e. for which their
266callback has not been invoked yet).
267
268Example: wait till there are no outstanding requests anymore:
269
270 IO::AIO::poll_wait, IO::AIO::poll_cb
271 while IO::AIO::nreqs;
272
273=item IO::AIO::flush
274
275Wait till all outstanding AIO requests have been handled.
276
277Strictly equivalent to:
278
279 IO::AIO::poll_wait, IO::AIO::poll_cb
280 while IO::AIO::nreqs;
281
282=item IO::AIO::poll
283
284Waits until some requests have been handled.
285
286Strictly equivalent to:
287
288 IO::AIO::poll_wait, IO::AIO::poll_cb
289 if IO::AIO::nreqs;
290
291=item IO::AIO::min_parallel $nthreads
292
293Set the minimum number of AIO threads to C<$nthreads>. The default is
294C<1>, which means a single asynchronous operation can be done at one time
295(the number of outstanding operations, however, is unlimited).
296
297It is recommended to keep the number of threads low, as some Linux
298kernel versions will scale negatively with the number of threads (higher
299parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32
300threads should be fine.
301
302Under normal circumstances you don't need to call this function, as this
303module automatically starts some threads (the exact number might change,
304and is currently 4).
305
306=item IO::AIO::max_parallel $nthreads
307
308Sets the maximum number of AIO threads to C<$nthreads>. If more than
309the specified number of threads are currently running, kill them. This
310function blocks until the limit is reached.
311
312This module automatically runs C<max_parallel 0> at program end, to ensure
313that all threads are killed and that there are no outstanding requests.
314
315Under normal circumstances you don't need to call this function.
316
317=item $oldnreqs = IO::AIO::max_outstanding $nreqs
318
319Sets the maximum number of outstanding requests to C<$nreqs>. If you
320try to queue up more than this number of requests, the caller will block until
321some requests have been handled.
322
323The default is very large, so normally there is no practical limit. If you
324queue up many requests in a loop it it often improves speed if you set
325this to a relatively low number, such as C<100>.
326
327Under normal circumstances you don't need to call this function.
328
329=back
224 330
225=cut 331=cut
226 332
227# support function to convert a fd into a perl filehandle 333# support function to convert a fd into a perl filehandle
228sub _fd2fh { 334sub _fd2fh {
229 return undef if $_[0] < 0; 335 return undef if $_[0] < 0;
230 336
231 # try to be perl5.6-compatible 337 # try to generate nice filehandles
232 local *AIO_FH; 338 my $sym = "IO::AIO::fd#$_[0]";
233 open AIO_FH, "+<&=$_[0]" 339 local *$sym;
340
341 open *$sym, "+<&=$_[0]" # usually works under any unix
342 or open *$sym, "<&=$_[0]" # cygwin needs this
343 or open *$sym, ">&=$_[0]" # or this
234 or return undef; 344 or return undef;
235 345
236 *AIO_FH 346 *$sym
237} 347}
238 348
239min_parallel 4; 349min_parallel 4;
240 350
241END { 351END {
242 max_parallel 0; 352 max_parallel 0;
243} 353}
244 354
2451; 3551;
246 356
247=back 357=head2 FORK BEHAVIOUR
248 358
249=head1 BUGS 359Before the fork IO::AIO enters a quiescent state where no requests can be
250 360added in other threads and no results will be processed. After the fork
251 - could be optimized to use more semaphores instead of filehandles. 361the parent simply leaves the quiescent state and continues request/result
362processing, while the child clears the request/result queue and starts the
363same number of threads as were in use by the parent.
252 364
253=head1 SEE ALSO 365=head1 SEE ALSO
254 366
255L<Coro>, L<Linux::AIO>. 367L<Coro>, L<Linux::AIO>.
256 368

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines