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 1.27 by root, Tue Aug 16 22:22:18 2005 UTC vs.
Revision 1.50 by root, Sat Jun 24 16:27:02 2006 UTC

14 aio_unlink "/tmp/file", sub { }; 14 aio_unlink "/tmp/file", sub { };
15 15
16 aio_read $fh, 30000, 1024, $buffer, 0, sub { 16 aio_read $fh, 30000, 1024, $buffer, 0, sub {
17 $_[0] > 0 or die "read error: $!"; 17 $_[0] > 0 or die "read error: $!";
18 }; 18 };
19
20 # AnyEvent
21 open my $fh, "<&=" . IO::AIO::poll_fileno or die "$!";
22 my $w = AnyEvent->io (fh => $fh, poll => 'r', cb => sub { IO::AIO::poll_cb });
19 23
20 # Event 24 # Event
21 Event->io (fd => IO::AIO::poll_fileno, 25 Event->io (fd => IO::AIO::poll_fileno,
22 poll => 'r', 26 poll => 'r',
23 cb => \&IO::AIO::poll_cb); 27 cb => \&IO::AIO::poll_cb);
63use base 'Exporter'; 67use base 'Exporter';
64 68
65use Fcntl (); 69use Fcntl ();
66 70
67BEGIN { 71BEGIN {
68 $VERSION = 1.1; 72 $VERSION = '1.8';
69 73
70 @EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink 74 @EXPORT = qw(aio_sendfile aio_read aio_write aio_open aio_close aio_stat
71 aio_rmdir aio_symlink aio_fsync aio_fdatasync aio_readahead); 75 aio_lstat aio_unlink aio_rmdir aio_readdir aio_scandir aio_symlink
76 aio_fsync aio_fdatasync aio_readahead aio_rename aio_link aio_move);
72 @EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel max_outstanding nreqs); 77 @EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel
78 max_outstanding nreqs);
73 79
74 require XSLoader; 80 require XSLoader;
75 XSLoader::load IO::AIO, $VERSION; 81 XSLoader::load IO::AIO, $VERSION;
76} 82}
77 83
88syscall has been executed asynchronously. 94syscall has been executed asynchronously.
89 95
90All functions expecting a filehandle keep a copy of the filehandle 96All functions expecting a filehandle keep a copy of the filehandle
91internally until the request has finished. 97internally until the request has finished.
92 98
93The filenames you pass to these routines I<must> be absolute. The reason 99The pathnames you pass to these routines I<must> be absolute and
94for this is that at the time the request is being executed, the current 100encoded in byte form. The reason for the former is that at the time the
95working directory could have changed. Alternatively, you can make sure 101request is being executed, the current working directory could have
96that you never change the current working directory. 102changed. Alternatively, you can make sure that you never change the
103current working directory.
104
105To encode pathnames to byte form, either make sure you either: a)
106always pass in filenames you got from outside (command line, readdir
107etc.), b) are ASCII or ISO 8859-1, c) use the Encode module and encode
108your pathnames to the locale (or other) encoding in effect in the user
109environment, d) use Glib::filename_from_unicode on unicode filenames or e)
110use something else.
97 111
98=over 4 112=over 4
99 113
100=item aio_open $pathname, $flags, $mode, $callback 114=item aio_open $pathname, $flags, $mode, $callback->($fh)
101 115
102Asynchronously open or create a file and call the callback with a newly 116Asynchronously open or create a file and call the callback with a newly
103created filehandle for the file. 117created filehandle for the file.
104 118
105The pathname passed to C<aio_open> must be absolute. See API NOTES, above, 119The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
122 } else { 136 } else {
123 die "open failed: $!\n"; 137 die "open failed: $!\n";
124 } 138 }
125 }; 139 };
126 140
127=item aio_close $fh, $callback 141=item aio_close $fh, $callback->($status)
128 142
129Asynchronously close a file and call the callback with the result 143Asynchronously close a file and call the callback with the result
130code. I<WARNING:> although accepted, you should not pass in a perl 144code. I<WARNING:> although accepted, you should not pass in a perl
131filehandle here, as perl will likely close the file descriptor another 145filehandle here, as perl will likely close the file descriptor another
132time when the filehandle is destroyed. Normally, you can safely call perls 146time when the filehandle is destroyed. Normally, you can safely call perls
133C<close> or just let filehandles go out of scope. 147C<close> or just let filehandles go out of scope.
134 148
135This is supposed to be a bug in the API, so that might change. It's 149This is supposed to be a bug in the API, so that might change. It's
136therefore best to avoid this function. 150therefore best to avoid this function.
137 151
138=item aio_read $fh,$offset,$length, $data,$dataoffset,$callback 152=item aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
139 153
140=item aio_write $fh,$offset,$length, $data,$dataoffset,$callback 154=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
141 155
142Reads or writes C<length> bytes from the specified C<fh> and C<offset> 156Reads or writes C<length> bytes from the specified C<fh> and C<offset>
143into the scalar given by C<data> and offset C<dataoffset> and calls the 157into the scalar given by C<data> and offset C<dataoffset> and calls the
144callback without the actual number of bytes read (or -1 on error, just 158callback without the actual number of bytes read (or -1 on error, just
145like the syscall). 159like the syscall).
146 160
161The C<$data> scalar I<MUST NOT> be modified in any way while the request
162is outstanding. Modifying it can result in segfaults or WW3 (if the
163necessary/optional hardware is installed).
164
147Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at 165Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
148offset C<0> within the scalar: 166offset C<0> within the scalar:
149 167
150 aio_read $fh, 7, 15, $buffer, 0, sub { 168 aio_read $fh, 7, 15, $buffer, 0, sub {
151 $_[0] > 0 or die "read error: $!"; 169 $_[0] > 0 or die "read error: $!";
152 print "read $_[0] bytes: <$buffer>\n"; 170 print "read $_[0] bytes: <$buffer>\n";
153 }; 171 };
154 172
173=item aio_move $srcpath, $dstpath, $callback->($status)
174
175[EXPERIMENTAL]
176
177Try to move the I<file> (directories not supported as either source or destination)
178from C<$srcpath> to C<$dstpath> and call the callback with the C<0> (error) or C<-1> ok.
179
180This is a composite request that tries to rename(2) the file first. If
181rename files with C<EXDEV>, it creates the destination file with mode 0200
182and copies the contents of the source file into it using C<aio_sendfile>,
183followed by restoring atime, mtime, access mode and uid/gid, in that
184order, and unlinking the C<$srcpath>.
185
186If an error occurs, the partial destination file will be unlinked, if
187possible, except when setting atime, mtime, access mode and uid/gid, where
188errors are being ignored.
189
190=cut
191
192sub aio_move($$$) {
193 my ($src, $dst, $cb) = @_;
194
195 aio_rename $src, $dst, sub {
196 if ($_[0] && $! == Errno::EXDEV) {
197 aio_open $src, O_RDONLY, 0, sub {
198 if (my $src_fh = $_[0]) {
199 my @stat = stat $src_fh;
200
201 aio_open $dst, O_WRONLY, 0200, sub {
202 if (my $dst_fh = $_[0]) {
203 aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
204 close $src_fh;
205
206 if ($_[0] == $stat[7]) {
207 utime $stat[8], $stat[9], $dst;
208 chmod $stat[2] & 07777, $dst_fh;
209 chown $stat[4], $stat[5], $dst_fh;
210 close $dst_fh;
211
212 aio_unlink $src, sub {
213 $cb->($_[0]);
214 };
215 } else {
216 my $errno = $!;
217 aio_unlink $dst, sub {
218 $! = $errno;
219 $cb->(-1);
220 };
221 }
222 };
223 } else {
224 $cb->(-1);
225 }
226 },
227
228 } else {
229 $cb->(-1);
230 }
231 };
232 } else {
233 $cb->($_[0]);
234 }
235 };
236}
237
238=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
239
240Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
241reading at byte offset C<$in_offset>, and starts writing at the current
242file offset of C<$out_fh>. Because of that, it is not safe to issue more
243than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
244other.
245
246This call tries to make use of a native C<sendfile> syscall to provide
247zero-copy operation. For this to work, C<$out_fh> should refer to a
248socket, and C<$in_fh> should refer to mmap'able file.
249
250If the native sendfile call fails or is not implemented, it will be
251emulated, so you can call C<aio_sendfile> on any type of filehandle
252regardless of the limitations of the operating system.
253
254Please note, however, that C<aio_sendfile> can read more bytes from
255C<$in_fh> than are written, and there is no way to find out how many
256bytes have been read from C<aio_sendfile> alone, as C<aio_sendfile> only
257provides the number of bytes written to C<$out_fh>. Only if the result
258value equals C<$length> one can assume that C<$length> bytes have been
259read.
260
155=item aio_readahead $fh,$offset,$length, $callback 261=item aio_readahead $fh,$offset,$length, $callback->($retval)
156 262
157C<aio_readahead> populates the page cache with data from a file so that 263C<aio_readahead> populates the page cache with data from a file so that
158subsequent reads from that file will not block on disk I/O. The C<$offset> 264subsequent reads from that file will not block on disk I/O. The C<$offset>
159argument specifies the starting point from which data is to be read and 265argument specifies the starting point from which data is to be read and
160C<$length> specifies the number of bytes to be read. I/O is performed in 266C<$length> specifies the number of bytes to be read. I/O is performed in
164file. The current file offset of the file is left unchanged. 270file. The current file offset of the file is left unchanged.
165 271
166If that syscall doesn't exist (likely if your OS isn't Linux) it will be 272If that syscall doesn't exist (likely if your OS isn't Linux) it will be
167emulated by simply reading the data, which would have a similar effect. 273emulated by simply reading the data, which would have a similar effect.
168 274
169=item aio_stat $fh_or_path, $callback 275=item aio_stat $fh_or_path, $callback->($status)
170 276
171=item aio_lstat $fh, $callback 277=item aio_lstat $fh, $callback->($status)
172 278
173Works like perl's C<stat> or C<lstat> in void context. The callback will 279Works like perl's C<stat> or C<lstat> in void context. The callback will
174be called after the stat and the results will be available using C<stat _> 280be called after the stat and the results will be available using C<stat _>
175or C<-s _> etc... 281or C<-s _> etc...
176 282
186 aio_stat "/etc/passwd", sub { 292 aio_stat "/etc/passwd", sub {
187 $_[0] and die "stat failed: $!"; 293 $_[0] and die "stat failed: $!";
188 print "size is ", -s _, "\n"; 294 print "size is ", -s _, "\n";
189 }; 295 };
190 296
191=item aio_unlink $pathname, $callback 297=item aio_unlink $pathname, $callback->($status)
192 298
193Asynchronously unlink (delete) a file and call the callback with the 299Asynchronously unlink (delete) a file and call the callback with the
194result code. 300result code.
195 301
302=item aio_link $srcpath, $dstpath, $callback->($status)
303
304Asynchronously create a new link to the existing object at C<$srcpath> at
305the path C<$dstpath> and call the callback with the result code.
306
307=item aio_symlink $srcpath, $dstpath, $callback->($status)
308
309Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
310the path C<$dstpath> and call the callback with the result code.
311
312=item aio_rename $srcpath, $dstpath, $callback->($status)
313
314Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
315rename(2) and call the callback with the result code.
316
196=item aio_rmdir $pathname, $callback 317=item aio_rmdir $pathname, $callback->($status)
197 318
198Asynchronously rmdir (delete) a directory and call the callback with the 319Asynchronously rmdir (delete) a directory and call the callback with the
199result code. 320result code.
200 321
322=item aio_readdir $pathname, $callback->($entries)
323
324Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
325directory (i.e. opendir + readdir + closedir). The entries will not be
326sorted, and will B<NOT> include the C<.> and C<..> entries.
327
328The callback a single argument which is either C<undef> or an array-ref
329with the filenames.
330
331=item aio_scandir $path, $maxreq, $callback->($dirs, $nondirs)
332
333Scans a directory (similar to C<aio_readdir>) and tries to separate the
334entries of directory C<$path> into two sets of names, ones you can recurse
335into (directories), and ones you cannot recurse into (everything else).
336
337C<aio_scandir> is a composite request that consists of many
338aio-primitives. C<$maxreq> specifies the maximum number of outstanding
339aio requests that this function generates. If it is C<< <= 0 >>, then a
340suitable default will be chosen (currently 8).
341
342On error, the callback is called without arguments, otherwise it receives
343two array-refs with path-relative entry names.
344
345Example:
346
347 aio_scandir $dir, 0, sub {
348 my ($dirs, $nondirs) = @_;
349 print "real directories: @$dirs\n";
350 print "everything else: @$nondirs\n";
351 };
352
353Implementation notes.
354
355The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.
356
357After reading the directory, the modification time, size etc. of the
358directory before and after the readdir is checked, and if they match, the
359link count will be used to decide how many entries are directories (if
360>= 2). Otherwise, no knowledge of the number of subdirectories will be
361assumed.
362
363Then entires will be sorted into likely directories (everything without a
364non-initial dot) and likely non-directories (everything else). Then every
365entry + C</.> will be C<stat>'ed, likely directories first. This is often
366faster because filesystems might detect the type of the entry without
367reading the inode data (e.g. ext2fs filetype feature). If that succeeds,
368it assumes that the entry is a directory or a symlink to directory (which
369will be checked seperately).
370
371If the known number of directories has been reached, the rest of the
372entries is assumed to be non-directories.
373
374=cut
375
376sub aio_scandir($$$) {
377 my ($path, $maxreq, $cb) = @_;
378
379 $maxreq = 8 if $maxreq <= 0;
380
381 # stat once
382 aio_stat $path, sub {
383 return $cb->() if $_[0];
384 my $hash1 = join ":", (stat _)[0,1,3,7,9];
385
386 # read the directory entries
387 aio_readdir $path, sub {
388 my $entries = shift
389 or return $cb->();
390
391 # stat the dir another time
392 aio_stat $path, sub {
393 my $hash2 = join ":", (stat _)[0,1,3,7,9];
394
395 my $ndirs;
396
397 # take the slow route if anything looks fishy
398 if ($hash1 ne $hash2) {
399 $ndirs = -1;
400 } else {
401 # if nlink == 2, we are finished
402 # on non-posix-fs's, we rely on nlink < 2
403 $ndirs = (stat _)[3] - 2
404 or return $cb->([], $entries);
405 }
406
407 # sort into likely dirs and likely nondirs
408 # dirs == files without ".", short entries first
409 $entries = [map $_->[0],
410 sort { $b->[1] cmp $a->[1] }
411 map [$_, sprintf "%s%04d", (/.\./ ? "1" : "0"), length],
412 @$entries];
413
414 my (@dirs, @nondirs);
415
416 my ($statcb, $schedcb);
417 my $nreq = 0;
418
419 $schedcb = sub {
420 if (@$entries) {
421 if ($nreq < $maxreq) {
422 my $ent = pop @$entries;
423 $nreq++;
424 aio_stat "$path/$ent/.", sub { $statcb->($_[0], $ent) };
425 }
426 } elsif (!$nreq) {
427 # finished
428 undef $statcb;
429 undef $schedcb;
430 $cb->(\@dirs, \@nondirs) if $cb;
431 undef $cb;
432 }
433 };
434 $statcb = sub {
435 my ($status, $entry) = @_;
436
437 if ($status < 0) {
438 $nreq--;
439 push @nondirs, $entry;
440 &$schedcb;
441 } else {
442 # need to check for real directory
443 aio_lstat "$path/$entry", sub {
444 $nreq--;
445
446 if (-d _) {
447 push @dirs, $entry;
448
449 if (!--$ndirs) {
450 push @nondirs, @$entries;
451 $entries = [];
452 }
453 } else {
454 push @nondirs, $entry;
455 }
456
457 &$schedcb;
458 }
459 }
460 };
461
462 &$schedcb while @$entries && $nreq < $maxreq;
463 };
464 };
465 };
466}
467
201=item aio_fsync $fh, $callback 468=item aio_fsync $fh, $callback->($status)
202 469
203Asynchronously call fsync on the given filehandle and call the callback 470Asynchronously call fsync on the given filehandle and call the callback
204with the fsync result code. 471with the fsync result code.
205 472
206=item aio_fdatasync $fh, $callback 473=item aio_fdatasync $fh, $callback->($status)
207 474
208Asynchronously call fdatasync on the given filehandle and call the 475Asynchronously call fdatasync on the given filehandle and call the
209callback with the fdatasync result code. 476callback with the fdatasync result code.
210 477
211If this call isn't available because your OS lacks it or it couldn't be 478If this call isn't available because your OS lacks it or it couldn't be
275 IO::AIO::poll_wait, IO::AIO::poll_cb 542 IO::AIO::poll_wait, IO::AIO::poll_cb
276 if IO::AIO::nreqs; 543 if IO::AIO::nreqs;
277 544
278=item IO::AIO::min_parallel $nthreads 545=item IO::AIO::min_parallel $nthreads
279 546
280Set the minimum number of AIO threads to C<$nthreads>. The default is 547Set the minimum number of AIO threads to C<$nthreads>. The current default
281C<1>, which means a single asynchronous operation can be done at one time 548is C<4>, which means four asynchronous operations can be done at one time
282(the number of outstanding operations, however, is unlimited). 549(the number of outstanding operations, however, is unlimited).
550
551IO::AIO starts threads only on demand, when an AIO request is queued and
552no free thread exists.
283 553
284It is recommended to keep the number of threads low, as some Linux 554It is recommended to keep the number of threads low, as some Linux
285kernel versions will scale negatively with the number of threads (higher 555kernel versions will scale negatively with the number of threads (higher
286parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32 556parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32
287threads should be fine. 557threads should be fine.
288 558
289Under normal circumstances you don't need to call this function, as this 559Under most circumstances you don't need to call this function, as the
290module automatically starts some threads (the exact number might change, 560module selects a default that is suitable for low to moderate load.
291and is currently 4).
292 561
293=item IO::AIO::max_parallel $nthreads 562=item IO::AIO::max_parallel $nthreads
294 563
295Sets the maximum number of AIO threads to C<$nthreads>. If more than 564Sets the maximum number of AIO threads to C<$nthreads>. If more than the
296the specified number of threads are currently running, kill them. This 565specified number of threads are currently running, this function kills
297function blocks until the limit is reached. 566them. This function blocks until the limit is reached.
567
568While C<$nthreads> are zero, aio requests get queued but not executed
569until the number of threads has been increased again.
298 570
299This module automatically runs C<max_parallel 0> at program end, to ensure 571This module automatically runs C<max_parallel 0> at program end, to ensure
300that all threads are killed and that there are no outstanding requests. 572that all threads are killed and that there are no outstanding requests.
301 573
302Under normal circumstances you don't need to call this function. 574Under normal circumstances you don't need to call this function.
306Sets the maximum number of outstanding requests to C<$nreqs>. If you 578Sets the maximum number of outstanding requests to C<$nreqs>. If you
307try to queue up more than this number of requests, the caller will block until 579try to queue up more than this number of requests, the caller will block until
308some requests have been handled. 580some requests have been handled.
309 581
310The default is very large, so normally there is no practical limit. If you 582The default is very large, so normally there is no practical limit. If you
311queue up many requests in a loop it it often improves speed if you set 583queue up many requests in a loop it often improves speed if you set
312this to a relatively low number, such as C<100>. 584this to a relatively low number, such as C<100>.
313 585
314Under normal circumstances you don't need to call this function. 586Under normal circumstances you don't need to call this function.
315 587
316=back 588=back
341 613
3421; 6141;
343 615
344=head2 FORK BEHAVIOUR 616=head2 FORK BEHAVIOUR
345 617
346IO::AIO handles all outstanding AIO requests before the fork, destroys all 618Before the fork, IO::AIO enters a quiescent state where no requests
347AIO threads, and recreates them in both the parent and the child after the 619can be added in other threads and no results will be processed. After
348fork. 620the fork the parent simply leaves the quiescent state and continues
349 621request/result processing, while the child clears the request/result
622queue (so the requests started before the fork will only be handled in
623the parent). Threats will be started on demand until the limit ste in the
624parent process has been reached again.
350 625
351=head1 SEE ALSO 626=head1 SEE ALSO
352 627
353L<Coro>, L<Linux::AIO>. 628L<Coro>, L<Linux::AIO>.
354 629

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines