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

Comparing IO-AIO/README (file contents):
Revision 1.8 by root, Sun Jul 31 17:24:05 2005 UTC vs.
Revision 1.61 by root, Sun Aug 12 06:07:06 2018 UTC

1NAME 1NAME
2 IO::AIO - Asynchronous Input/Output 2 IO::AIO - Asynchronous/Advanced Input/Output
3 3
4SYNOPSIS 4SYNOPSIS
5 use IO::AIO; 5 use IO::AIO;
6 6
7 aio_open "/etc/passwd", O_RDONLY, 0, sub { 7 aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
8 my ($fh) = @_; 8 my $fh = shift
9 or die "/etc/passwd: $!";
9 ... 10 ...
10 }; 11 };
11 12
12 aio_unlink "/tmp/file", sub { }; 13 aio_unlink "/tmp/file", sub { };
13 14
14 aio_read $fh, 30000, 1024, $buffer, 0, sub { 15 aio_read $fh, 30000, 1024, $buffer, 0, sub {
15 $_[0] > 0 or die "read error: $!"; 16 $_[0] > 0 or die "read error: $!";
16 }; 17 };
17 18
18 # Event 19 # version 2+ has request and group objects
19 Event->io (fd => IO::AIO::poll_fileno, 20 use IO::AIO 2;
20 poll => 'r',
21 cb => \&IO::AIO::poll_cb);
22 21
23 # Glib/Gtk2 22 aioreq_pri 4; # give next request a very high priority
24 add_watch Glib::IO IO::AIO::poll_fileno, 23 my $req = aio_unlink "/tmp/file", sub { };
25 in => sub { IO::AIO::poll_cb; 1 }; 24 $req->cancel; # cancel request if still in queue
26 25
27 # Tk 26 my $grp = aio_group sub { print "all stats done\n" };
28 Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "", 27 add $grp aio_stat "..." for ...;
29 readable => \&IO::AIO::poll_cb);
30
31 # Danga::Socket
32 Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
33 \&IO::AIO::poll_cb);
34 28
35DESCRIPTION 29DESCRIPTION
36 This module implements asynchronous I/O using whatever means your 30 This module implements asynchronous I/O using whatever means your
37 operating system supports. 31 operating system supports. It is implemented as an interface to "libeio"
32 (<http://software.schmorp.de/pkg/libeio.html>).
38 33
34 Asynchronous means that operations that can normally block your program
35 (e.g. reading from disk) will be done asynchronously: the operation will
36 still block, but you can do something else in the meantime. This is
37 extremely useful for programs that need to stay interactive even when
38 doing heavy I/O (GUI programs, high performance network servers etc.),
39 but can also be used to easily do operations in parallel that are
40 normally done sequentially, e.g. stat'ing many files, which is much
41 faster on a RAID volume or over NFS when you do a number of stat
42 operations concurrently.
43
44 While most of this works on all types of file descriptors (for example
45 sockets), using these functions on file descriptors that support
46 nonblocking operation (again, sockets, pipes etc.) is very inefficient.
47 Use an event loop for that (such as the EV module): IO::AIO will
48 naturally fit into such an event loop itself.
49
39 Currently, a number of threads are started that execute your read/writes 50 In this version, a number of threads are started that execute your
40 and signal their completion. You don't need thread support in your libc 51 requests and signal their completion. You don't need thread support in
41 or perl, and the threads created by this module will not be visible to 52 perl, and the threads created by this module will not be visible to
42 the pthreads library. In the future, this module might make use of the 53 perl. In the future, this module might make use of the native aio
43 native aio functions available on many operating systems. However, they 54 functions available on many operating systems. However, they are often
44 are often not well-supported (Linux doesn't allow them on normal files 55 not well-supported or restricted (GNU/Linux doesn't allow them on normal
45 currently, for example), and they would only support aio_read and 56 files currently, for example), and they would only support aio_read and
46 aio_write, so the remaining functionality would have to be implemented 57 aio_write, so the remaining functionality would have to be implemented
47 using threads anyway. 58 using threads anyway.
48 59
60 In addition to asynchronous I/O, this module also exports some rather
61 arcane interfaces, such as "madvise" or linux's "splice" system call,
62 which is why the "A" in "AIO" can also mean *advanced*.
63
49 Although the module will work with in the presence of other threads, it 64 Although the module will work in the presence of other (Perl-) threads,
50 is currently not reentrant, so use appropriate locking yourself, always 65 it is currently not reentrant in any way, so use appropriate locking
51 call "poll_cb" from within the same thread, or never call "poll_cb" (or 66 yourself, always call "poll_cb" from within the same thread, or never
52 other "aio_" functions) recursively. 67 call "poll_cb" (or other "aio_" functions) recursively.
68
69 EXAMPLE
70 This is a simple example that uses the EV module and loads /etc/passwd
71 asynchronously:
72
73 use EV;
74 use IO::AIO;
75
76 # register the IO::AIO callback with EV
77 my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
78
79 # queue the request to open /etc/passwd
80 aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
81 my $fh = shift
82 or die "error while opening: $!";
83
84 # stat'ing filehandles is generally non-blocking
85 my $size = -s $fh;
86
87 # queue a request to read the file
88 my $contents;
89 aio_read $fh, 0, $size, $contents, 0, sub {
90 $_[0] == $size
91 or die "short read: $!";
92
93 close $fh;
94
95 # file contents now in $contents
96 print $contents;
97
98 # exit event loop and program
99 EV::break;
100 };
101 };
102
103 # possibly queue up other requests, or open GUI windows,
104 # check for sockets etc. etc.
105
106 # process events as long as there are some:
107 EV::run;
108
109REQUEST ANATOMY AND LIFETIME
110 Every "aio_*" function creates a request. which is a C data structure
111 not directly visible to Perl.
112
113 If called in non-void context, every request function returns a Perl
114 object representing the request. In void context, nothing is returned,
115 which saves a bit of memory.
116
117 The perl object is a fairly standard ref-to-hash object. The hash
118 contents are not used by IO::AIO so you are free to store anything you
119 like in it.
120
121 During their existance, aio requests travel through the following
122 states, in order:
123
124 ready
125 Immediately after a request is created it is put into the ready
126 state, waiting for a thread to execute it.
127
128 execute
129 A thread has accepted the request for processing and is currently
130 executing it (e.g. blocking in read).
131
132 pending
133 The request has been executed and is waiting for result processing.
134
135 While request submission and execution is fully asynchronous, result
136 processing is not and relies on the perl interpreter calling
137 "poll_cb" (or another function with the same effect).
138
139 result
140 The request results are processed synchronously by "poll_cb".
141
142 The "poll_cb" function will process all outstanding aio requests by
143 calling their callbacks, freeing memory associated with them and
144 managing any groups they are contained in.
145
146 done
147 Request has reached the end of its lifetime and holds no resources
148 anymore (except possibly for the Perl object, but its connection to
149 the actual aio request is severed and calling its methods will
150 either do nothing or result in a runtime error).
53 151
54FUNCTIONS 152FUNCTIONS
55 AIO FUNCTIONS 153 QUICK OVERVIEW
154 This section simply lists the prototypes most of the functions for quick
155 reference. See the following sections for function-by-function
156 documentation.
157
158 aio_wd $pathname, $callback->($wd)
159 aio_open $pathname, $flags, $mode, $callback->($fh)
160 aio_close $fh, $callback->($status)
161 aio_seek $fh,$offset,$whence, $callback->($offs)
162 aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
163 aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
164 aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
165 aio_readahead $fh,$offset,$length, $callback->($retval)
166 aio_stat $fh_or_path, $callback->($status)
167 aio_lstat $fh, $callback->($status)
168 aio_statvfs $fh_or_path, $callback->($statvfs)
169 aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
170 aio_chown $fh_or_path, $uid, $gid, $callback->($status)
171 aio_chmod $fh_or_path, $mode, $callback->($status)
172 aio_truncate $fh_or_path, $offset, $callback->($status)
173 aio_allocate $fh, $mode, $offset, $len, $callback->($status)
174 aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
175 aio_unlink $pathname, $callback->($status)
176 aio_mknod $pathname, $mode, $dev, $callback->($status)
177 aio_link $srcpath, $dstpath, $callback->($status)
178 aio_symlink $srcpath, $dstpath, $callback->($status)
179 aio_readlink $pathname, $callback->($link)
180 aio_realpath $pathname, $callback->($path)
181 aio_rename $srcpath, $dstpath, $callback->($status)
182 aio_rename2 $srcpath, $dstpath, $flags, $callback->($status)
183 aio_mkdir $pathname, $mode, $callback->($status)
184 aio_rmdir $pathname, $callback->($status)
185 aio_readdir $pathname, $callback->($entries)
186 aio_readdirx $pathname, $flags, $callback->($entries, $flags)
187 IO::AIO::READDIR_DENTS IO::AIO::READDIR_DIRS_FIRST
188 IO::AIO::READDIR_STAT_ORDER IO::AIO::READDIR_FOUND_UNKNOWN
189 aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
190 aio_load $pathname, $data, $callback->($status)
191 aio_copy $srcpath, $dstpath, $callback->($status)
192 aio_move $srcpath, $dstpath, $callback->($status)
193 aio_rmtree $pathname, $callback->($status)
194 aio_fcntl $fh, $cmd, $arg, $callback->($status)
195 aio_ioctl $fh, $request, $buf, $callback->($status)
196 aio_sync $callback->($status)
197 aio_syncfs $fh, $callback->($status)
198 aio_fsync $fh, $callback->($status)
199 aio_fdatasync $fh, $callback->($status)
200 aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
201 aio_pathsync $pathname, $callback->($status)
202 aio_msync $scalar, $offset = 0, $length = undef, flags = MS_SYNC, $callback->($status)
203 aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
204 aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
205 aio_mlockall $flags, $callback->($status)
206 aio_group $callback->(...)
207 aio_nop $callback->()
208
209 $prev_pri = aioreq_pri [$pri]
210 aioreq_nice $pri_adjust
211
212 IO::AIO::poll_wait
213 IO::AIO::poll_cb
214 IO::AIO::poll
215 IO::AIO::flush
216 IO::AIO::max_poll_reqs $nreqs
217 IO::AIO::max_poll_time $seconds
218 IO::AIO::min_parallel $nthreads
219 IO::AIO::max_parallel $nthreads
220 IO::AIO::max_idle $nthreads
221 IO::AIO::idle_timeout $seconds
222 IO::AIO::max_outstanding $maxreqs
223 IO::AIO::nreqs
224 IO::AIO::nready
225 IO::AIO::npending
226 $nfd = IO::AIO::get_fdlimit [EXPERIMENTAL]
227 IO::AIO::min_fdlimit $nfd [EXPERIMENTAL]
228
229 IO::AIO::sendfile $ofh, $ifh, $offset, $count
230 IO::AIO::fadvise $fh, $offset, $len, $advice
231 IO::AIO::mmap $scalar, $length, $prot, $flags[, $fh[, $offset]]
232 IO::AIO::munmap $scalar
233 IO::AIO::mremap $scalar, $new_length, $flags[, $new_address]
234 IO::AIO::madvise $scalar, $offset, $length, $advice
235 IO::AIO::mprotect $scalar, $offset, $length, $protect
236 IO::AIO::munlock $scalar, $offset = 0, $length = undef
237 IO::AIO::munlockall
238
239 API NOTES
56 All the "aio_*" calls are more or less thin wrappers around the syscall 240 All the "aio_*" calls are more or less thin wrappers around the syscall
57 with the same name (sans "aio_"). The arguments are similar or 241 with the same name (sans "aio_"). The arguments are similar or
58 identical, and they all accept an additional (and optional) $callback 242 identical, and they all accept an additional (and optional) $callback
59 argument which must be a code reference. This code reference will get 243 argument which must be a code reference. This code reference will be
60 called with the syscall return code (e.g. most syscalls return -1 on 244 called after the syscall has been executed in an asynchronous fashion.
61 error, unlike perl, which usually delivers "false") as it's sole 245 The results of the request will be passed as arguments to the callback
62 argument when the given syscall has been executed asynchronously. 246 (and, if an error occured, in $!) - for most requests the syscall return
247 code (e.g. most syscalls return -1 on error, unlike perl, which usually
248 delivers "false").
249
250 Some requests (such as "aio_readdir") pass the actual results and
251 communicate failures by passing "undef".
63 252
64 All functions expecting a filehandle keep a copy of the filehandle 253 All functions expecting a filehandle keep a copy of the filehandle
65 internally until the request has finished. 254 internally until the request has finished.
66 255
256 All functions return request objects of type IO::AIO::REQ that allow
257 further manipulation of those requests while they are in-flight.
258
67 The filenames you pass to these routines *must* be absolute. The reason 259 The pathnames you pass to these routines *should* be absolute. The
68 for this is that at the time the request is being executed, the current 260 reason for this is that at the time the request is being executed, the
69 working directory could have changed. Alternatively, you can make sure 261 current working directory could have changed. Alternatively, you can
70 that you never change the current working directory. 262 make sure that you never change the current working directory anywhere
263 in the program and then use relative paths. You can also take advantage
264 of IO::AIOs working directory abstraction, that lets you specify paths
265 relative to some previously-opened "working directory object" - see the
266 description of the "IO::AIO::WD" class later in this document.
71 267
268 To encode pathnames as octets, either make sure you either: a) always
269 pass in filenames you got from outside (command line, readdir etc.)
270 without tinkering, b) are in your native filesystem encoding, c) use the
271 Encode module and encode your pathnames to the locale (or other)
272 encoding in effect in the user environment, d) use
273 Glib::filename_from_unicode on unicode filenames or e) use something
274 else to ensure your scalar has the correct contents.
275
276 This works, btw. independent of the internal UTF-8 bit, which IO::AIO
277 handles correctly whether it is set or not.
278
279 AIO REQUEST FUNCTIONS
280 $prev_pri = aioreq_pri [$pri]
281 Returns the priority value that would be used for the next request
282 and, if $pri is given, sets the priority for the next aio request.
283
284 The default priority is 0, the minimum and maximum priorities are -4
285 and 4, respectively. Requests with higher priority will be serviced
286 first.
287
288 The priority will be reset to 0 after each call to one of the
289 "aio_*" functions.
290
291 Example: open a file with low priority, then read something from it
292 with higher priority so the read request is serviced before other
293 low priority open requests (potentially spamming the cache):
294
295 aioreq_pri -3;
296 aio_open ..., sub {
297 return unless $_[0];
298
299 aioreq_pri -2;
300 aio_read $_[0], ..., sub {
301 ...
302 };
303 };
304
305 aioreq_nice $pri_adjust
306 Similar to "aioreq_pri", but subtracts the given value from the
307 current priority, so the effect is cumulative.
308
72 aio_open $pathname, $flags, $mode, $callback 309 aio_open $pathname, $flags, $mode, $callback->($fh)
73 Asynchronously open or create a file and call the callback with a 310 Asynchronously open or create a file and call the callback with a
74 newly created filehandle for the file. 311 newly created filehandle for the file (or "undef" in case of an
312 error).
75 313
76 The pathname passed to "aio_open" must be absolute. See API NOTES, 314 The pathname passed to "aio_open" must be absolute. See API NOTES,
77 above, for an explanation. 315 above, for an explanation.
78 316
79 The $flags argument is a bitmask. See the "Fcntl" module for a list. 317 The $flags argument is a bitmask. See the "Fcntl" module for a list.
80 They are the same as used by "sysopen". 318 They are the same as used by "sysopen".
81 319
82 Likewise, $mode specifies the mode of the newly created file, if it 320 Likewise, $mode specifies the mode of the newly created file, if it
83 didn't exist and "O_CREAT" has been given, just like perl's 321 didn't exist and "O_CREAT" has been given, just like perl's
84 "sysopen", except that it is mandatory (i.e. use 0 if you don't 322 "sysopen", except that it is mandatory (i.e. use 0 if you don't
85 create new files, and 0666 or 0777 if you do). 323 create new files, and 0666 or 0777 if you do). Note that the $mode
324 will be modified by the umask in effect then the request is being
325 executed, so better never change the umask.
86 326
87 Example: 327 Example:
88 328
89 aio_open "/etc/passwd", O_RDONLY, 0, sub { 329 aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub {
90 if ($_[0]) { 330 if ($_[0]) {
91 print "open successful, fh is $_[0]\n"; 331 print "open successful, fh is $_[0]\n";
92 ... 332 ...
93 } else { 333 } else {
94 die "open failed: $!\n"; 334 die "open failed: $!\n";
95 } 335 }
96 }; 336 };
97 337
338 In addition to all the common open modes/flags ("O_RDONLY",
339 "O_WRONLY", "O_RDWR", "O_CREAT", "O_TRUNC", "O_EXCL" and
340 "O_APPEND"), the following POSIX and non-POSIX constants are
341 available (missing ones on your system are, as usual, 0):
342
343 "O_ASYNC", "O_DIRECT", "O_NOATIME", "O_CLOEXEC", "O_NOCTTY",
344 "O_NOFOLLOW", "O_NONBLOCK", "O_EXEC", "O_SEARCH", "O_DIRECTORY",
345 "O_DSYNC", "O_RSYNC", "O_SYNC", "O_PATH", "O_TMPFILE", "O_TTY_INIT"
346 and "O_ACCMODE".
347
98 aio_close $fh, $callback 348 aio_close $fh, $callback->($status)
99 Asynchronously close a file and call the callback with the result 349 Asynchronously close a file and call the callback with the result
100 code. *WARNING:* although accepted, you should not pass in a perl 350 code.
101 filehandle here, as perl will likely close the file descriptor
102 another time when the filehandle is destroyed. Normally, you can
103 safely call perls "close" or just let filehandles go out of scope.
104 351
105 This is supposed to be a bug in the API, so that might change. It's 352 Unfortunately, you can't do this to perl. Perl *insists* very
106 therefore best to avoid this function. 353 strongly on closing the file descriptor associated with the
354 filehandle itself.
107 355
356 Therefore, "aio_close" will not close the filehandle - instead it
357 will use dup2 to overwrite the file descriptor with the write-end of
358 a pipe (the pipe fd will be created on demand and will be cached).
359
360 Or in other words: the file descriptor will be closed, but it will
361 not be free for reuse until the perl filehandle is closed.
362
363 aio_seek $fh, $offset, $whence, $callback->($offs)
364 Seeks the filehandle to the new $offset, similarly to perl's
365 "sysseek". The $whence can use the traditional values (0 for
366 "IO::AIO::SEEK_SET", 1 for "IO::AIO::SEEK_CUR" or 2 for
367 "IO::AIO::SEEK_END").
368
369 The resulting absolute offset will be passed to the callback, or -1
370 in case of an error.
371
372 In theory, the $whence constants could be different than the
373 corresponding values from Fcntl, but perl guarantees they are the
374 same, so don't panic.
375
376 As a GNU/Linux (and maybe Solaris) extension, also the constants
377 "IO::AIO::SEEK_DATA" and "IO::AIO::SEEK_HOLE" are available, if they
378 could be found. No guarantees about suitability for use in
379 "aio_seek" or Perl's "sysseek" can be made though, although I would
380 naively assume they "just work".
381
108 aio_read $fh,$offset,$length, $data,$dataoffset,$callback 382 aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
109 aio_write $fh,$offset,$length, $data,$dataoffset,$callback 383 aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
110 Reads or writes "length" bytes from the specified "fh" and "offset" 384 Reads or writes $length bytes from or to the specified $fh and
111 into the scalar given by "data" and offset "dataoffset" and calls 385 $offset into the scalar given by $data and offset $dataoffset and
112 the callback without the actual number of bytes read (or -1 on 386 calls the callback with the actual number of bytes transferred (or
113 error, just like the syscall). 387 -1 on error, just like the syscall).
388
389 "aio_read" will, like "sysread", shrink or grow the $data scalar to
390 offset plus the actual number of bytes read.
391
392 If $offset is undefined, then the current file descriptor offset
393 will be used (and updated), otherwise the file descriptor offset
394 will not be changed by these calls.
395
396 If $length is undefined in "aio_write", use the remaining length of
397 $data.
398
399 If $dataoffset is less than zero, it will be counted from the end of
400 $data.
401
402 The $data scalar *MUST NOT* be modified in any way while the request
403 is outstanding. Modifying it can result in segfaults or World War
404 III (if the necessary/optional hardware is installed).
114 405
115 Example: Read 15 bytes at offset 7 into scalar $buffer, starting at 406 Example: Read 15 bytes at offset 7 into scalar $buffer, starting at
116 offset 0 within the scalar: 407 offset 0 within the scalar:
117 408
118 aio_read $fh, 7, 15, $buffer, 0, sub { 409 aio_read $fh, 7, 15, $buffer, 0, sub {
119 $_[0] > 0 or die "read error: $!"; 410 $_[0] > 0 or die "read error: $!";
120 print "read $_[0] bytes: <$buffer>\n"; 411 print "read $_[0] bytes: <$buffer>\n";
121 }; 412 };
122 413
414 aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
415 Tries to copy $length bytes from $in_fh to $out_fh. It starts
416 reading at byte offset $in_offset, and starts writing at the current
417 file offset of $out_fh. Because of that, it is not safe to issue
418 more than one "aio_sendfile" per $out_fh, as they will interfere
419 with each other. The same $in_fh works fine though, as this function
420 does not move or use the file offset of $in_fh.
421
422 Please note that "aio_sendfile" can read more bytes from $in_fh than
423 are written, and there is no way to find out how many more bytes
424 have been read from "aio_sendfile" alone, as "aio_sendfile" only
425 provides the number of bytes written to $out_fh. Only if the result
426 value equals $length one can assume that $length bytes have been
427 read.
428
429 Unlike with other "aio_" functions, it makes a lot of sense to use
430 "aio_sendfile" on non-blocking sockets, as long as one end
431 (typically the $in_fh) is a file - the file I/O will then be
432 asynchronous, while the socket I/O will be non-blocking. Note,
433 however, that you can run into a trap where "aio_sendfile" reads
434 some data with readahead, then fails to write all data, and when the
435 socket is ready the next time, the data in the cache is already
436 lost, forcing "aio_sendfile" to again hit the disk. Explicit
437 "aio_read" + "aio_write" let's you better control resource usage.
438
439 This call tries to make use of a native "sendfile"-like syscall to
440 provide zero-copy operation. For this to work, $out_fh should refer
441 to a socket, and $in_fh should refer to an mmap'able file.
442
443 If a native sendfile cannot be found or it fails with "ENOSYS",
444 "EINVAL", "ENOTSUP", "EOPNOTSUPP", "EAFNOSUPPORT", "EPROTOTYPE" or
445 "ENOTSOCK", it will be emulated, so you can call "aio_sendfile" on
446 any type of filehandle regardless of the limitations of the
447 operating system.
448
449 As native sendfile syscalls (as practically any non-POSIX interface
450 hacked together in a hurry to improve benchmark numbers) tend to be
451 rather buggy on many systems, this implementation tries to work
452 around some known bugs in Linux and FreeBSD kernels (probably
453 others, too), but that might fail, so you really really should check
454 the return value of "aio_sendfile" - fewer bytes than expected might
455 have been transferred.
456
123 aio_readahead $fh,$offset,$length, $callback 457 aio_readahead $fh,$offset,$length, $callback->($retval)
124 Asynchronously reads the specified byte range into the page cache,
125 using the "readahead" syscall. If that syscall doesn't exist (likely
126 if your OS isn't Linux) the status will be -1 and $! is set to
127 "ENOSYS".
128
129 "aio_readahead" populates the page cache with data from a file so 458 "aio_readahead" populates the page cache with data from a file so
130 that subsequent reads from that file will not block on disk I/O. The 459 that subsequent reads from that file will not block on disk I/O. The
131 $offset argument specifies the starting point from which data is to 460 $offset argument specifies the starting point from which data is to
132 be read and $length specifies the number of bytes to be read. I/O is 461 be read and $length specifies the number of bytes to be read. I/O is
133 performed in whole pages, so that offset is effectively rounded down 462 performed in whole pages, so that offset is effectively rounded down
134 to a page boundary and bytes are read up to the next page boundary 463 to a page boundary and bytes are read up to the next page boundary
135 greater than or equal to (off-set+length). "aio_readahead" does not 464 greater than or equal to (off-set+length). "aio_readahead" does not
136 read beyond the end of the file. The current file offset of the file 465 read beyond the end of the file. The current file offset of the file
137 is left unchanged. 466 is left unchanged.
138 467
468 If that syscall doesn't exist (likely if your kernel isn't Linux) it
469 will be emulated by simply reading the data, which would have a
470 similar effect.
471
139 aio_stat $fh_or_path, $callback 472 aio_stat $fh_or_path, $callback->($status)
140 aio_lstat $fh, $callback 473 aio_lstat $fh, $callback->($status)
141 Works like perl's "stat" or "lstat" in void context. The callback 474 Works like perl's "stat" or "lstat" in void context. The callback
142 will be called after the stat and the results will be available 475 will be called after the stat and the results will be available
143 using "stat _" or "-s _" etc... 476 using "stat _" or "-s _" etc...
144 477
145 The pathname passed to "aio_stat" must be absolute. See API NOTES, 478 The pathname passed to "aio_stat" must be absolute. See API NOTES,
… …
148 Currently, the stats are always 64-bit-stats, i.e. instead of 481 Currently, the stats are always 64-bit-stats, i.e. instead of
149 returning an error when stat'ing a large file, the results will be 482 returning an error when stat'ing a large file, the results will be
150 silently truncated unless perl itself is compiled with large file 483 silently truncated unless perl itself is compiled with large file
151 support. 484 support.
152 485
486 To help interpret the mode and dev/rdev stat values, IO::AIO offers
487 the following constants and functions (if not implemented, the
488 constants will be 0 and the functions will either "croak" or fall
489 back on traditional behaviour).
490
491 "S_IFMT", "S_IFIFO", "S_IFCHR", "S_IFBLK", "S_IFLNK", "S_IFREG",
492 "S_IFDIR", "S_IFWHT", "S_IFSOCK", "IO::AIO::major $dev_t",
493 "IO::AIO::minor $dev_t", "IO::AIO::makedev $major, $minor".
494
495 To access higher resolution stat timestamps, see "SUBSECOND STAT
496 TIME ACCESS".
497
153 Example: Print the length of /etc/passwd: 498 Example: Print the length of /etc/passwd:
154 499
155 aio_stat "/etc/passwd", sub { 500 aio_stat "/etc/passwd", sub {
156 $_[0] and die "stat failed: $!"; 501 $_[0] and die "stat failed: $!";
157 print "size is ", -s _, "\n"; 502 print "size is ", -s _, "\n";
158 }; 503 };
159 504
505 aio_statvfs $fh_or_path, $callback->($statvfs)
506 Works like the POSIX "statvfs" or "fstatvfs" syscalls, depending on
507 whether a file handle or path was passed.
508
509 On success, the callback is passed a hash reference with the
510 following members: "bsize", "frsize", "blocks", "bfree", "bavail",
511 "files", "ffree", "favail", "fsid", "flag" and "namemax". On
512 failure, "undef" is passed.
513
514 The following POSIX IO::AIO::ST_* constants are defined: "ST_RDONLY"
515 and "ST_NOSUID".
516
517 The following non-POSIX IO::AIO::ST_* flag masks are defined to
518 their correct value when available, or to 0 on systems that do not
519 support them: "ST_NODEV", "ST_NOEXEC", "ST_SYNCHRONOUS",
520 "ST_MANDLOCK", "ST_WRITE", "ST_APPEND", "ST_IMMUTABLE",
521 "ST_NOATIME", "ST_NODIRATIME" and "ST_RELATIME".
522
523 Example: stat "/wd" and dump out the data if successful.
524
525 aio_statvfs "/wd", sub {
526 my $f = $_[0]
527 or die "statvfs: $!";
528
529 use Data::Dumper;
530 say Dumper $f;
531 };
532
533 # result:
534 {
535 bsize => 1024,
536 bfree => 4333064312,
537 blocks => 10253828096,
538 files => 2050765568,
539 flag => 4096,
540 favail => 2042092649,
541 bavail => 4333064312,
542 ffree => 2042092649,
543 namemax => 255,
544 frsize => 1024,
545 fsid => 1810
546 }
547
548 aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
549 Works like perl's "utime" function (including the special case of
550 $atime and $mtime being undef). Fractional times are supported if
551 the underlying syscalls support them.
552
553 When called with a pathname, uses utimes(2) if available, otherwise
554 utime(2). If called on a file descriptor, uses futimes(2) if
555 available, otherwise returns ENOSYS, so this is not portable.
556
557 Examples:
558
559 # set atime and mtime to current time (basically touch(1)):
560 aio_utime "path", undef, undef;
561 # set atime to current time and mtime to beginning of the epoch:
562 aio_utime "path", time, undef; # undef==0
563
564 aio_chown $fh_or_path, $uid, $gid, $callback->($status)
565 Works like perl's "chown" function, except that "undef" for either
566 $uid or $gid is being interpreted as "do not change" (but -1 can
567 also be used).
568
569 Examples:
570
571 # same as "chown root path" in the shell:
572 aio_chown "path", 0, -1;
573 # same as above:
574 aio_chown "path", 0, undef;
575
576 aio_truncate $fh_or_path, $offset, $callback->($status)
577 Works like truncate(2) or ftruncate(2).
578
579 aio_allocate $fh, $mode, $offset, $len, $callback->($status)
580 Allocates or frees disk space according to the $mode argument. See
581 the linux "fallocate" documentation for details.
582
583 $mode is usually 0 or "IO::AIO::FALLOC_FL_KEEP_SIZE" to allocate
584 space, or "IO::AIO::FALLOC_FL_PUNCH_HOLE |
585 IO::AIO::FALLOC_FL_KEEP_SIZE", to deallocate a file range.
586
587 IO::AIO also supports "FALLOC_FL_COLLAPSE_RANGE", to remove a range
588 (without leaving a hole), "FALLOC_FL_ZERO_RANGE", to zero a range,
589 "FALLOC_FL_INSERT_RANGE" to insert a range and
590 "FALLOC_FL_UNSHARE_RANGE" to unshare shared blocks (see your
591 fallocate(2) manpage).
592
593 The file system block size used by "fallocate" is presumably the
594 "f_bsize" returned by "statvfs", but different filesystems and
595 filetypes can dictate other limitations.
596
597 If "fallocate" isn't available or cannot be emulated (currently no
598 emulation will be attempted), passes -1 and sets $! to "ENOSYS".
599
600 aio_chmod $fh_or_path, $mode, $callback->($status)
601 Works like perl's "chmod" function.
602
160 aio_unlink $pathname, $callback 603 aio_unlink $pathname, $callback->($status)
161 Asynchronously unlink (delete) a file and call the callback with the 604 Asynchronously unlink (delete) a file and call the callback with the
162 result code. 605 result code.
163 606
607 aio_mknod $pathname, $mode, $dev, $callback->($status)
608 [EXPERIMENTAL]
609
610 Asynchronously create a device node (or fifo). See mknod(2).
611
612 The only (POSIX-) portable way of calling this function is:
613
614 aio_mknod $pathname, IO::AIO::S_IFIFO | $mode, 0, sub { ...
615
616 See "aio_stat" for info about some potentially helpful extra
617 constants and functions.
618
619 aio_link $srcpath, $dstpath, $callback->($status)
620 Asynchronously create a new link to the existing object at $srcpath
621 at the path $dstpath and call the callback with the result code.
622
623 aio_symlink $srcpath, $dstpath, $callback->($status)
624 Asynchronously create a new symbolic link to the existing object at
625 $srcpath at the path $dstpath and call the callback with the result
626 code.
627
628 aio_readlink $pathname, $callback->($link)
629 Asynchronously read the symlink specified by $path and pass it to
630 the callback. If an error occurs, nothing or undef gets passed to
631 the callback.
632
633 aio_realpath $pathname, $callback->($path)
634 Asynchronously make the path absolute and resolve any symlinks in
635 $path. The resulting path only consists of directories (same as
636 Cwd::realpath).
637
638 This request can be used to get the absolute path of the current
639 working directory by passing it a path of . (a single dot).
640
641 aio_rename $srcpath, $dstpath, $callback->($status)
642 Asynchronously rename the object at $srcpath to $dstpath, just as
643 rename(2) and call the callback with the result code.
644
645 On systems that support the AIO::WD working directory abstraction
646 natively, the case "[$wd, "."]" as $srcpath is specialcased -
647 instead of failing, "rename" is called on the absolute path of $wd.
648
649 aio_rename2 $srcpath, $dstpath, $flags, $callback->($status)
650 Basically a version of "aio_rename" with an additional $flags
651 argument. Calling this with "$flags=0" is the same as calling
652 "aio_rename".
653
654 Non-zero flags are currently only supported on GNU/Linux systems
655 that support renameat2. Other systems fail with "ENOSYS" in this
656 case.
657
658 The following constants are available (missing ones are, as usual
659 0), see renameat2(2) for details:
660
661 "IO::AIO::RENAME_NOREPLACE", "IO::AIO::RENAME_EXCHANGE" and
662 "IO::AIO::RENAME_WHITEOUT".
663
664 aio_mkdir $pathname, $mode, $callback->($status)
665 Asynchronously mkdir (create) a directory and call the callback with
666 the result code. $mode will be modified by the umask at the time the
667 request is executed, so do not change your umask.
668
669 aio_rmdir $pathname, $callback->($status)
670 Asynchronously rmdir (delete) a directory and call the callback with
671 the result code.
672
673 On systems that support the AIO::WD working directory abstraction
674 natively, the case "[$wd, "."]" is specialcased - instead of
675 failing, "rmdir" is called on the absolute path of $wd.
676
677 aio_readdir $pathname, $callback->($entries)
678 Unlike the POSIX call of the same name, "aio_readdir" reads an
679 entire directory (i.e. opendir + readdir + closedir). The entries
680 will not be sorted, and will NOT include the "." and ".." entries.
681
682 The callback is passed a single argument which is either "undef" or
683 an array-ref with the filenames.
684
685 aio_readdirx $pathname, $flags, $callback->($entries, $flags)
686 Quite similar to "aio_readdir", but the $flags argument allows one
687 to tune behaviour and output format. In case of an error, $entries
688 will be "undef".
689
690 The flags are a combination of the following constants, ORed
691 together (the flags will also be passed to the callback, possibly
692 modified):
693
694 IO::AIO::READDIR_DENTS
695 Normally the callback gets an arrayref consisting of names only
696 (as with "aio_readdir"). If this flag is set, then the callback
697 gets an arrayref with "[$name, $type, $inode]" arrayrefs, each
698 describing a single directory entry in more detail:
699
700 $name is the name of the entry.
701
702 $type is one of the "IO::AIO::DT_xxx" constants:
703
704 "IO::AIO::DT_UNKNOWN", "IO::AIO::DT_FIFO", "IO::AIO::DT_CHR",
705 "IO::AIO::DT_DIR", "IO::AIO::DT_BLK", "IO::AIO::DT_REG",
706 "IO::AIO::DT_LNK", "IO::AIO::DT_SOCK", "IO::AIO::DT_WHT".
707
708 "IO::AIO::DT_UNKNOWN" means just that: readdir does not know. If
709 you need to know, you have to run stat yourself. Also, for
710 speed/memory reasons, the $type scalars are read-only: you must
711 not modify them.
712
713 $inode is the inode number (which might not be exact on systems
714 with 64 bit inode numbers and 32 bit perls). This field has
715 unspecified content on systems that do not deliver the inode
716 information.
717
718 IO::AIO::READDIR_DIRS_FIRST
719 When this flag is set, then the names will be returned in an
720 order where likely directories come first, in optimal stat
721 order. This is useful when you need to quickly find directories,
722 or you want to find all directories while avoiding to stat()
723 each entry.
724
725 If the system returns type information in readdir, then this is
726 used to find directories directly. Otherwise, likely directories
727 are names beginning with ".", or otherwise names with no dots,
728 of which names with short names are tried first.
729
730 IO::AIO::READDIR_STAT_ORDER
731 When this flag is set, then the names will be returned in an
732 order suitable for stat()'ing each one. That is, when you plan
733 to stat() most or all files in the given directory, then the
734 returned order will likely be faster.
735
736 If both this flag and "IO::AIO::READDIR_DIRS_FIRST" are
737 specified, then the likely dirs come first, resulting in a less
738 optimal stat order for stat'ing all entries, but likely a more
739 optimal order for finding subdirectories.
740
741 IO::AIO::READDIR_FOUND_UNKNOWN
742 This flag should not be set when calling "aio_readdirx".
743 Instead, it is being set by "aio_readdirx", when any of the
744 $type's found were "IO::AIO::DT_UNKNOWN". The absence of this
745 flag therefore indicates that all $type's are known, which can
746 be used to speed up some algorithms.
747
748 aio_slurp $pathname, $offset, $length, $data, $callback->($status)
749 Opens, reads and closes the given file. The data is put into $data,
750 which is resized as required.
751
752 If $offset is negative, then it is counted from the end of the file.
753
754 If $length is zero, then the remaining length of the file is used.
755 Also, in this case, the same limitations to modifying $data apply as
756 when IO::AIO::mmap is used, i.e. it must only be modified in-place
757 with "substr". If the size of the file is known, specifying a
758 non-zero $length results in a performance advantage.
759
760 This request is similar to the older "aio_load" request, but since
761 it is a single request, it might be more efficient to use.
762
763 Example: load /etc/passwd into $passwd.
764
765 my $passwd;
766 aio_slurp "/etc/passwd", 0, 0, $passwd, sub {
767 $_[0] >= 0
768 or die "/etc/passwd: $!\n";
769
770 printf "/etc/passwd is %d bytes long, and contains:\n", length $passwd;
771 print $passwd;
772 };
773 IO::AIO::flush;
774
775 aio_load $pathname, $data, $callback->($status)
776 This is a composite request that tries to fully load the given file
777 into memory. Status is the same as with aio_read.
778
779 Using "aio_slurp" might be more efficient, as it is a single
780 request.
781
782 aio_copy $srcpath, $dstpath, $callback->($status)
783 Try to copy the *file* (directories not supported as either source
784 or destination) from $srcpath to $dstpath and call the callback with
785 a status of 0 (ok) or -1 (error, see $!).
786
787 Existing destination files will be truncated.
788
789 This is a composite request that creates the destination file with
790 mode 0200 and copies the contents of the source file into it using
791 "aio_sendfile", followed by restoring atime, mtime, access mode and
792 uid/gid, in that order.
793
794 If an error occurs, the partial destination file will be unlinked,
795 if possible, except when setting atime, mtime, access mode and
796 uid/gid, where errors are being ignored.
797
798 aio_move $srcpath, $dstpath, $callback->($status)
799 Try to move the *file* (directories not supported as either source
800 or destination) from $srcpath to $dstpath and call the callback with
801 a status of 0 (ok) or -1 (error, see $!).
802
803 This is a composite request that tries to rename(2) the file first;
804 if rename fails with "EXDEV", it copies the file with "aio_copy"
805 and, if that is successful, unlinks the $srcpath.
806
807 aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
808 Scans a directory (similar to "aio_readdir") but additionally tries
809 to efficiently separate the entries of directory $path into two sets
810 of names, directories you can recurse into (directories), and ones
811 you cannot recurse into (everything else, including symlinks to
812 directories).
813
814 "aio_scandir" is a composite request that generates many sub
815 requests. $maxreq specifies the maximum number of outstanding aio
816 requests that this function generates. If it is "<= 0", then a
817 suitable default will be chosen (currently 4).
818
819 On error, the callback is called without arguments, otherwise it
820 receives two array-refs with path-relative entry names.
821
822 Example:
823
824 aio_scandir $dir, 0, sub {
825 my ($dirs, $nondirs) = @_;
826 print "real directories: @$dirs\n";
827 print "everything else: @$nondirs\n";
828 };
829
830 Implementation notes.
831
832 The "aio_readdir" cannot be avoided, but "stat()"'ing every entry
833 can.
834
835 If readdir returns file type information, then this is used directly
836 to find directories.
837
838 Otherwise, after reading the directory, the modification time, size
839 etc. of the directory before and after the readdir is checked, and
840 if they match (and isn't the current time), the link count will be
841 used to decide how many entries are directories (if >= 2).
842 Otherwise, no knowledge of the number of subdirectories will be
843 assumed.
844
845 Then entries will be sorted into likely directories a non-initial
846 dot currently) and likely non-directories (see "aio_readdirx"). Then
847 every entry plus an appended "/." will be "stat"'ed, likely
848 directories first, in order of their inode numbers. If that
849 succeeds, it assumes that the entry is a directory or a symlink to
850 directory (which will be checked separately). This is often faster
851 than stat'ing the entry itself because filesystems might detect the
852 type of the entry without reading the inode data (e.g. ext2fs
853 filetype feature), even on systems that cannot return the filetype
854 information on readdir.
855
856 If the known number of directories (link count - 2) has been
857 reached, the rest of the entries is assumed to be non-directories.
858
859 This only works with certainty on POSIX (= UNIX) filesystems, which
860 fortunately are the vast majority of filesystems around.
861
862 It will also likely work on non-POSIX filesystems with reduced
863 efficiency as those tend to return 0 or 1 as link counts, which
864 disables the directory counting heuristic.
865
866 aio_rmtree $pathname, $callback->($status)
867 Delete a directory tree starting (and including) $path, return the
868 status of the final "rmdir" only. This is a composite request that
869 uses "aio_scandir" to recurse into and rmdir directories, and unlink
870 everything else.
871
872 aio_fcntl $fh, $cmd, $arg, $callback->($status)
873 aio_ioctl $fh, $request, $buf, $callback->($status)
874 These work just like the "fcntl" and "ioctl" built-in functions,
875 except they execute asynchronously and pass the return value to the
876 callback.
877
878 Both calls can be used for a lot of things, some of which make more
879 sense to run asynchronously in their own thread, while some others
880 make less sense. For example, calls that block waiting for external
881 events, such as locking, will also lock down an I/O thread while it
882 is waiting, which can deadlock the whole I/O system. At the same
883 time, there might be no alternative to using a thread to wait.
884
885 So in general, you should only use these calls for things that do
886 (filesystem) I/O, not for things that wait for other events
887 (network, other processes), although if you are careful and know
888 what you are doing, you still can.
889
890 The following constants are available (missing ones are, as usual
891 0):
892
893 "F_DUPFD_CLOEXEC",
894
895 "F_OFD_GETLK", "F_OFD_SETLK", "F_OFD_GETLKW",
896
897 "FIFREEZE", "FITHAW", "FITRIM", "FICLONE", "FICLONERANGE",
898 "FIDEDUPERANGE".
899
900 "FS_IOC_GETFLAGS", "FS_IOC_SETFLAGS", "FS_IOC_GETVERSION",
901 "FS_IOC_SETVERSION", "FS_IOC_FIEMAP".
902
903 "FS_IOC_FSGETXATTR", "FS_IOC_FSSETXATTR",
904 "FS_IOC_SET_ENCRYPTION_POLICY", "FS_IOC_GET_ENCRYPTION_PWSALT",
905 "FS_IOC_GET_ENCRYPTION_POLICY", "FS_KEY_DESCRIPTOR_SIZE".
906
907 "FS_SECRM_FL", "FS_UNRM_FL", "FS_COMPR_FL", "FS_SYNC_FL",
908 "FS_IMMUTABLE_FL", "FS_APPEND_FL", "FS_NODUMP_FL", "FS_NOATIME_FL",
909 "FS_DIRTY_FL", "FS_COMPRBLK_FL", "FS_NOCOMP_FL", "FS_ENCRYPT_FL",
910 "FS_BTREE_FL", "FS_INDEX_FL", "FS_JOURNAL_DATA_FL", "FS_NOTAIL_FL",
911 "FS_DIRSYNC_FL", "FS_TOPDIR_FL", "FS_FL_USER_MODIFIABLE".
912
913 "FS_XFLAG_REALTIME", "FS_XFLAG_PREALLOC", "FS_XFLAG_IMMUTABLE",
914 "FS_XFLAG_APPEND", "FS_XFLAG_SYNC", "FS_XFLAG_NOATIME",
915 "FS_XFLAG_NODUMP", "FS_XFLAG_RTINHERIT", "FS_XFLAG_PROJINHERIT",
916 "FS_XFLAG_NOSYMLINKS", "FS_XFLAG_EXTSIZE", "FS_XFLAG_EXTSZINHERIT",
917 "FS_XFLAG_NODEFRAG", "FS_XFLAG_FILESTREAM", "FS_XFLAG_DAX",
918 "FS_XFLAG_HASATTR",
919
920 aio_sync $callback->($status)
921 Asynchronously call sync and call the callback when finished.
922
164 aio_fsync $fh, $callback 923 aio_fsync $fh, $callback->($status)
165 Asynchronously call fsync on the given filehandle and call the 924 Asynchronously call fsync on the given filehandle and call the
166 callback with the fsync result code. 925 callback with the fsync result code.
167 926
168 aio_fdatasync $fh, $callback 927 aio_fdatasync $fh, $callback->($status)
169 Asynchronously call fdatasync on the given filehandle and call the 928 Asynchronously call fdatasync on the given filehandle and call the
170 callback with the fdatasync result code. Might set $! to "ENOSYS" if 929 callback with the fdatasync result code.
171 "fdatasync" is not available. 930
931 If this call isn't available because your OS lacks it or it couldn't
932 be detected, it will be emulated by calling "fsync" instead.
933
934 aio_syncfs $fh, $callback->($status)
935 Asynchronously call the syncfs syscall to sync the filesystem
936 associated to the given filehandle and call the callback with the
937 syncfs result code. If syncfs is not available, calls sync(), but
938 returns -1 and sets errno to "ENOSYS" nevertheless.
939
940 aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
941 Sync the data portion of the file specified by $offset and $length
942 to disk (but NOT the metadata), by calling the Linux-specific
943 sync_file_range call. If sync_file_range is not available or it
944 returns ENOSYS, then fdatasync or fsync is being substituted.
945
946 $flags can be a combination of
947 "IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE",
948 "IO::AIO::SYNC_FILE_RANGE_WRITE" and
949 "IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER": refer to the sync_file_range
950 manpage for details.
951
952 aio_pathsync $pathname, $callback->($status)
953 This request tries to open, fsync and close the given path. This is
954 a composite request intended to sync directories after directory
955 operations (E.g. rename). This might not work on all operating
956 systems or have any specific effect, but usually it makes sure that
957 directory changes get written to disc. It works for anything that
958 can be opened for read-only, not just directories.
959
960 Future versions of this function might fall back to other methods
961 when "fsync" on the directory fails (such as calling "sync").
962
963 Passes 0 when everything went ok, and -1 on error.
964
965 aio_msync $scalar, $offset = 0, $length = undef, flags = MS_SYNC,
966 $callback->($status)
967 This is a rather advanced IO::AIO call, which only works on
968 mmap(2)ed scalars (see the "IO::AIO::mmap" function, although it
969 also works on data scalars managed by the Sys::Mmap or Mmap modules,
970 note that the scalar must only be modified in-place while an aio
971 operation is pending on it).
972
973 It calls the "msync" function of your OS, if available, with the
974 memory area starting at $offset in the string and ending $length
975 bytes later. If $length is negative, counts from the end, and if
976 $length is "undef", then it goes till the end of the string. The
977 flags can be either "IO::AIO::MS_ASYNC" or "IO::AIO::MS_SYNC", plus
978 an optional "IO::AIO::MS_INVALIDATE".
979
980 aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0,
981 $callback->($status)
982 This is a rather advanced IO::AIO call, which works best on
983 mmap(2)ed scalars.
984
985 It touches (reads or writes) all memory pages in the specified range
986 inside the scalar. All caveats and parameters are the same as for
987 "aio_msync", above, except for flags, which must be either 0 (which
988 reads all pages and ensures they are instantiated) or
989 "IO::AIO::MT_MODIFY", which modifies the memory pages (by reading
990 and writing an octet from it, which dirties the page).
991
992 aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
993 This is a rather advanced IO::AIO call, which works best on
994 mmap(2)ed scalars.
995
996 It reads in all the pages of the underlying storage into memory (if
997 any) and locks them, so they are not getting swapped/paged out or
998 removed.
999
1000 If $length is undefined, then the scalar will be locked till the
1001 end.
1002
1003 On systems that do not implement "mlock", this function returns -1
1004 and sets errno to "ENOSYS".
1005
1006 Note that the corresponding "munlock" is synchronous and is
1007 documented under "MISCELLANEOUS FUNCTIONS".
1008
1009 Example: open a file, mmap and mlock it - both will be undone when
1010 $data gets destroyed.
1011
1012 open my $fh, "<", $path or die "$path: $!";
1013 my $data;
1014 IO::AIO::mmap $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh;
1015 aio_mlock $data; # mlock in background
1016
1017 aio_mlockall $flags, $callback->($status)
1018 Calls the "mlockall" function with the given $flags (a combination
1019 of "IO::AIO::MCL_CURRENT" and "IO::AIO::MCL_FUTURE").
1020
1021 On systems that do not implement "mlockall", this function returns
1022 -1 and sets errno to "ENOSYS".
1023
1024 Note that the corresponding "munlockall" is synchronous and is
1025 documented under "MISCELLANEOUS FUNCTIONS".
1026
1027 Example: asynchronously lock all current and future pages into
1028 memory.
1029
1030 aio_mlockall IO::AIO::MCL_FUTURE;
1031
1032 aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents)
1033 Queries the extents of the given file (by calling the Linux "FIEMAP"
1034 ioctl, see <http://cvs.schmorp.de/IO-AIO/doc/fiemap.txt> for
1035 details). If the ioctl is not available on your OS, then this
1036 request will fail with "ENOSYS".
1037
1038 $start is the starting offset to query extents for, $length is the
1039 size of the range to query - if it is "undef", then the whole file
1040 will be queried.
1041
1042 $flags is a combination of flags ("IO::AIO::FIEMAP_FLAG_SYNC" or
1043 "IO::AIO::FIEMAP_FLAG_XATTR" - "IO::AIO::FIEMAP_FLAGS_COMPAT" is
1044 also exported), and is normally 0 or "IO::AIO::FIEMAP_FLAG_SYNC" to
1045 query the data portion.
1046
1047 $count is the maximum number of extent records to return. If it is
1048 "undef", then IO::AIO queries all extents of the range. As a very
1049 special case, if it is 0, then the callback receives the number of
1050 extents instead of the extents themselves (which is unreliable, see
1051 below).
1052
1053 If an error occurs, the callback receives no arguments. The special
1054 "errno" value "IO::AIO::EBADR" is available to test for flag errors.
1055
1056 Otherwise, the callback receives an array reference with extent
1057 structures. Each extent structure is an array reference itself, with
1058 the following members:
1059
1060 [$logical, $physical, $length, $flags]
1061
1062 Flags is any combination of the following flag values (typically
1063 either 0 or "IO::AIO::FIEMAP_EXTENT_LAST" (1)):
1064
1065 "IO::AIO::FIEMAP_EXTENT_LAST", "IO::AIO::FIEMAP_EXTENT_UNKNOWN",
1066 "IO::AIO::FIEMAP_EXTENT_DELALLOC", "IO::AIO::FIEMAP_EXTENT_ENCODED",
1067 "IO::AIO::FIEMAP_EXTENT_DATA_ENCRYPTED",
1068 "IO::AIO::FIEMAP_EXTENT_NOT_ALIGNED",
1069 "IO::AIO::FIEMAP_EXTENT_DATA_INLINE",
1070 "IO::AIO::FIEMAP_EXTENT_DATA_TAIL",
1071 "IO::AIO::FIEMAP_EXTENT_UNWRITTEN", "IO::AIO::FIEMAP_EXTENT_MERGED"
1072 or "IO::AIO::FIEMAP_EXTENT_SHARED".
1073
1074 At the time of this writing (Linux 3.2), this request is unreliable
1075 unless $count is "undef", as the kernel has all sorts of bugs
1076 preventing it to return all extents of a range for files with a
1077 large number of extents. The code (only) works around all these
1078 issues if $count is "undef".
1079
1080 aio_group $callback->(...)
1081 This is a very special aio request: Instead of doing something, it
1082 is a container for other aio requests, which is useful if you want
1083 to bundle many requests into a single, composite, request with a
1084 definite callback and the ability to cancel the whole request with
1085 its subrequests.
1086
1087 Returns an object of class IO::AIO::GRP. See its documentation below
1088 for more info.
1089
1090 Example:
1091
1092 my $grp = aio_group sub {
1093 print "all stats done\n";
1094 };
1095
1096 add $grp
1097 (aio_stat ...),
1098 (aio_stat ...),
1099 ...;
1100
1101 aio_nop $callback->()
1102 This is a special request - it does nothing in itself and is only
1103 used for side effects, such as when you want to add a dummy request
1104 to a group so that finishing the requests in the group depends on
1105 executing the given code.
1106
1107 While this request does nothing, it still goes through the execution
1108 phase and still requires a worker thread. Thus, the callback will
1109 not be executed immediately but only after other requests in the
1110 queue have entered their execution phase. This can be used to
1111 measure request latency.
1112
1113 IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED*
1114 Mainly used for debugging and benchmarking, this aio request puts
1115 one of the request workers to sleep for the given time.
1116
1117 While it is theoretically handy to have simple I/O scheduling
1118 requests like sleep and file handle readable/writable, the overhead
1119 this creates is immense (it blocks a thread for a long time) so do
1120 not use this function except to put your application under
1121 artificial I/O pressure.
1122
1123 IO::AIO::WD - multiple working directories
1124 Your process only has one current working directory, which is used by
1125 all threads. This makes it hard to use relative paths (some other
1126 component could call "chdir" at any time, and it is hard to control when
1127 the path will be used by IO::AIO).
1128
1129 One solution for this is to always use absolute paths. This usually
1130 works, but can be quite slow (the kernel has to walk the whole path on
1131 every access), and can also be a hassle to implement.
1132
1133 Newer POSIX systems have a number of functions (openat, fdopendir,
1134 futimensat and so on) that make it possible to specify working
1135 directories per operation.
1136
1137 For portability, and because the clowns who "designed", or shall I
1138 write, perpetrated this new interface were obviously half-drunk, this
1139 abstraction cannot be perfect, though.
1140
1141 IO::AIO allows you to convert directory paths into a so-called
1142 IO::AIO::WD object. This object stores the canonicalised, absolute
1143 version of the path, and on systems that allow it, also a directory file
1144 descriptor.
1145
1146 Everywhere where a pathname is accepted by IO::AIO (e.g. in "aio_stat"
1147 or "aio_unlink"), one can specify an array reference with an IO::AIO::WD
1148 object and a pathname instead (or the IO::AIO::WD object alone, which
1149 gets interpreted as "[$wd, "."]"). If the pathname is absolute, the
1150 IO::AIO::WD object is ignored, otherwise the pathname is resolved
1151 relative to that IO::AIO::WD object.
1152
1153 For example, to get a wd object for /etc and then stat passwd inside,
1154 you would write:
1155
1156 aio_wd "/etc", sub {
1157 my $etcdir = shift;
1158
1159 # although $etcdir can be undef on error, there is generally no reason
1160 # to check for errors here, as aio_stat will fail with ENOENT
1161 # when $etcdir is undef.
1162
1163 aio_stat [$etcdir, "passwd"], sub {
1164 # yay
1165 };
1166 };
1167
1168 The fact that "aio_wd" is a request and not a normal function shows that
1169 creating an IO::AIO::WD object is itself a potentially blocking
1170 operation, which is why it is done asynchronously.
1171
1172 To stat the directory obtained with "aio_wd" above, one could write
1173 either of the following three request calls:
1174
1175 aio_lstat "/etc" , sub { ... # pathname as normal string
1176 aio_lstat [$wd, "."], sub { ... # "." relative to $wd (i.e. $wd itself)
1177 aio_lstat $wd , sub { ... # shorthand for the previous
1178
1179 As with normal pathnames, IO::AIO keeps a copy of the working directory
1180 object and the pathname string, so you could write the following without
1181 causing any issues due to $path getting reused:
1182
1183 my $path = [$wd, undef];
1184
1185 for my $name (qw(abc def ghi)) {
1186 $path->[1] = $name;
1187 aio_stat $path, sub {
1188 # ...
1189 };
1190 }
1191
1192 There are some caveats: when directories get renamed (or deleted), the
1193 pathname string doesn't change, so will point to the new directory (or
1194 nowhere at all), while the directory fd, if available on the system,
1195 will still point to the original directory. Most functions accepting a
1196 pathname will use the directory fd on newer systems, and the string on
1197 older systems. Some functions (such as "aio_realpath") will always rely
1198 on the string form of the pathname.
1199
1200 So this functionality is mainly useful to get some protection against
1201 "chdir", to easily get an absolute path out of a relative path for
1202 future reference, and to speed up doing many operations in the same
1203 directory (e.g. when stat'ing all files in a directory).
1204
1205 The following functions implement this working directory abstraction:
1206
1207 aio_wd $pathname, $callback->($wd)
1208 Asynchonously canonicalise the given pathname and convert it to an
1209 IO::AIO::WD object representing it. If possible and supported on the
1210 system, also open a directory fd to speed up pathname resolution
1211 relative to this working directory.
1212
1213 If something goes wrong, then "undef" is passwd to the callback
1214 instead of a working directory object and $! is set appropriately.
1215 Since passing "undef" as working directory component of a pathname
1216 fails the request with "ENOENT", there is often no need for error
1217 checking in the "aio_wd" callback, as future requests using the
1218 value will fail in the expected way.
1219
1220 IO::AIO::CWD
1221 This is a compiletime constant (object) that represents the process
1222 current working directory.
1223
1224 Specifying this object as working directory object for a pathname is
1225 as if the pathname would be specified directly, without a directory
1226 object. For example, these calls are functionally identical:
1227
1228 aio_stat "somefile", sub { ... };
1229 aio_stat [IO::AIO::CWD, "somefile"], sub { ... };
1230
1231 To recover the path associated with an IO::AIO::WD object, you can use
1232 "aio_realpath":
1233
1234 aio_realpath $wd, sub {
1235 warn "path is $_[0]\n";
1236 };
1237
1238 Currently, "aio_statvfs" always, and "aio_rename" and "aio_rmdir"
1239 sometimes, fall back to using an absolue path.
1240
1241 IO::AIO::REQ CLASS
1242 All non-aggregate "aio_*" functions return an object of this class when
1243 called in non-void context.
1244
1245 cancel $req
1246 Cancels the request, if possible. Has the effect of skipping
1247 execution when entering the execute state and skipping calling the
1248 callback when entering the the result state, but will leave the
1249 request otherwise untouched (with the exception of readdir). That
1250 means that requests that currently execute will not be stopped and
1251 resources held by the request will not be freed prematurely.
1252
1253 cb $req $callback->(...)
1254 Replace (or simply set) the callback registered to the request.
1255
1256 IO::AIO::GRP CLASS
1257 This class is a subclass of IO::AIO::REQ, so all its methods apply to
1258 objects of this class, too.
1259
1260 A IO::AIO::GRP object is a special request that can contain multiple
1261 other aio requests.
1262
1263 You create one by calling the "aio_group" constructing function with a
1264 callback that will be called when all contained requests have entered
1265 the "done" state:
1266
1267 my $grp = aio_group sub {
1268 print "all requests are done\n";
1269 };
1270
1271 You add requests by calling the "add" method with one or more
1272 "IO::AIO::REQ" objects:
1273
1274 $grp->add (aio_unlink "...");
1275
1276 add $grp aio_stat "...", sub {
1277 $_[0] or return $grp->result ("error");
1278
1279 # add another request dynamically, if first succeeded
1280 add $grp aio_open "...", sub {
1281 $grp->result ("ok");
1282 };
1283 };
1284
1285 This makes it very easy to create composite requests (see the source of
1286 "aio_move" for an application) that work and feel like simple requests.
1287
1288 * The IO::AIO::GRP objects will be cleaned up during calls to
1289 "IO::AIO::poll_cb", just like any other request.
1290
1291 * They can be canceled like any other request. Canceling will cancel
1292 not only the request itself, but also all requests it contains.
1293
1294 * They can also can also be added to other IO::AIO::GRP objects.
1295
1296 * You must not add requests to a group from within the group callback
1297 (or any later time).
1298
1299 Their lifetime, simplified, looks like this: when they are empty, they
1300 will finish very quickly. If they contain only requests that are in the
1301 "done" state, they will also finish. Otherwise they will continue to
1302 exist.
1303
1304 That means after creating a group you have some time to add requests
1305 (precisely before the callback has been invoked, which is only done
1306 within the "poll_cb"). And in the callbacks of those requests, you can
1307 add further requests to the group. And only when all those requests have
1308 finished will the the group itself finish.
1309
1310 add $grp ...
1311 $grp->add (...)
1312 Add one or more requests to the group. Any type of IO::AIO::REQ can
1313 be added, including other groups, as long as you do not create
1314 circular dependencies.
1315
1316 Returns all its arguments.
1317
1318 $grp->cancel_subs
1319 Cancel all subrequests and clears any feeder, but not the group
1320 request itself. Useful when you queued a lot of events but got a
1321 result early.
1322
1323 The group request will finish normally (you cannot add requests to
1324 the group).
1325
1326 $grp->result (...)
1327 Set the result value(s) that will be passed to the group callback
1328 when all subrequests have finished and set the groups errno to the
1329 current value of errno (just like calling "errno" without an error
1330 number). By default, no argument will be passed and errno is zero.
1331
1332 $grp->errno ([$errno])
1333 Sets the group errno value to $errno, or the current value of errno
1334 when the argument is missing.
1335
1336 Every aio request has an associated errno value that is restored
1337 when the callback is invoked. This method lets you change this value
1338 from its default (0).
1339
1340 Calling "result" will also set errno, so make sure you either set $!
1341 before the call to "result", or call c<errno> after it.
1342
1343 feed $grp $callback->($grp)
1344 Sets a feeder/generator on this group: every group can have an
1345 attached generator that generates requests if idle. The idea behind
1346 this is that, although you could just queue as many requests as you
1347 want in a group, this might starve other requests for a potentially
1348 long time. For example, "aio_scandir" might generate hundreds of
1349 thousands of "aio_stat" requests, delaying any later requests for a
1350 long time.
1351
1352 To avoid this, and allow incremental generation of requests, you can
1353 instead a group and set a feeder on it that generates those
1354 requests. The feed callback will be called whenever there are few
1355 enough (see "limit", below) requests active in the group itself and
1356 is expected to queue more requests.
1357
1358 The feed callback can queue as many requests as it likes (i.e. "add"
1359 does not impose any limits).
1360
1361 If the feed does not queue more requests when called, it will be
1362 automatically removed from the group.
1363
1364 If the feed limit is 0 when this method is called, it will be set to
1365 2 automatically.
1366
1367 Example:
1368
1369 # stat all files in @files, but only ever use four aio requests concurrently:
1370
1371 my $grp = aio_group sub { print "finished\n" };
1372 limit $grp 4;
1373 feed $grp sub {
1374 my $file = pop @files
1375 or return;
1376
1377 add $grp aio_stat $file, sub { ... };
1378 };
1379
1380 limit $grp $num
1381 Sets the feeder limit for the group: The feeder will be called
1382 whenever the group contains less than this many requests.
1383
1384 Setting the limit to 0 will pause the feeding process.
1385
1386 The default value for the limit is 0, but note that setting a feeder
1387 automatically bumps it up to 2.
172 1388
173 SUPPORT FUNCTIONS 1389 SUPPORT FUNCTIONS
1390 EVENT PROCESSING AND EVENT LOOP INTEGRATION
174 $fileno = IO::AIO::poll_fileno 1391 $fileno = IO::AIO::poll_fileno
175 Return the *request result pipe file descriptor*. This filehandle 1392 Return the *request result pipe file descriptor*. This filehandle
176 must be polled for reading by some mechanism outside this module 1393 must be polled for reading by some mechanism outside this module
177 (e.g. Event or select, see below or the SYNOPSIS). If the pipe 1394 (e.g. EV, Glib, select and so on, see below or the SYNOPSIS). If the
178 becomes readable you have to call "poll_cb" to check the results. 1395 pipe becomes readable you have to call "poll_cb" to check the
1396 results.
179 1397
180 See "poll_cb" for an example. 1398 See "poll_cb" for an example.
181 1399
182 IO::AIO::poll_cb 1400 IO::AIO::poll_cb
183 Process all outstanding events on the result pipe. You have to call 1401 Process some requests that have reached the result phase (i.e. they
184 this regularly. Returns the number of events processed. Returns 1402 have been executed but the results are not yet reported). You have
1403 to call this "regularly" to finish outstanding requests.
1404
1405 Returns 0 if all events could be processed (or there were no events
1406 to process), or -1 if it returned earlier for whatever reason.
185 immediately when no events are outstanding. 1407 Returns immediately when no events are outstanding. The amount of
1408 events processed depends on the settings of "IO::AIO::max_poll_req",
1409 "IO::AIO::max_poll_time" and "IO::AIO::max_outstanding".
1410
1411 If not all requests were processed for whatever reason, the poll
1412 file descriptor will still be ready when "poll_cb" returns, so
1413 normally you don't have to do anything special to have it called
1414 later.
1415
1416 Apart from calling "IO::AIO::poll_cb" when the event filehandle
1417 becomes ready, it can be beneficial to call this function from loops
1418 which submit a lot of requests, to make sure the results get
1419 processed when they become available and not just when the loop is
1420 finished and the event loop takes over again. This function returns
1421 very fast when there are no outstanding requests.
186 1422
187 Example: Install an Event watcher that automatically calls 1423 Example: Install an Event watcher that automatically calls
188 IO::AIO::poll_cb with high priority: 1424 IO::AIO::poll_cb with high priority (more examples can be found in
1425 the SYNOPSIS section, at the top of this document):
189 1426
190 Event->io (fd => IO::AIO::poll_fileno, 1427 Event->io (fd => IO::AIO::poll_fileno,
191 poll => 'r', async => 1, 1428 poll => 'r', async => 1,
192 cb => \&IO::AIO::poll_cb); 1429 cb => \&IO::AIO::poll_cb);
193 1430
194 IO::AIO::poll_wait 1431 IO::AIO::poll_wait
195 Wait till the result filehandle becomes ready for reading (simply 1432 Wait until either at least one request is in the result phase or no
196 does a "select" on the filehandle. This is useful if you want to 1433 requests are outstanding anymore.
197 synchronously wait for some requests to finish). 1434
1435 This is useful if you want to synchronously wait for some requests
1436 to become ready, without actually handling them.
198 1437
199 See "nreqs" for an example. 1438 See "nreqs" for an example.
200 1439
1440 IO::AIO::poll
1441 Waits until some requests have been handled.
1442
1443 Returns the number of requests processed, but is otherwise strictly
1444 equivalent to:
1445
1446 IO::AIO::poll_wait, IO::AIO::poll_cb
1447
201 IO::AIO::nreqs 1448 IO::AIO::flush
202 Returns the number of requests currently outstanding (i.e. for which 1449 Wait till all outstanding AIO requests have been handled.
203 their callback has not been invoked yet).
204 1450
205 Example: wait till there are no outstanding requests anymore: 1451 Strictly equivalent to:
206 1452
207 IO::AIO::poll_wait, IO::AIO::poll_cb 1453 IO::AIO::poll_wait, IO::AIO::poll_cb
208 while IO::AIO::nreqs; 1454 while IO::AIO::nreqs;
209 1455
210 IO::AIO::flush 1456 IO::AIO::max_poll_reqs $nreqs
211 Wait till all outstanding AIO requests have been handled. 1457 IO::AIO::max_poll_time $seconds
1458 These set the maximum number of requests (default 0, meaning
1459 infinity) that are being processed by "IO::AIO::poll_cb" in one
1460 call, respectively the maximum amount of time (default 0, meaning
1461 infinity) spent in "IO::AIO::poll_cb" to process requests (more
1462 correctly the mininum amount of time "poll_cb" is allowed to use).
212 1463
213 Strictly equivalent to: 1464 Setting "max_poll_time" to a non-zero value creates an overhead of
1465 one syscall per request processed, which is not normally a problem
1466 unless your callbacks are really really fast or your OS is really
1467 really slow (I am not mentioning Solaris here). Using
1468 "max_poll_reqs" incurs no overhead.
214 1469
215 IO::AIO::poll_wait, IO::AIO::poll_cb 1470 Setting these is useful if you want to ensure some level of
216 while IO::AIO::nreqs; 1471 interactiveness when perl is not fast enough to process all requests
1472 in time.
217 1473
218 IO::AIO::poll 1474 For interactive programs, values such as 0.01 to 0.1 should be fine.
219 Waits until some requests have been handled.
220 1475
221 Strictly equivalent to: 1476 Example: Install an Event watcher that automatically calls
1477 IO::AIO::poll_cb with low priority, to ensure that other parts of
1478 the program get the CPU sometimes even under high AIO load.
222 1479
223 IO::AIO::poll_wait, IO::AIO::poll_cb 1480 # try not to spend much more than 0.1s in poll_cb
224 if IO::AIO::nreqs; 1481 IO::AIO::max_poll_time 0.1;
225 1482
1483 # use a low priority so other tasks have priority
1484 Event->io (fd => IO::AIO::poll_fileno,
1485 poll => 'r', nice => 1,
1486 cb => &IO::AIO::poll_cb);
1487
1488 CONTROLLING THE NUMBER OF THREADS
226 IO::AIO::min_parallel $nthreads 1489 IO::AIO::min_parallel $nthreads
227 Set the minimum number of AIO threads to $nthreads. The default is 1490 Set the minimum number of AIO threads to $nthreads. The current
228 1, which means a single asynchronous operation can be done at one 1491 default is 8, which means eight asynchronous operations can execute
229 time (the number of outstanding operations, however, is unlimited). 1492 concurrently at any one time (the number of outstanding requests,
1493 however, is unlimited).
230 1494
1495 IO::AIO starts threads only on demand, when an AIO request is queued
1496 and no free thread exists. Please note that queueing up a hundred
1497 requests can create demand for a hundred threads, even if it turns
1498 out that everything is in the cache and could have been processed
1499 faster by a single thread.
1500
231 It is recommended to keep the number of threads low, as some Linux 1501 It is recommended to keep the number of threads relatively low, as
232 kernel versions will scale negatively with the number of threads 1502 some Linux kernel versions will scale negatively with the number of
233 (higher parallelity => MUCH higher latency). With current Linux 2.6 1503 threads (higher parallelity => MUCH higher latency). With current
234 versions, 4-32 threads should be fine. 1504 Linux 2.6 versions, 4-32 threads should be fine.
235 1505
236 Under normal circumstances you don't need to call this function, as 1506 Under most circumstances you don't need to call this function, as
237 this module automatically starts some threads (the exact number 1507 the module selects a default that is suitable for low to moderate
238 might change, and is currently 4). 1508 load.
239 1509
240 IO::AIO::max_parallel $nthreads 1510 IO::AIO::max_parallel $nthreads
241 Sets the maximum number of AIO threads to $nthreads. If more than 1511 Sets the maximum number of AIO threads to $nthreads. If more than
242 the specified number of threads are currently running, kill them. 1512 the specified number of threads are currently running, this function
243 This function blocks until the limit is reached. 1513 kills them. This function blocks until the limit is reached.
1514
1515 While $nthreads are zero, aio requests get queued but not executed
1516 until the number of threads has been increased again.
244 1517
245 This module automatically runs "max_parallel 0" at program end, to 1518 This module automatically runs "max_parallel 0" at program end, to
246 ensure that all threads are killed and that there are no outstanding 1519 ensure that all threads are killed and that there are no outstanding
247 requests. 1520 requests.
248 1521
249 Under normal circumstances you don't need to call this function. 1522 Under normal circumstances you don't need to call this function.
250 1523
1524 IO::AIO::max_idle $nthreads
1525 Limit the number of threads (default: 4) that are allowed to idle
1526 (i.e., threads that did not get a request to process within the idle
1527 timeout (default: 10 seconds). That means if a thread becomes idle
1528 while $nthreads other threads are also idle, it will free its
1529 resources and exit.
1530
1531 This is useful when you allow a large number of threads (e.g. 100 or
1532 1000) to allow for extremely high load situations, but want to free
1533 resources under normal circumstances (1000 threads can easily
1534 consume 30MB of RAM).
1535
1536 The default is probably ok in most situations, especially if thread
1537 creation is fast. If thread creation is very slow on your system you
1538 might want to use larger values.
1539
1540 IO::AIO::idle_timeout $seconds
1541 Sets the minimum idle timeout (default 10) after which worker
1542 threads are allowed to exit. SEe "IO::AIO::max_idle".
1543
251 $oldnreqs = IO::AIO::max_outstanding $nreqs 1544 IO::AIO::max_outstanding $maxreqs
252 Sets the maximum number of outstanding requests to $nreqs. If you 1545 Sets the maximum number of outstanding requests to $nreqs. If you do
253 try to queue up more than this number of requests, the caller will 1546 queue up more than this number of requests, the next call to
1547 "IO::AIO::poll_cb" (and other functions calling "poll_cb", such as
1548 "IO::AIO::flush" or "IO::AIO::poll") will block until the limit is
1549 no longer exceeded.
1550
1551 In other words, this setting does not enforce a queue limit, but can
1552 be used to make poll functions block if the limit is exceeded.
1553
1554 This is a very bad function to use in interactive programs because
1555 it blocks, and a bad way to reduce concurrency because it is
1556 inexact: Better use an "aio_group" together with a feed callback.
1557
1558 Its main use is in scripts without an event loop - when you want to
1559 stat a lot of files, you can write something like this:
1560
1561 IO::AIO::max_outstanding 32;
1562
1563 for my $path (...) {
1564 aio_stat $path , ...;
1565 IO::AIO::poll_cb;
1566 }
1567
1568 IO::AIO::flush;
1569
1570 The call to "poll_cb" inside the loop will normally return
1571 instantly, but as soon as more thna 32 reqeusts are in-flight, it
254 block until some requests have been handled. 1572 will block until some requests have been handled. This keeps the
1573 loop from pushing a large number of "aio_stat" requests onto the
1574 queue.
255 1575
256 The default is very large, so normally there is no practical limit. 1576 The default value for "max_outstanding" is very large, so there is
257 If you queue up many requests in a loop it it often improves speed 1577 no practical limit on the number of outstanding requests.
258 if you set this to a relatively low number, such as 100.
259 1578
260 Under normal circumstances you don't need to call this function. 1579 STATISTICAL INFORMATION
1580 IO::AIO::nreqs
1581 Returns the number of requests currently in the ready, execute or
1582 pending states (i.e. for which their callback has not been invoked
1583 yet).
1584
1585 Example: wait till there are no outstanding requests anymore:
1586
1587 IO::AIO::poll_wait, IO::AIO::poll_cb
1588 while IO::AIO::nreqs;
1589
1590 IO::AIO::nready
1591 Returns the number of requests currently in the ready state (not yet
1592 executed).
1593
1594 IO::AIO::npending
1595 Returns the number of requests currently in the pending state
1596 (executed, but not yet processed by poll_cb).
1597
1598 SUBSECOND STAT TIME ACCESS
1599 Both "aio_stat"/"aio_lstat" and perl's "stat"/"lstat" functions can
1600 generally find access/modification and change times with subsecond time
1601 accuracy of the system supports it, but perl's built-in functions only
1602 return the integer part.
1603
1604 The following functions return the timestamps of the most recent stat
1605 with subsecond precision on most systems and work both after
1606 "aio_stat"/"aio_lstat" and perl's "stat"/"lstat" calls. Their return
1607 value is only meaningful after a successful "stat"/"lstat" call, or
1608 during/after a successful "aio_stat"/"aio_lstat" callback.
1609
1610 This is similar to the Time::HiRes "stat" functions, but can return full
1611 resolution without rounding and work with standard perl "stat",
1612 alleviating the need to call the special "Time::HiRes" functions, which
1613 do not act like their perl counterparts.
1614
1615 On operating systems or file systems where subsecond time resolution is
1616 not supported or could not be detected, a fractional part of 0 is
1617 returned, so it is always safe to call these functions.
1618
1619 $seconds = IO::AIO::st_atime, IO::AIO::st_mtime, IO::AIO::st_ctime
1620 Return the access, modication or change time, respectively,
1621 including fractional part. Due to the limited precision of floating
1622 point, the accuracy on most platforms is only a bit better than
1623 milliseconds for times around now - see the *nsec* function family,
1624 below, for full accuracy.
1625
1626 ($atime, $mtime, $ctime, ...) = IO::AIO::st_xtime
1627 Returns access, modification and change time all in one go, and
1628 maybe more times in the future version.
1629
1630 $nanoseconds = IO::AIO::st_atimensec, IO::AIO::st_mtimensec,
1631 IO::AIO::st_ctimensec
1632 Return the fractional access, modifcation or change time, in
1633 nanoseconds, as an integer in the range 0 to 999999999.
1634
1635 ($atime, $mtime, $ctime, ...) = IO::AIO::st_xtimensec
1636 Like the functions above, but returns all three times in one go (and
1637 maybe more in future versions).
1638
1639 Example: print the high resolution modification time of /etc, using
1640 "stat", and "IO::AIO::aio_stat".
1641
1642 if (stat "/etc") {
1643 printf "stat(/etc) mtime: %f\n", IO::AIO::st_mtime;
1644 }
1645
1646 IO::AIO::aio_stat "/etc", sub {
1647 $_[0]
1648 and return;
1649
1650 printf "aio_stat(/etc) mtime: %d.%09d\n", (stat _)[9], IO::AIO::st_mtimensec;
1651 };
1652
1653 IO::AIO::flush;
1654
1655 Output of the awbove on my system, showing reduced and full accuracy:
1656
1657 stat(/etc) mtime: 1534043702.020808
1658 aio_stat(/etc) mtime: 1534043702.020807792
1659
1660 MISCELLANEOUS FUNCTIONS
1661 IO::AIO implements some functions that are useful when you want to use
1662 some "Advanced I/O" function not available to in Perl, without going the
1663 "Asynchronous I/O" route. Many of these have an asynchronous "aio_*"
1664 counterpart.
1665
1666 $numfd = IO::AIO::get_fdlimit
1667 This function is *EXPERIMENTAL* and subject to change.
1668
1669 Tries to find the current file descriptor limit and returns it, or
1670 "undef" and sets $! in case of an error. The limit is one larger
1671 than the highest valid file descriptor number.
1672
1673 IO::AIO::min_fdlimit [$numfd]
1674 This function is *EXPERIMENTAL* and subject to change.
1675
1676 Try to increase the current file descriptor limit(s) to at least
1677 $numfd by changing the soft or hard file descriptor resource limit.
1678 If $numfd is missing, it will try to set a very high limit, although
1679 this is not recommended when you know the actual minimum that you
1680 require.
1681
1682 If the limit cannot be raised enough, the function makes a
1683 best-effort attempt to increase the limit as much as possible, using
1684 various tricks, while still failing. You can query the resulting
1685 limit using "IO::AIO::get_fdlimit".
1686
1687 If an error occurs, returns "undef" and sets $!, otherwise returns
1688 true.
1689
1690 IO::AIO::sendfile $ofh, $ifh, $offset, $count
1691 Calls the "eio_sendfile_sync" function, which is like
1692 "aio_sendfile", but is blocking (this makes most sense if you know
1693 the input data is likely cached already and the output filehandle is
1694 set to non-blocking operations).
1695
1696 Returns the number of bytes copied, or -1 on error.
1697
1698 IO::AIO::fadvise $fh, $offset, $len, $advice
1699 Simply calls the "posix_fadvise" function (see its manpage for
1700 details). The following advice constants are available:
1701 "IO::AIO::FADV_NORMAL", "IO::AIO::FADV_SEQUENTIAL",
1702 "IO::AIO::FADV_RANDOM", "IO::AIO::FADV_NOREUSE",
1703 "IO::AIO::FADV_WILLNEED", "IO::AIO::FADV_DONTNEED".
1704
1705 On systems that do not implement "posix_fadvise", this function
1706 returns ENOSYS, otherwise the return value of "posix_fadvise".
1707
1708 IO::AIO::madvise $scalar, $offset, $len, $advice
1709 Simply calls the "posix_madvise" function (see its manpage for
1710 details). The following advice constants are available:
1711 "IO::AIO::MADV_NORMAL", "IO::AIO::MADV_SEQUENTIAL",
1712 "IO::AIO::MADV_RANDOM", "IO::AIO::MADV_WILLNEED",
1713 "IO::AIO::MADV_DONTNEED".
1714
1715 If $offset is negative, counts from the end. If $length is negative,
1716 the remaining length of the $scalar is used. If possible, $length
1717 will be reduced to fit into the $scalar.
1718
1719 On systems that do not implement "posix_madvise", this function
1720 returns ENOSYS, otherwise the return value of "posix_madvise".
1721
1722 IO::AIO::mprotect $scalar, $offset, $len, $protect
1723 Simply calls the "mprotect" function on the preferably AIO::mmap'ed
1724 $scalar (see its manpage for details). The following protect
1725 constants are available: "IO::AIO::PROT_NONE", "IO::AIO::PROT_READ",
1726 "IO::AIO::PROT_WRITE", "IO::AIO::PROT_EXEC".
1727
1728 If $offset is negative, counts from the end. If $length is negative,
1729 the remaining length of the $scalar is used. If possible, $length
1730 will be reduced to fit into the $scalar.
1731
1732 On systems that do not implement "mprotect", this function returns
1733 ENOSYS, otherwise the return value of "mprotect".
1734
1735 IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset]
1736 Memory-maps a file (or anonymous memory range) and attaches it to
1737 the given $scalar, which will act like a string scalar. Returns true
1738 on success, and false otherwise.
1739
1740 The scalar must exist, but its contents do not matter - this means
1741 you cannot use a nonexistant array or hash element. When in doubt,
1742 "undef" the scalar first.
1743
1744 The only operations allowed on the mmapped scalar are
1745 "substr"/"vec", which don't change the string length, and most
1746 read-only operations such as copying it or searching it with regexes
1747 and so on.
1748
1749 Anything else is unsafe and will, at best, result in memory leaks.
1750
1751 The memory map associated with the $scalar is automatically removed
1752 when the $scalar is undef'd or destroyed, or when the
1753 "IO::AIO::mmap" or "IO::AIO::munmap" functions are called on it.
1754
1755 This calls the "mmap"(2) function internally. See your system's
1756 manual page for details on the $length, $prot and $flags parameters.
1757
1758 The $length must be larger than zero and smaller than the actual
1759 filesize.
1760
1761 $prot is a combination of "IO::AIO::PROT_NONE",
1762 "IO::AIO::PROT_EXEC", "IO::AIO::PROT_READ" and/or
1763 "IO::AIO::PROT_WRITE",
1764
1765 $flags can be a combination of "IO::AIO::MAP_SHARED" or
1766 "IO::AIO::MAP_PRIVATE", or a number of system-specific flags (when
1767 not available, the are 0): "IO::AIO::MAP_ANONYMOUS" (which is set to
1768 "MAP_ANON" if your system only provides this constant),
1769 "IO::AIO::MAP_LOCKED", "IO::AIO::MAP_NORESERVE",
1770 "IO::AIO::MAP_POPULATE", "IO::AIO::MAP_NONBLOCK",
1771 "IO::AIO::MAP_FIXED", "IO::AIO::MAP_GROWSDOWN",
1772 "IO::AIO::MAP_32BIT", "IO::AIO::MAP_HUGETLB" or
1773 "IO::AIO::MAP_STACK".
1774
1775 If $fh is "undef", then a file descriptor of -1 is passed.
1776
1777 $offset is the offset from the start of the file - it generally must
1778 be a multiple of "IO::AIO::PAGESIZE" and defaults to 0.
1779
1780 Example:
1781
1782 use Digest::MD5;
1783 use IO::AIO;
1784
1785 open my $fh, "<verybigfile"
1786 or die "$!";
1787
1788 IO::AIO::mmap my $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh
1789 or die "verybigfile: $!";
1790
1791 my $fast_md5 = md5 $data;
1792
1793 IO::AIO::munmap $scalar
1794 Removes a previous mmap and undefines the $scalar.
1795
1796 IO::AIO::mremap $scalar, $new_length, $flags = MREMAP_MAYMOVE[,
1797 $new_address = 0]
1798 Calls the Linux-specific mremap(2) system call. The $scalar must
1799 have been mapped by "IO::AIO::mmap", and $flags must currently
1800 either be 0 or "IO::AIO::MREMAP_MAYMOVE".
1801
1802 Returns true if successful, and false otherwise. If the underlying
1803 mmapped region has changed address, then the true value has the
1804 numerical value 1, otherwise it has the numerical value 0:
1805
1806 my $success = IO::AIO::mremap $mmapped, 8192, IO::AIO::MREMAP_MAYMOVE
1807 or die "mremap: $!";
1808
1809 if ($success*1) {
1810 warn "scalar has chanegd address in memory\n";
1811 }
1812
1813 "IO::AIO::MREMAP_FIXED" and the $new_address argument are currently
1814 implemented, but not supported and might go away in a future
1815 version.
1816
1817 On systems where this call is not supported or is not emulated, this
1818 call returns falls and sets $! to "ENOSYS".
1819
1820 IO::AIO::munlock $scalar, $offset = 0, $length = undef
1821 Calls the "munlock" function, undoing the effects of a previous
1822 "aio_mlock" call (see its description for details).
1823
1824 IO::AIO::munlockall
1825 Calls the "munlockall" function.
1826
1827 On systems that do not implement "munlockall", this function returns
1828 ENOSYS, otherwise the return value of "munlockall".
1829
1830 IO::AIO::splice $r_fh, $r_off, $w_fh, $w_off, $length, $flags
1831 Calls the GNU/Linux splice(2) syscall, if available. If $r_off or
1832 $w_off are "undef", then "NULL" is passed for these, otherwise they
1833 should be the file offset.
1834
1835 $r_fh and $w_fh should not refer to the same file, as splice might
1836 silently corrupt the data in this case.
1837
1838 The following symbol flag values are available:
1839 "IO::AIO::SPLICE_F_MOVE", "IO::AIO::SPLICE_F_NONBLOCK",
1840 "IO::AIO::SPLICE_F_MORE" and "IO::AIO::SPLICE_F_GIFT".
1841
1842 See the splice(2) manpage for details.
1843
1844 IO::AIO::tee $r_fh, $w_fh, $length, $flags
1845 Calls the GNU/Linux tee(2) syscall, see its manpage and the
1846 description for "IO::AIO::splice" above for details.
1847
1848 $actual_size = IO::AIO::pipesize $r_fh[, $new_size]
1849 Attempts to query or change the pipe buffer size. Obviously works
1850 only on pipes, and currently works only on GNU/Linux systems, and
1851 fails with -1/"ENOSYS" everywhere else. If anybody knows how to
1852 influence pipe buffer size on other systems, drop me a note.
1853
1854 ($rfh, $wfh) = IO::AIO::pipe2 [$flags]
1855 This is a direct interface to the Linux pipe2(2) system call. If
1856 $flags is missing or 0, then this should be the same as a call to
1857 perl's built-in "pipe" function and create a new pipe, and works on
1858 systems that lack the pipe2 syscall. On win32, this case invokes
1859 "_pipe (..., 4096, O_BINARY)".
1860
1861 If $flags is non-zero, it tries to invoke the pipe2 system call with
1862 the given flags (Linux 2.6.27, glibc 2.9).
1863
1864 On success, the read and write file handles are returned.
1865
1866 On error, nothing will be returned. If the pipe2 syscall is missing
1867 and $flags is non-zero, fails with "ENOSYS".
1868
1869 Please refer to pipe2(2) for more info on the $flags, but at the
1870 time of this writing, "IO::AIO::O_CLOEXEC", "IO::AIO::O_NONBLOCK"
1871 and "IO::AIO::O_DIRECT" (Linux 3.4, for packet-based pipes) were
1872 supported.
1873
1874 Example: create a pipe race-free w.r.t. threads and fork:
1875
1876 my ($rfh, $wfh) = IO::AIO::pipe2 IO::AIO::O_CLOEXEC
1877 or die "pipe2: $!\n";
1878
1879 $fh = IO::AIO::eventfd [$initval, [$flags]]
1880 This is a direct interface to the Linux eventfd(2) system call. The
1881 (unhelpful) defaults for $initval and $flags are 0 for both.
1882
1883 On success, the new eventfd filehandle is returned, otherwise
1884 returns "undef". If the eventfd syscall is missing, fails with
1885 "ENOSYS".
1886
1887 Please refer to eventfd(2) for more info on this call.
1888
1889 The following symbol flag values are available:
1890 "IO::AIO::EFD_CLOEXEC", "IO::AIO::EFD_NONBLOCK" and
1891 "IO::AIO::EFD_SEMAPHORE" (Linux 2.6.30).
1892
1893 Example: create a new eventfd filehandle:
1894
1895 $fh = IO::AIO::eventfd 0, IO::AIO::O_CLOEXEC
1896 or die "eventfd: $!\n";
1897
1898 $fh = IO::AIO::timerfd_create $clockid[, $flags]
1899 This is a direct interface to the Linux timerfd_create(2) system
1900 call. The (unhelpful) default for $flags is 0.
1901
1902 On success, the new timerfd filehandle is returned, otherwise
1903 returns "undef". If the eventfd syscall is missing, fails with
1904 "ENOSYS".
1905
1906 Please refer to timerfd_create(2) for more info on this call.
1907
1908 The following $clockid values are available:
1909 "IO::AIO::CLOCK_REALTIME", "IO::AIO::CLOCK_MONOTONIC"
1910 "IO::AIO::CLOCK_CLOCK_BOOTTIME" (Linux 3.15)
1911 "IO::AIO::CLOCK_CLOCK_REALTIME_ALARM" (Linux 3.11) and
1912 "IO::AIO::CLOCK_CLOCK_BOOTTIME_ALARM" (Linux 3.11).
1913
1914 The following $flags values are available (Linux 2.6.27):
1915 "IO::AIO::TFD_NONBLOCK" and "IO::AIO::TFD_CLOEXEC".
1916
1917 Example: create a new timerfd and set it to one-second repeated
1918 alarms, then wait for two alarms:
1919
1920 my $fh = IO::AIO::timerfd_create IO::AIO::CLOCK_BOOTTIME, IO::AIO::TFD_CLOEXEC
1921 or die "timerfd_create: $!\n";
1922
1923 defined IO::AIO::timerfd_settime $fh, 0, 1, 1
1924 or die "timerfd_settime: $!\n";
1925
1926 for (1..2) {
1927 8 == sysread $fh, my $buf, 8
1928 or die "timerfd read failure\n";
1929
1930 printf "number of expirations (likely 1): %d\n",
1931 unpack "Q", $buf;
1932 }
1933
1934 ($cur_interval, $cur_value) = IO::AIO::timerfd_settime $fh, $flags,
1935 $new_interval, $nbw_value
1936 This is a direct interface to the Linux timerfd_settime(2) system
1937 call. Please refer to its manpage for more info on this call.
1938
1939 The new itimerspec is specified using two (possibly fractional)
1940 second values, $new_interval and $new_value).
1941
1942 On success, the current interval and value are returned (as per
1943 "timerfd_gettime"). On failure, the empty list is returned.
1944
1945 The following $flags values are available:
1946 "IO::AIO::TFD_TIMER_ABSTIME" and "IO::AIO::TFD_TIMER_CANCEL_ON_SET".
1947
1948 See "IO::AIO::timerfd_create" for a full example.
1949
1950 ($cur_interval, $cur_value) = IO::AIO::timerfd_gettime $fh
1951 This is a direct interface to the Linux timerfd_gettime(2) system
1952 call. Please refer to its manpage for more info on this call.
1953
1954 On success, returns the current values of interval and value for the
1955 given timerfd (as potentially fractional second values). On failure,
1956 the empty list is returned.
1957
1958EVENT LOOP INTEGRATION
1959 It is recommended to use AnyEvent::AIO to integrate IO::AIO
1960 automatically into many event loops:
1961
1962 # AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...)
1963 use AnyEvent::AIO;
1964
1965 You can also integrate IO::AIO manually into many event loops, here are
1966 some examples of how to do this:
1967
1968 # EV integration
1969 my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb;
1970
1971 # Event integration
1972 Event->io (fd => IO::AIO::poll_fileno,
1973 poll => 'r',
1974 cb => \&IO::AIO::poll_cb);
1975
1976 # Glib/Gtk2 integration
1977 add_watch Glib::IO IO::AIO::poll_fileno,
1978 in => sub { IO::AIO::poll_cb; 1 };
1979
1980 # Tk integration
1981 Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
1982 readable => \&IO::AIO::poll_cb);
1983
1984 # Danga::Socket integration
1985 Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
1986 \&IO::AIO::poll_cb);
1987
1988 FORK BEHAVIOUR
1989 Usage of pthreads in a program changes the semantics of fork
1990 considerably. Specifically, only async-safe functions can be called
1991 after fork. Perl doesn't know about this, so in general, you cannot call
1992 fork with defined behaviour in perl if pthreads are involved. IO::AIO
1993 uses pthreads, so this applies, but many other extensions and (for
1994 inexplicable reasons) perl itself often is linked against pthreads, so
1995 this limitation applies to quite a lot of perls.
1996
1997 This module no longer tries to fight your OS, or POSIX. That means
1998 IO::AIO only works in the process that loaded it. Forking is fully
1999 supported, but using IO::AIO in the child is not.
2000
2001 You might get around by not *using* IO::AIO before (or after) forking.
2002 You could also try to call the IO::AIO::reinit function in the child:
2003
2004 IO::AIO::reinit
2005 Abandons all current requests and I/O threads and simply
2006 reinitialises all data structures. This is not an operation
2007 supported by any standards, but happens to work on GNU/Linux and
2008 some newer BSD systems.
2009
2010 The only reasonable use for this function is to call it after
2011 forking, if "IO::AIO" was used in the parent. Calling it while
2012 IO::AIO is active in the process will result in undefined behaviour.
2013 Calling it at any time will also result in any undefined (by POSIX)
2014 behaviour.
2015
2016 LINUX-SPECIFIC CALLS
2017 When a call is documented as "linux-specific" then this means it
2018 originated on GNU/Linux. "IO::AIO" will usually try to autodetect the
2019 availability and compatibility of such calls regardless of the platform
2020 it is compiled on, so platforms such as FreeBSD which often implement
2021 these calls will work. When in doubt, call them and see if they fail wth
2022 "ENOSYS".
2023
2024 MEMORY USAGE
2025 Per-request usage:
2026
2027 Each aio request uses - depending on your architecture - around 100-200
2028 bytes of memory. In addition, stat requests need a stat buffer (possibly
2029 a few hundred bytes), readdir requires a result buffer and so on. Perl
2030 scalars and other data passed into aio requests will also be locked and
2031 will consume memory till the request has entered the done state.
2032
2033 This is not awfully much, so queuing lots of requests is not usually a
2034 problem.
2035
2036 Per-thread usage:
2037
2038 In the execution phase, some aio requests require more memory for
2039 temporary buffers, and each thread requires a stack and other data
2040 structures (usually around 16k-128k, depending on the OS).
2041
2042KNOWN BUGS
2043 Known bugs will be fixed in the next release :)
2044
2045KNOWN ISSUES
2046 Calls that try to "import" foreign memory areas (such as "IO::AIO::mmap"
2047 or "IO::AIO::aio_slurp") do not work with generic lvalues, such as
2048 non-created hash slots or other scalars I didn't think of. It's best to
2049 avoid such and either use scalar variables or making sure that the
2050 scalar exists (e.g. by storing "undef") and isn't "funny" (e.g. tied).
2051
2052 I am not sure anything can be done about this, so this is considered a
2053 known issue, rather than a bug.
261 2054
262SEE ALSO 2055SEE ALSO
263 Coro, Linux::AIO. 2056 AnyEvent::AIO for easy integration into event loops, Coro::AIO for a
2057 more natural syntax.
264 2058
265AUTHOR 2059AUTHOR
266 Marc Lehmann <schmorp@schmorp.de> 2060 Marc Lehmann <schmorp@schmorp.de>
267 http://home.schmorp.de/ 2061 http://home.schmorp.de/
268 2062

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines