1 |
NAME |
2 |
IO::AIO - Asynchronous/Advanced Input/Output |
3 |
|
4 |
SYNOPSIS |
5 |
use IO::AIO; |
6 |
|
7 |
aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub { |
8 |
my $fh = shift |
9 |
or die "/etc/passwd: $!"; |
10 |
... |
11 |
}; |
12 |
|
13 |
aio_unlink "/tmp/file", sub { }; |
14 |
|
15 |
aio_read $fh, 30000, 1024, $buffer, 0, sub { |
16 |
$_[0] > 0 or die "read error: $!"; |
17 |
}; |
18 |
|
19 |
# version 2+ has request and group objects |
20 |
use IO::AIO 2; |
21 |
|
22 |
aioreq_pri 4; # give next request a very high priority |
23 |
my $req = aio_unlink "/tmp/file", sub { }; |
24 |
$req->cancel; # cancel request if still in queue |
25 |
|
26 |
my $grp = aio_group sub { print "all stats done\n" }; |
27 |
add $grp aio_stat "..." for ...; |
28 |
|
29 |
DESCRIPTION |
30 |
This module implements asynchronous I/O using whatever means your |
31 |
operating system supports. It is implemented as an interface to "libeio" |
32 |
(<http://software.schmorp.de/pkg/libeio.html>). |
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 |
|
50 |
In this version, a number of threads are started that execute your |
51 |
requests and signal their completion. You don't need thread support in |
52 |
perl, and the threads created by this module will not be visible to |
53 |
perl. In the future, this module might make use of the native aio |
54 |
functions available on many operating systems. However, they are often |
55 |
not well-supported or restricted (GNU/Linux doesn't allow them on normal |
56 |
files currently, for example), and they would only support aio_read and |
57 |
aio_write, so the remaining functionality would have to be implemented |
58 |
using threads anyway. |
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 |
|
64 |
Although the module will work in the presence of other (Perl-) threads, |
65 |
it is currently not reentrant in any way, so use appropriate locking |
66 |
yourself, always call "poll_cb" from within the same thread, or never |
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 |
|
109 |
REQUEST 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). |
151 |
|
152 |
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 |
IO::AIO::reinit |
227 |
|
228 |
$nfd = IO::AIO::get_fdlimit |
229 |
IO::AIO::min_fdlimit $nfd |
230 |
|
231 |
IO::AIO::sendfile $ofh, $ifh, $offset, $count |
232 |
IO::AIO::fadvise $fh, $offset, $len, $advice |
233 |
IO::AIO::fexecve $fh, $argv, $envp |
234 |
|
235 |
IO::AIO::mmap $scalar, $length, $prot, $flags[, $fh[, $offset]] |
236 |
IO::AIO::munmap $scalar |
237 |
IO::AIO::mremap $scalar, $new_length, $flags[, $new_address] |
238 |
IO::AIO::madvise $scalar, $offset, $length, $advice |
239 |
IO::AIO::mprotect $scalar, $offset, $length, $protect |
240 |
IO::AIO::munlock $scalar, $offset = 0, $length = undef |
241 |
IO::AIO::munlockall |
242 |
|
243 |
# stat extensions |
244 |
$counter = IO::AIO::st_gen |
245 |
$seconds = IO::AIO::st_atime, IO::AIO::st_mtime, IO::AIO::st_ctime, IO::AIO::st_btime |
246 |
($atime, $mtime, $ctime, $btime, ...) = IO::AIO::st_xtime |
247 |
$nanoseconds = IO::AIO::st_atimensec, IO::AIO::st_mtimensec, IO::AIO::st_ctimensec, IO::AIO::st_btimensec |
248 |
$seconds = IO::AIO::st_btimesec |
249 |
($atime, $mtime, $ctime, $btime, ...) = IO::AIO::st_xtimensec |
250 |
|
251 |
# very much unportable syscalls |
252 |
IO::AIO::accept4 $r_fh, $sockaddr, $sockaddr_len, $flags |
253 |
IO::AIO::splice $r_fh, $r_off, $w_fh, $w_off, $length, $flags |
254 |
IO::AIO::tee $r_fh, $w_fh, $length, $flags |
255 |
|
256 |
$actual_size = IO::AIO::pipesize $r_fh[, $new_size] |
257 |
($rfh, $wfh) = IO::AIO::pipe2 [$flags] |
258 |
|
259 |
$fh = IO::AIO::eventfd [$initval, [$flags]] |
260 |
$fh = IO::AIO::memfd_create $pathname[, $flags] |
261 |
|
262 |
$fh = IO::AIO::timerfd_create $clockid[, $flags] |
263 |
($cur_interval, $cur_value) = IO::AIO::timerfd_settime $fh, $flags, $new_interval, $nbw_value |
264 |
($cur_interval, $cur_value) = IO::AIO::timerfd_gettime $fh |
265 |
|
266 |
$fh = IO::AIO::pidfd_open $pid[, $flags] |
267 |
$status = IO::AIO::pidfd_send_signal $pidfh, $signal[, $siginfo[, $flags]] |
268 |
$fh = IO::AIO::pidfd_getfd $pidfh, $targetfd[, $flags] |
269 |
|
270 |
$retval = IO::AIO::mount $special, $path, $fstype, $flags = 0, $data = undef |
271 |
$retval = IO::AIO::umount $path, $flags = 0 |
272 |
|
273 |
API NOTES |
274 |
All the "aio_*" calls are more or less thin wrappers around the syscall |
275 |
with the same name (sans "aio_"). The arguments are similar or |
276 |
identical, and they all accept an additional (and optional) $callback |
277 |
argument which must be a code reference. This code reference will be |
278 |
called after the syscall has been executed in an asynchronous fashion. |
279 |
The results of the request will be passed as arguments to the callback |
280 |
(and, if an error occured, in $!) - for most requests the syscall return |
281 |
code (e.g. most syscalls return -1 on error, unlike perl, which usually |
282 |
delivers "false"). |
283 |
|
284 |
Some requests (such as "aio_readdir") pass the actual results and |
285 |
communicate failures by passing "undef". |
286 |
|
287 |
All functions expecting a filehandle keep a copy of the filehandle |
288 |
internally until the request has finished. |
289 |
|
290 |
All functions return request objects of type IO::AIO::REQ that allow |
291 |
further manipulation of those requests while they are in-flight. |
292 |
|
293 |
The pathnames you pass to these routines *should* be absolute. The |
294 |
reason for this is that at the time the request is being executed, the |
295 |
current working directory could have changed. Alternatively, you can |
296 |
make sure that you never change the current working directory anywhere |
297 |
in the program and then use relative paths. You can also take advantage |
298 |
of IO::AIOs working directory abstraction, that lets you specify paths |
299 |
relative to some previously-opened "working directory object" - see the |
300 |
description of the "IO::AIO::WD" class later in this document. |
301 |
|
302 |
To encode pathnames as octets, either make sure you either: a) always |
303 |
pass in filenames you got from outside (command line, readdir etc.) |
304 |
without tinkering, b) are in your native filesystem encoding, c) use the |
305 |
Encode module and encode your pathnames to the locale (or other) |
306 |
encoding in effect in the user environment, d) use |
307 |
Glib::filename_from_unicode on unicode filenames or e) use something |
308 |
else to ensure your scalar has the correct contents. |
309 |
|
310 |
This works, btw. independent of the internal UTF-8 bit, which IO::AIO |
311 |
handles correctly whether it is set or not. |
312 |
|
313 |
AIO REQUEST FUNCTIONS |
314 |
$prev_pri = aioreq_pri [$pri] |
315 |
Returns the priority value that would be used for the next request |
316 |
and, if $pri is given, sets the priority for the next aio request. |
317 |
|
318 |
The default priority is 0, the minimum and maximum priorities are -4 |
319 |
and 4, respectively. Requests with higher priority will be serviced |
320 |
first. |
321 |
|
322 |
The priority will be reset to 0 after each call to one of the |
323 |
"aio_*" functions. |
324 |
|
325 |
Example: open a file with low priority, then read something from it |
326 |
with higher priority so the read request is serviced before other |
327 |
low priority open requests (potentially spamming the cache): |
328 |
|
329 |
aioreq_pri -3; |
330 |
aio_open ..., sub { |
331 |
return unless $_[0]; |
332 |
|
333 |
aioreq_pri -2; |
334 |
aio_read $_[0], ..., sub { |
335 |
... |
336 |
}; |
337 |
}; |
338 |
|
339 |
aioreq_nice $pri_adjust |
340 |
Similar to "aioreq_pri", but subtracts the given value from the |
341 |
current priority, so the effect is cumulative. |
342 |
|
343 |
aio_open $pathname, $flags, $mode, $callback->($fh) |
344 |
Asynchronously open or create a file and call the callback with a |
345 |
newly created filehandle for the file (or "undef" in case of an |
346 |
error). |
347 |
|
348 |
The $flags argument is a bitmask. See the "Fcntl" module for a list. |
349 |
They are the same as used by "sysopen". |
350 |
|
351 |
Likewise, $mode specifies the mode of the newly created file, if it |
352 |
didn't exist and "O_CREAT" has been given, just like perl's |
353 |
"sysopen", except that it is mandatory (i.e. use 0 if you don't |
354 |
create new files, and 0666 or 0777 if you do). Note that the $mode |
355 |
will be modified by the umask in effect then the request is being |
356 |
executed, so better never change the umask. |
357 |
|
358 |
Example: |
359 |
|
360 |
aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub { |
361 |
if ($_[0]) { |
362 |
print "open successful, fh is $_[0]\n"; |
363 |
... |
364 |
} else { |
365 |
die "open failed: $!\n"; |
366 |
} |
367 |
}; |
368 |
|
369 |
In addition to all the common open modes/flags ("O_RDONLY", |
370 |
"O_WRONLY", "O_RDWR", "O_CREAT", "O_TRUNC", "O_EXCL" and |
371 |
"O_APPEND"), the following POSIX and non-POSIX constants are |
372 |
available (missing ones on your system are, as usual, 0): |
373 |
|
374 |
"O_ASYNC", "O_DIRECT", "O_NOATIME", "O_CLOEXEC", "O_NOCTTY", |
375 |
"O_NOFOLLOW", "O_NONBLOCK", "O_EXEC", "O_SEARCH", "O_DIRECTORY", |
376 |
"O_DSYNC", "O_RSYNC", "O_SYNC", "O_PATH", "O_TMPFILE", "O_TTY_INIT" |
377 |
and "O_ACCMODE". |
378 |
|
379 |
aio_close $fh, $callback->($status) |
380 |
Asynchronously close a file and call the callback with the result |
381 |
code. |
382 |
|
383 |
Unfortunately, you can't do this to perl. Perl *insists* very |
384 |
strongly on closing the file descriptor associated with the |
385 |
filehandle itself. |
386 |
|
387 |
Therefore, "aio_close" will not close the filehandle - instead it |
388 |
will use dup2 to overwrite the file descriptor with the write-end of |
389 |
a pipe (the pipe fd will be created on demand and will be cached). |
390 |
|
391 |
Or in other words: the file descriptor will be closed, but it will |
392 |
not be free for reuse until the perl filehandle is closed. |
393 |
|
394 |
aio_seek $fh, $offset, $whence, $callback->($offs) |
395 |
Seeks the filehandle to the new $offset, similarly to perl's |
396 |
"sysseek". The $whence can use the traditional values (0 for |
397 |
"IO::AIO::SEEK_SET", 1 for "IO::AIO::SEEK_CUR" or 2 for |
398 |
"IO::AIO::SEEK_END"). |
399 |
|
400 |
The resulting absolute offset will be passed to the callback, or -1 |
401 |
in case of an error. |
402 |
|
403 |
In theory, the $whence constants could be different than the |
404 |
corresponding values from Fcntl, but perl guarantees they are the |
405 |
same, so don't panic. |
406 |
|
407 |
As a GNU/Linux (and maybe Solaris) extension, also the constants |
408 |
"IO::AIO::SEEK_DATA" and "IO::AIO::SEEK_HOLE" are available, if they |
409 |
could be found. No guarantees about suitability for use in |
410 |
"aio_seek" or Perl's "sysseek" can be made though, although I would |
411 |
naively assume they "just work". |
412 |
|
413 |
aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
414 |
aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
415 |
Reads or writes $length bytes from or to the specified $fh and |
416 |
$offset into the scalar given by $data and offset $dataoffset and |
417 |
calls the callback with the actual number of bytes transferred (or |
418 |
-1 on error, just like the syscall). |
419 |
|
420 |
"aio_read" will, like "sysread", shrink or grow the $data scalar to |
421 |
offset plus the actual number of bytes read. |
422 |
|
423 |
If $offset is undefined, then the current file descriptor offset |
424 |
will be used (and updated), otherwise the file descriptor offset |
425 |
will not be changed by these calls. |
426 |
|
427 |
If $length is undefined in "aio_write", use the remaining length of |
428 |
$data. |
429 |
|
430 |
If $dataoffset is less than zero, it will be counted from the end of |
431 |
$data. |
432 |
|
433 |
The $data scalar *MUST NOT* be modified in any way while the request |
434 |
is outstanding. Modifying it can result in segfaults or World War |
435 |
III (if the necessary/optional hardware is installed). |
436 |
|
437 |
Example: Read 15 bytes at offset 7 into scalar $buffer, starting at |
438 |
offset 0 within the scalar: |
439 |
|
440 |
aio_read $fh, 7, 15, $buffer, 0, sub { |
441 |
$_[0] > 0 or die "read error: $!"; |
442 |
print "read $_[0] bytes: <$buffer>\n"; |
443 |
}; |
444 |
|
445 |
aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval) |
446 |
Tries to copy $length bytes from $in_fh to $out_fh. It starts |
447 |
reading at byte offset $in_offset, and starts writing at the current |
448 |
file offset of $out_fh. Because of that, it is not safe to issue |
449 |
more than one "aio_sendfile" per $out_fh, as they will interfere |
450 |
with each other. The same $in_fh works fine though, as this function |
451 |
does not move or use the file offset of $in_fh. |
452 |
|
453 |
Please note that "aio_sendfile" can read more bytes from $in_fh than |
454 |
are written, and there is no way to find out how many more bytes |
455 |
have been read from "aio_sendfile" alone, as "aio_sendfile" only |
456 |
provides the number of bytes written to $out_fh. Only if the result |
457 |
value equals $length one can assume that $length bytes have been |
458 |
read. |
459 |
|
460 |
Unlike with other "aio_" functions, it makes a lot of sense to use |
461 |
"aio_sendfile" on non-blocking sockets, as long as one end |
462 |
(typically the $in_fh) is a file - the file I/O will then be |
463 |
asynchronous, while the socket I/O will be non-blocking. Note, |
464 |
however, that you can run into a trap where "aio_sendfile" reads |
465 |
some data with readahead, then fails to write all data, and when the |
466 |
socket is ready the next time, the data in the cache is already |
467 |
lost, forcing "aio_sendfile" to again hit the disk. Explicit |
468 |
"aio_read" + "aio_write" let's you better control resource usage. |
469 |
|
470 |
This call tries to make use of a native "sendfile"-like syscall to |
471 |
provide zero-copy operation. For this to work, $out_fh should refer |
472 |
to a socket, and $in_fh should refer to an mmap'able file. |
473 |
|
474 |
If a native sendfile cannot be found or it fails with "ENOSYS", |
475 |
"EINVAL", "ENOTSUP", "EOPNOTSUPP", "EAFNOSUPPORT", "EPROTOTYPE" or |
476 |
"ENOTSOCK", it will be emulated, so you can call "aio_sendfile" on |
477 |
any type of filehandle regardless of the limitations of the |
478 |
operating system. |
479 |
|
480 |
As native sendfile syscalls (as practically any non-POSIX interface |
481 |
hacked together in a hurry to improve benchmark numbers) tend to be |
482 |
rather buggy on many systems, this implementation tries to work |
483 |
around some known bugs in Linux and FreeBSD kernels (probably |
484 |
others, too), but that might fail, so you really really should check |
485 |
the return value of "aio_sendfile" - fewer bytes than expected might |
486 |
have been transferred. |
487 |
|
488 |
aio_readahead $fh,$offset,$length, $callback->($retval) |
489 |
"aio_readahead" populates the page cache with data from a file so |
490 |
that subsequent reads from that file will not block on disk I/O. The |
491 |
$offset argument specifies the starting point from which data is to |
492 |
be read and $length specifies the number of bytes to be read. I/O is |
493 |
performed in whole pages, so that offset is effectively rounded down |
494 |
to a page boundary and bytes are read up to the next page boundary |
495 |
greater than or equal to (off-set+length). "aio_readahead" does not |
496 |
read beyond the end of the file. The current file offset of the file |
497 |
is left unchanged. |
498 |
|
499 |
If that syscall doesn't exist (likely if your kernel isn't Linux) it |
500 |
will be emulated by simply reading the data, which would have a |
501 |
similar effect. |
502 |
|
503 |
aio_stat $fh_or_path, $callback->($status) |
504 |
aio_lstat $fh, $callback->($status) |
505 |
Works almost exactly like perl's "stat" or "lstat" in void context. |
506 |
The callback will be called after the stat and the results will be |
507 |
available using "stat _" or "-s _" and other tests (with the |
508 |
exception of "-B" and "-T"). |
509 |
|
510 |
Currently, the stats are always 64-bit-stats, i.e. instead of |
511 |
returning an error when stat'ing a large file, the results will be |
512 |
silently truncated unless perl itself is compiled with large file |
513 |
support. |
514 |
|
515 |
To help interpret the mode and dev/rdev stat values, IO::AIO offers |
516 |
the following constants and functions (if not implemented, the |
517 |
constants will be 0 and the functions will either "croak" or fall |
518 |
back on traditional behaviour). |
519 |
|
520 |
"S_IFMT", "S_IFIFO", "S_IFCHR", "S_IFBLK", "S_IFLNK", "S_IFREG", |
521 |
"S_IFDIR", "S_IFWHT", "S_IFSOCK", "IO::AIO::major $dev_t", |
522 |
"IO::AIO::minor $dev_t", "IO::AIO::makedev $major, $minor". |
523 |
|
524 |
To access higher resolution stat timestamps, see "SUBSECOND STAT |
525 |
TIME ACCESS". |
526 |
|
527 |
Example: Print the length of /etc/passwd: |
528 |
|
529 |
aio_stat "/etc/passwd", sub { |
530 |
$_[0] and die "stat failed: $!"; |
531 |
print "size is ", -s _, "\n"; |
532 |
}; |
533 |
|
534 |
aio_statvfs $fh_or_path, $callback->($statvfs) |
535 |
Works like the POSIX "statvfs" or "fstatvfs" syscalls, depending on |
536 |
whether a file handle or path was passed. |
537 |
|
538 |
On success, the callback is passed a hash reference with the |
539 |
following members: "bsize", "frsize", "blocks", "bfree", "bavail", |
540 |
"files", "ffree", "favail", "fsid", "flag" and "namemax". On |
541 |
failure, "undef" is passed. |
542 |
|
543 |
The following POSIX IO::AIO::ST_* constants are defined: "ST_RDONLY" |
544 |
and "ST_NOSUID". |
545 |
|
546 |
The following non-POSIX IO::AIO::ST_* flag masks are defined to |
547 |
their correct value when available, or to 0 on systems that do not |
548 |
support them: "ST_NODEV", "ST_NOEXEC", "ST_SYNCHRONOUS", |
549 |
"ST_MANDLOCK", "ST_WRITE", "ST_APPEND", "ST_IMMUTABLE", |
550 |
"ST_NOATIME", "ST_NODIRATIME" and "ST_RELATIME". |
551 |
|
552 |
Example: stat "/wd" and dump out the data if successful. |
553 |
|
554 |
aio_statvfs "/wd", sub { |
555 |
my $f = $_[0] |
556 |
or die "statvfs: $!"; |
557 |
|
558 |
use Data::Dumper; |
559 |
say Dumper $f; |
560 |
}; |
561 |
|
562 |
# result: |
563 |
{ |
564 |
bsize => 1024, |
565 |
bfree => 4333064312, |
566 |
blocks => 10253828096, |
567 |
files => 2050765568, |
568 |
flag => 4096, |
569 |
favail => 2042092649, |
570 |
bavail => 4333064312, |
571 |
ffree => 2042092649, |
572 |
namemax => 255, |
573 |
frsize => 1024, |
574 |
fsid => 1810 |
575 |
} |
576 |
|
577 |
aio_utime $fh_or_path, $atime, $mtime, $callback->($status) |
578 |
Works like perl's "utime" function (including the special case of |
579 |
$atime and $mtime being undef). Fractional times are supported if |
580 |
the underlying syscalls support them. |
581 |
|
582 |
When called with a pathname, uses utimensat(2) or utimes(2) if |
583 |
available, otherwise utime(2). If called on a file descriptor, uses |
584 |
futimens(2) or futimes(2) if available, otherwise returns ENOSYS, so |
585 |
this is not portable. |
586 |
|
587 |
Examples: |
588 |
|
589 |
# set atime and mtime to current time (basically touch(1)): |
590 |
aio_utime "path", undef, undef; |
591 |
# set atime to current time and mtime to beginning of the epoch: |
592 |
aio_utime "path", time, undef; # undef==0 |
593 |
|
594 |
aio_chown $fh_or_path, $uid, $gid, $callback->($status) |
595 |
Works like perl's "chown" function, except that "undef" for either |
596 |
$uid or $gid is being interpreted as "do not change" (but -1 can |
597 |
also be used). |
598 |
|
599 |
Examples: |
600 |
|
601 |
# same as "chown root path" in the shell: |
602 |
aio_chown "path", 0, -1; |
603 |
# same as above: |
604 |
aio_chown "path", 0, undef; |
605 |
|
606 |
aio_truncate $fh_or_path, $offset, $callback->($status) |
607 |
Works like truncate(2) or ftruncate(2). |
608 |
|
609 |
aio_allocate $fh, $mode, $offset, $len, $callback->($status) |
610 |
Allocates or frees disk space according to the $mode argument. See |
611 |
the linux "fallocate" documentation for details. |
612 |
|
613 |
$mode is usually 0 or "IO::AIO::FALLOC_FL_KEEP_SIZE" to allocate |
614 |
space, or "IO::AIO::FALLOC_FL_PUNCH_HOLE | |
615 |
IO::AIO::FALLOC_FL_KEEP_SIZE", to deallocate a file range. |
616 |
|
617 |
IO::AIO also supports "FALLOC_FL_COLLAPSE_RANGE", to remove a range |
618 |
(without leaving a hole), "FALLOC_FL_ZERO_RANGE", to zero a range, |
619 |
"FALLOC_FL_INSERT_RANGE" to insert a range and |
620 |
"FALLOC_FL_UNSHARE_RANGE" to unshare shared blocks (see your |
621 |
fallocate(2) manpage). |
622 |
|
623 |
The file system block size used by "fallocate" is presumably the |
624 |
"f_bsize" returned by "statvfs", but different filesystems and |
625 |
filetypes can dictate other limitations. |
626 |
|
627 |
If "fallocate" isn't available or cannot be emulated (currently no |
628 |
emulation will be attempted), passes -1 and sets $! to "ENOSYS". |
629 |
|
630 |
aio_chmod $fh_or_path, $mode, $callback->($status) |
631 |
Works like perl's "chmod" function. |
632 |
|
633 |
aio_unlink $pathname, $callback->($status) |
634 |
Asynchronously unlink (delete) a file and call the callback with the |
635 |
result code. |
636 |
|
637 |
aio_mknod $pathname, $mode, $dev, $callback->($status) |
638 |
[EXPERIMENTAL] |
639 |
|
640 |
Asynchronously create a device node (or fifo). See mknod(2). |
641 |
|
642 |
The only (POSIX-) portable way of calling this function is: |
643 |
|
644 |
aio_mknod $pathname, IO::AIO::S_IFIFO | $mode, 0, sub { ... |
645 |
|
646 |
See "aio_stat" for info about some potentially helpful extra |
647 |
constants and functions. |
648 |
|
649 |
aio_link $srcpath, $dstpath, $callback->($status) |
650 |
Asynchronously create a new link to the existing object at $srcpath |
651 |
at the path $dstpath and call the callback with the result code. |
652 |
|
653 |
aio_symlink $srcpath, $dstpath, $callback->($status) |
654 |
Asynchronously create a new symbolic link to the existing object at |
655 |
$srcpath at the path $dstpath and call the callback with the result |
656 |
code. |
657 |
|
658 |
aio_readlink $pathname, $callback->($link) |
659 |
Asynchronously read the symlink specified by $path and pass it to |
660 |
the callback. If an error occurs, nothing or undef gets passed to |
661 |
the callback. |
662 |
|
663 |
aio_realpath $pathname, $callback->($path) |
664 |
Asynchronously make the path absolute and resolve any symlinks in |
665 |
$path. The resulting path only consists of directories (same as |
666 |
Cwd::realpath). |
667 |
|
668 |
This request can be used to get the absolute path of the current |
669 |
working directory by passing it a path of . (a single dot). |
670 |
|
671 |
aio_rename $srcpath, $dstpath, $callback->($status) |
672 |
Asynchronously rename the object at $srcpath to $dstpath, just as |
673 |
rename(2) and call the callback with the result code. |
674 |
|
675 |
On systems that support the AIO::WD working directory abstraction |
676 |
natively, the case "[$wd, "."]" as $srcpath is specialcased - |
677 |
instead of failing, "rename" is called on the absolute path of $wd. |
678 |
|
679 |
aio_rename2 $srcpath, $dstpath, $flags, $callback->($status) |
680 |
Basically a version of "aio_rename" with an additional $flags |
681 |
argument. Calling this with "$flags=0" is the same as calling |
682 |
"aio_rename". |
683 |
|
684 |
Non-zero flags are currently only supported on GNU/Linux systems |
685 |
that support renameat2. Other systems fail with "ENOSYS" in this |
686 |
case. |
687 |
|
688 |
The following constants are available (missing ones are, as usual |
689 |
0), see renameat2(2) for details: |
690 |
|
691 |
"IO::AIO::RENAME_NOREPLACE", "IO::AIO::RENAME_EXCHANGE" and |
692 |
"IO::AIO::RENAME_WHITEOUT". |
693 |
|
694 |
aio_mkdir $pathname, $mode, $callback->($status) |
695 |
Asynchronously mkdir (create) a directory and call the callback with |
696 |
the result code. $mode will be modified by the umask at the time the |
697 |
request is executed, so do not change your umask. |
698 |
|
699 |
aio_rmdir $pathname, $callback->($status) |
700 |
Asynchronously rmdir (delete) a directory and call the callback with |
701 |
the result code. |
702 |
|
703 |
On systems that support the AIO::WD working directory abstraction |
704 |
natively, the case "[$wd, "."]" is specialcased - instead of |
705 |
failing, "rmdir" is called on the absolute path of $wd. |
706 |
|
707 |
aio_readdir $pathname, $callback->($entries) |
708 |
Unlike the POSIX call of the same name, "aio_readdir" reads an |
709 |
entire directory (i.e. opendir + readdir + closedir). The entries |
710 |
will not be sorted, and will NOT include the "." and ".." entries. |
711 |
|
712 |
The callback is passed a single argument which is either "undef" or |
713 |
an array-ref with the filenames. |
714 |
|
715 |
aio_readdirx $pathname, $flags, $callback->($entries, $flags) |
716 |
Quite similar to "aio_readdir", but the $flags argument allows one |
717 |
to tune behaviour and output format. In case of an error, $entries |
718 |
will be "undef". |
719 |
|
720 |
The flags are a combination of the following constants, ORed |
721 |
together (the flags will also be passed to the callback, possibly |
722 |
modified): |
723 |
|
724 |
IO::AIO::READDIR_DENTS |
725 |
Normally the callback gets an arrayref consisting of names only |
726 |
(as with "aio_readdir"). If this flag is set, then the callback |
727 |
gets an arrayref with "[$name, $type, $inode]" arrayrefs, each |
728 |
describing a single directory entry in more detail: |
729 |
|
730 |
$name is the name of the entry. |
731 |
|
732 |
$type is one of the "IO::AIO::DT_xxx" constants: |
733 |
|
734 |
"IO::AIO::DT_UNKNOWN", "IO::AIO::DT_FIFO", "IO::AIO::DT_CHR", |
735 |
"IO::AIO::DT_DIR", "IO::AIO::DT_BLK", "IO::AIO::DT_REG", |
736 |
"IO::AIO::DT_LNK", "IO::AIO::DT_SOCK", "IO::AIO::DT_WHT". |
737 |
|
738 |
"IO::AIO::DT_UNKNOWN" means just that: readdir does not know. If |
739 |
you need to know, you have to run stat yourself. Also, for |
740 |
speed/memory reasons, the $type scalars are read-only: you must |
741 |
not modify them. |
742 |
|
743 |
$inode is the inode number (which might not be exact on systems |
744 |
with 64 bit inode numbers and 32 bit perls). This field has |
745 |
unspecified content on systems that do not deliver the inode |
746 |
information. |
747 |
|
748 |
IO::AIO::READDIR_DIRS_FIRST |
749 |
When this flag is set, then the names will be returned in an |
750 |
order where likely directories come first, in optimal stat |
751 |
order. This is useful when you need to quickly find directories, |
752 |
or you want to find all directories while avoiding to stat() |
753 |
each entry. |
754 |
|
755 |
If the system returns type information in readdir, then this is |
756 |
used to find directories directly. Otherwise, likely directories |
757 |
are names beginning with ".", or otherwise names with no dots, |
758 |
of which names with short names are tried first. |
759 |
|
760 |
IO::AIO::READDIR_STAT_ORDER |
761 |
When this flag is set, then the names will be returned in an |
762 |
order suitable for stat()'ing each one. That is, when you plan |
763 |
to stat() most or all files in the given directory, then the |
764 |
returned order will likely be faster. |
765 |
|
766 |
If both this flag and "IO::AIO::READDIR_DIRS_FIRST" are |
767 |
specified, then the likely dirs come first, resulting in a less |
768 |
optimal stat order for stat'ing all entries, but likely a more |
769 |
optimal order for finding subdirectories. |
770 |
|
771 |
IO::AIO::READDIR_FOUND_UNKNOWN |
772 |
This flag should not be set when calling "aio_readdirx". |
773 |
Instead, it is being set by "aio_readdirx", when any of the |
774 |
$type's found were "IO::AIO::DT_UNKNOWN". The absence of this |
775 |
flag therefore indicates that all $type's are known, which can |
776 |
be used to speed up some algorithms. |
777 |
|
778 |
aio_slurp $pathname, $offset, $length, $data, $callback->($status) |
779 |
Opens, reads and closes the given file. The data is put into $data, |
780 |
which is resized as required. |
781 |
|
782 |
If $offset is negative, then it is counted from the end of the file. |
783 |
|
784 |
If $length is zero, then the remaining length of the file is used. |
785 |
Also, in this case, the same limitations to modifying $data apply as |
786 |
when IO::AIO::mmap is used, i.e. it must only be modified in-place |
787 |
with "substr". If the size of the file is known, specifying a |
788 |
non-zero $length results in a performance advantage. |
789 |
|
790 |
This request is similar to the older "aio_load" request, but since |
791 |
it is a single request, it might be more efficient to use. |
792 |
|
793 |
Example: load /etc/passwd into $passwd. |
794 |
|
795 |
my $passwd; |
796 |
aio_slurp "/etc/passwd", 0, 0, $passwd, sub { |
797 |
$_[0] >= 0 |
798 |
or die "/etc/passwd: $!\n"; |
799 |
|
800 |
printf "/etc/passwd is %d bytes long, and contains:\n", length $passwd; |
801 |
print $passwd; |
802 |
}; |
803 |
IO::AIO::flush; |
804 |
|
805 |
aio_load $pathname, $data, $callback->($status) |
806 |
This is a composite request that tries to fully load the given file |
807 |
into memory. Status is the same as with aio_read. |
808 |
|
809 |
Using "aio_slurp" might be more efficient, as it is a single |
810 |
request. |
811 |
|
812 |
aio_copy $srcpath, $dstpath, $callback->($status) |
813 |
Try to copy the *file* (directories not supported as either source |
814 |
or destination) from $srcpath to $dstpath and call the callback with |
815 |
a status of 0 (ok) or -1 (error, see $!). |
816 |
|
817 |
Existing destination files will be truncated. |
818 |
|
819 |
This is a composite request that creates the destination file with |
820 |
mode 0200 and copies the contents of the source file into it using |
821 |
"aio_sendfile", followed by restoring atime, mtime, access mode and |
822 |
uid/gid, in that order. |
823 |
|
824 |
If an error occurs, the partial destination file will be unlinked, |
825 |
if possible, except when setting atime, mtime, access mode and |
826 |
uid/gid, where errors are being ignored. |
827 |
|
828 |
aio_move $srcpath, $dstpath, $callback->($status) |
829 |
Try to move the *file* (directories not supported as either source |
830 |
or destination) from $srcpath to $dstpath and call the callback with |
831 |
a status of 0 (ok) or -1 (error, see $!). |
832 |
|
833 |
This is a composite request that tries to rename(2) the file first; |
834 |
if rename fails with "EXDEV", it copies the file with "aio_copy" |
835 |
and, if that is successful, unlinks the $srcpath. |
836 |
|
837 |
aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs) |
838 |
Scans a directory (similar to "aio_readdir") but additionally tries |
839 |
to efficiently separate the entries of directory $path into two sets |
840 |
of names, directories you can recurse into (directories), and ones |
841 |
you cannot recurse into (everything else, including symlinks to |
842 |
directories). |
843 |
|
844 |
"aio_scandir" is a composite request that generates many sub |
845 |
requests. $maxreq specifies the maximum number of outstanding aio |
846 |
requests that this function generates. If it is "<= 0", then a |
847 |
suitable default will be chosen (currently 4). |
848 |
|
849 |
On error, the callback is called without arguments, otherwise it |
850 |
receives two array-refs with path-relative entry names. |
851 |
|
852 |
Example: |
853 |
|
854 |
aio_scandir $dir, 0, sub { |
855 |
my ($dirs, $nondirs) = @_; |
856 |
print "real directories: @$dirs\n"; |
857 |
print "everything else: @$nondirs\n"; |
858 |
}; |
859 |
|
860 |
Implementation notes. |
861 |
|
862 |
The "aio_readdir" cannot be avoided, but stat()'ing every entry can. |
863 |
|
864 |
If readdir returns file type information, then this is used directly |
865 |
to find directories. |
866 |
|
867 |
Otherwise, after reading the directory, the modification time, size |
868 |
etc. of the directory before and after the readdir is checked, and |
869 |
if they match (and isn't the current time), the link count will be |
870 |
used to decide how many entries are directories (if >= 2). |
871 |
Otherwise, no knowledge of the number of subdirectories will be |
872 |
assumed. |
873 |
|
874 |
Then entries will be sorted into likely directories a non-initial |
875 |
dot currently) and likely non-directories (see "aio_readdirx"). Then |
876 |
every entry plus an appended "/." will be "stat"'ed, likely |
877 |
directories first, in order of their inode numbers. If that |
878 |
succeeds, it assumes that the entry is a directory or a symlink to |
879 |
directory (which will be checked separately). This is often faster |
880 |
than stat'ing the entry itself because filesystems might detect the |
881 |
type of the entry without reading the inode data (e.g. ext2fs |
882 |
filetype feature), even on systems that cannot return the filetype |
883 |
information on readdir. |
884 |
|
885 |
If the known number of directories (link count - 2) has been |
886 |
reached, the rest of the entries is assumed to be non-directories. |
887 |
|
888 |
This only works with certainty on POSIX (= UNIX) filesystems, which |
889 |
fortunately are the vast majority of filesystems around. |
890 |
|
891 |
It will also likely work on non-POSIX filesystems with reduced |
892 |
efficiency as those tend to return 0 or 1 as link counts, which |
893 |
disables the directory counting heuristic. |
894 |
|
895 |
aio_rmtree $pathname, $callback->($status) |
896 |
Delete a directory tree starting (and including) $path, return the |
897 |
status of the final "rmdir" only. This is a composite request that |
898 |
uses "aio_scandir" to recurse into and rmdir directories, and unlink |
899 |
everything else. |
900 |
|
901 |
aio_fcntl $fh, $cmd, $arg, $callback->($status) |
902 |
aio_ioctl $fh, $request, $buf, $callback->($status) |
903 |
These work just like the "fcntl" and "ioctl" built-in functions, |
904 |
except they execute asynchronously and pass the return value to the |
905 |
callback. |
906 |
|
907 |
Both calls can be used for a lot of things, some of which make more |
908 |
sense to run asynchronously in their own thread, while some others |
909 |
make less sense. For example, calls that block waiting for external |
910 |
events, such as locking, will also lock down an I/O thread while it |
911 |
is waiting, which can deadlock the whole I/O system. At the same |
912 |
time, there might be no alternative to using a thread to wait. |
913 |
|
914 |
So in general, you should only use these calls for things that do |
915 |
(filesystem) I/O, not for things that wait for other events |
916 |
(network, other processes), although if you are careful and know |
917 |
what you are doing, you still can. |
918 |
|
919 |
The following constants are available and can be used for normal |
920 |
"ioctl" and "fcntl" as well (missing ones are, as usual 0): |
921 |
|
922 |
"F_DUPFD_CLOEXEC", |
923 |
|
924 |
"F_OFD_GETLK", "F_OFD_SETLK", "F_OFD_GETLKW", |
925 |
|
926 |
"FIFREEZE", "FITHAW", "FITRIM", "FICLONE", "FICLONERANGE", |
927 |
"FIDEDUPERANGE". |
928 |
|
929 |
"F_ADD_SEALS", "F_GET_SEALS", "F_SEAL_SEAL", "F_SEAL_SHRINK", |
930 |
"F_SEAL_GROW" and "F_SEAL_WRITE". |
931 |
|
932 |
"FS_IOC_GETFLAGS", "FS_IOC_SETFLAGS", "FS_IOC_GETVERSION", |
933 |
"FS_IOC_SETVERSION", "FS_IOC_FIEMAP". |
934 |
|
935 |
"FS_IOC_FSGETXATTR", "FS_IOC_FSSETXATTR", |
936 |
"FS_IOC_SET_ENCRYPTION_POLICY", "FS_IOC_GET_ENCRYPTION_PWSALT", |
937 |
"FS_IOC_GET_ENCRYPTION_POLICY", "FS_KEY_DESCRIPTOR_SIZE". |
938 |
|
939 |
"FS_SECRM_FL", "FS_UNRM_FL", "FS_COMPR_FL", "FS_SYNC_FL", |
940 |
"FS_IMMUTABLE_FL", "FS_APPEND_FL", "FS_NODUMP_FL", "FS_NOATIME_FL", |
941 |
"FS_DIRTY_FL", "FS_COMPRBLK_FL", "FS_NOCOMP_FL", "FS_ENCRYPT_FL", |
942 |
"FS_BTREE_FL", "FS_INDEX_FL", "FS_JOURNAL_DATA_FL", "FS_NOTAIL_FL", |
943 |
"FS_DIRSYNC_FL", "FS_TOPDIR_FL", "FS_FL_USER_MODIFIABLE". |
944 |
|
945 |
"FS_XFLAG_REALTIME", "FS_XFLAG_PREALLOC", "FS_XFLAG_IMMUTABLE", |
946 |
"FS_XFLAG_APPEND", "FS_XFLAG_SYNC", "FS_XFLAG_NOATIME", |
947 |
"FS_XFLAG_NODUMP", "FS_XFLAG_RTINHERIT", "FS_XFLAG_PROJINHERIT", |
948 |
"FS_XFLAG_NOSYMLINKS", "FS_XFLAG_EXTSIZE", "FS_XFLAG_EXTSZINHERIT", |
949 |
"FS_XFLAG_NODEFRAG", "FS_XFLAG_FILESTREAM", "FS_XFLAG_DAX", |
950 |
"FS_XFLAG_HASATTR", |
951 |
|
952 |
"BLKROSET", "BLKROGET", "BLKRRPART", "BLKGETSIZE", "BLKFLSBUF", |
953 |
"BLKRASET", "BLKRAGET", "BLKFRASET", "BLKFRAGET", "BLKSECTSET", |
954 |
"BLKSECTGET", "BLKSSZGET", "BLKBSZGET", "BLKBSZSET", "BLKGETSIZE64", |
955 |
|
956 |
aio_sync $callback->($status) |
957 |
Asynchronously call sync and call the callback when finished. |
958 |
|
959 |
aio_fsync $fh, $callback->($status) |
960 |
Asynchronously call fsync on the given filehandle and call the |
961 |
callback with the fsync result code. |
962 |
|
963 |
aio_fdatasync $fh, $callback->($status) |
964 |
Asynchronously call fdatasync on the given filehandle and call the |
965 |
callback with the fdatasync result code. |
966 |
|
967 |
If this call isn't available because your OS lacks it or it couldn't |
968 |
be detected, it will be emulated by calling "fsync" instead. |
969 |
|
970 |
aio_syncfs $fh, $callback->($status) |
971 |
Asynchronously call the syncfs syscall to sync the filesystem |
972 |
associated to the given filehandle and call the callback with the |
973 |
syncfs result code. If syncfs is not available, calls sync(), but |
974 |
returns -1 and sets errno to "ENOSYS" nevertheless. |
975 |
|
976 |
aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status) |
977 |
Sync the data portion of the file specified by $offset and $length |
978 |
to disk (but NOT the metadata), by calling the Linux-specific |
979 |
sync_file_range call. If sync_file_range is not available or it |
980 |
returns ENOSYS, then fdatasync or fsync is being substituted. |
981 |
|
982 |
$flags can be a combination of |
983 |
"IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE", |
984 |
"IO::AIO::SYNC_FILE_RANGE_WRITE" and |
985 |
"IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER": refer to the sync_file_range |
986 |
manpage for details. |
987 |
|
988 |
aio_pathsync $pathname, $callback->($status) |
989 |
This request tries to open, fsync and close the given path. This is |
990 |
a composite request intended to sync directories after directory |
991 |
operations (E.g. rename). This might not work on all operating |
992 |
systems or have any specific effect, but usually it makes sure that |
993 |
directory changes get written to disc. It works for anything that |
994 |
can be opened for read-only, not just directories. |
995 |
|
996 |
Future versions of this function might fall back to other methods |
997 |
when "fsync" on the directory fails (such as calling "sync"). |
998 |
|
999 |
Passes 0 when everything went ok, and -1 on error. |
1000 |
|
1001 |
aio_msync $scalar, $offset = 0, $length = undef, flags = MS_SYNC, |
1002 |
$callback->($status) |
1003 |
This is a rather advanced IO::AIO call, which only works on |
1004 |
mmap(2)ed scalars (see the "IO::AIO::mmap" function, although it |
1005 |
also works on data scalars managed by the Sys::Mmap or Mmap modules, |
1006 |
note that the scalar must only be modified in-place while an aio |
1007 |
operation is pending on it). |
1008 |
|
1009 |
It calls the "msync" function of your OS, if available, with the |
1010 |
memory area starting at $offset in the string and ending $length |
1011 |
bytes later. If $length is negative, counts from the end, and if |
1012 |
$length is "undef", then it goes till the end of the string. The |
1013 |
flags can be either "IO::AIO::MS_ASYNC" or "IO::AIO::MS_SYNC", plus |
1014 |
an optional "IO::AIO::MS_INVALIDATE". |
1015 |
|
1016 |
aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, |
1017 |
$callback->($status) |
1018 |
This is a rather advanced IO::AIO call, which works best on |
1019 |
mmap(2)ed scalars. |
1020 |
|
1021 |
It touches (reads or writes) all memory pages in the specified range |
1022 |
inside the scalar. All caveats and parameters are the same as for |
1023 |
"aio_msync", above, except for flags, which must be either 0 (which |
1024 |
reads all pages and ensures they are instantiated) or |
1025 |
"IO::AIO::MT_MODIFY", which modifies the memory pages (by reading |
1026 |
and writing an octet from it, which dirties the page). |
1027 |
|
1028 |
aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status) |
1029 |
This is a rather advanced IO::AIO call, which works best on |
1030 |
mmap(2)ed scalars. |
1031 |
|
1032 |
It reads in all the pages of the underlying storage into memory (if |
1033 |
any) and locks them, so they are not getting swapped/paged out or |
1034 |
removed. |
1035 |
|
1036 |
If $length is undefined, then the scalar will be locked till the |
1037 |
end. |
1038 |
|
1039 |
On systems that do not implement "mlock", this function returns -1 |
1040 |
and sets errno to "ENOSYS". |
1041 |
|
1042 |
Note that the corresponding "munlock" is synchronous and is |
1043 |
documented under "MISCELLANEOUS FUNCTIONS". |
1044 |
|
1045 |
Example: open a file, mmap and mlock it - both will be undone when |
1046 |
$data gets destroyed. |
1047 |
|
1048 |
open my $fh, "<", $path or die "$path: $!"; |
1049 |
my $data; |
1050 |
IO::AIO::mmap $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh; |
1051 |
aio_mlock $data; # mlock in background |
1052 |
|
1053 |
aio_mlockall $flags, $callback->($status) |
1054 |
Calls the "mlockall" function with the given $flags (a combination |
1055 |
of "IO::AIO::MCL_CURRENT", "IO::AIO::MCL_FUTURE" and |
1056 |
"IO::AIO::MCL_ONFAULT"). |
1057 |
|
1058 |
On systems that do not implement "mlockall", this function returns |
1059 |
-1 and sets errno to "ENOSYS". Similarly, flag combinations not |
1060 |
supported by the system result in a return value of -1 with errno |
1061 |
being set to "EINVAL". |
1062 |
|
1063 |
Note that the corresponding "munlockall" is synchronous and is |
1064 |
documented under "MISCELLANEOUS FUNCTIONS". |
1065 |
|
1066 |
Example: asynchronously lock all current and future pages into |
1067 |
memory. |
1068 |
|
1069 |
aio_mlockall IO::AIO::MCL_FUTURE; |
1070 |
|
1071 |
aio_fiemap $fh, $start, $length, $flags, $count, $cb->(\@extents) |
1072 |
Queries the extents of the given file (by calling the Linux "FIEMAP" |
1073 |
ioctl, see <http://cvs.schmorp.de/IO-AIO/doc/fiemap.txt> for |
1074 |
details). If the ioctl is not available on your OS, then this |
1075 |
request will fail with "ENOSYS". |
1076 |
|
1077 |
$start is the starting offset to query extents for, $length is the |
1078 |
size of the range to query - if it is "undef", then the whole file |
1079 |
will be queried. |
1080 |
|
1081 |
$flags is a combination of flags ("IO::AIO::FIEMAP_FLAG_SYNC" or |
1082 |
"IO::AIO::FIEMAP_FLAG_XATTR" - "IO::AIO::FIEMAP_FLAGS_COMPAT" is |
1083 |
also exported), and is normally 0 or "IO::AIO::FIEMAP_FLAG_SYNC" to |
1084 |
query the data portion. |
1085 |
|
1086 |
$count is the maximum number of extent records to return. If it is |
1087 |
"undef", then IO::AIO queries all extents of the range. As a very |
1088 |
special case, if it is 0, then the callback receives the number of |
1089 |
extents instead of the extents themselves (which is unreliable, see |
1090 |
below). |
1091 |
|
1092 |
If an error occurs, the callback receives no arguments. The special |
1093 |
"errno" value "IO::AIO::EBADR" is available to test for flag errors. |
1094 |
|
1095 |
Otherwise, the callback receives an array reference with extent |
1096 |
structures. Each extent structure is an array reference itself, with |
1097 |
the following members: |
1098 |
|
1099 |
[$logical, $physical, $length, $flags] |
1100 |
|
1101 |
Flags is any combination of the following flag values (typically |
1102 |
either 0 or "IO::AIO::FIEMAP_EXTENT_LAST" (1)): |
1103 |
|
1104 |
"IO::AIO::FIEMAP_EXTENT_LAST", "IO::AIO::FIEMAP_EXTENT_UNKNOWN", |
1105 |
"IO::AIO::FIEMAP_EXTENT_DELALLOC", "IO::AIO::FIEMAP_EXTENT_ENCODED", |
1106 |
"IO::AIO::FIEMAP_EXTENT_DATA_ENCRYPTED", |
1107 |
"IO::AIO::FIEMAP_EXTENT_NOT_ALIGNED", |
1108 |
"IO::AIO::FIEMAP_EXTENT_DATA_INLINE", |
1109 |
"IO::AIO::FIEMAP_EXTENT_DATA_TAIL", |
1110 |
"IO::AIO::FIEMAP_EXTENT_UNWRITTEN", "IO::AIO::FIEMAP_EXTENT_MERGED" |
1111 |
or "IO::AIO::FIEMAP_EXTENT_SHARED". |
1112 |
|
1113 |
At the time of this writing (Linux 3.2), this request is unreliable |
1114 |
unless $count is "undef", as the kernel has all sorts of bugs |
1115 |
preventing it to return all extents of a range for files with a |
1116 |
large number of extents. The code (only) works around all these |
1117 |
issues if $count is "undef". |
1118 |
|
1119 |
aio_group $callback->(...) |
1120 |
This is a very special aio request: Instead of doing something, it |
1121 |
is a container for other aio requests, which is useful if you want |
1122 |
to bundle many requests into a single, composite, request with a |
1123 |
definite callback and the ability to cancel the whole request with |
1124 |
its subrequests. |
1125 |
|
1126 |
Returns an object of class IO::AIO::GRP. See its documentation below |
1127 |
for more info. |
1128 |
|
1129 |
Example: |
1130 |
|
1131 |
my $grp = aio_group sub { |
1132 |
print "all stats done\n"; |
1133 |
}; |
1134 |
|
1135 |
add $grp |
1136 |
(aio_stat ...), |
1137 |
(aio_stat ...), |
1138 |
...; |
1139 |
|
1140 |
aio_nop $callback->() |
1141 |
This is a special request - it does nothing in itself and is only |
1142 |
used for side effects, such as when you want to add a dummy request |
1143 |
to a group so that finishing the requests in the group depends on |
1144 |
executing the given code. |
1145 |
|
1146 |
While this request does nothing, it still goes through the execution |
1147 |
phase and still requires a worker thread. Thus, the callback will |
1148 |
not be executed immediately but only after other requests in the |
1149 |
queue have entered their execution phase. This can be used to |
1150 |
measure request latency. |
1151 |
|
1152 |
IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED* |
1153 |
Mainly used for debugging and benchmarking, this aio request puts |
1154 |
one of the request workers to sleep for the given time. |
1155 |
|
1156 |
While it is theoretically handy to have simple I/O scheduling |
1157 |
requests like sleep and file handle readable/writable, the overhead |
1158 |
this creates is immense (it blocks a thread for a long time) so do |
1159 |
not use this function except to put your application under |
1160 |
artificial I/O pressure. |
1161 |
|
1162 |
IO::AIO::WD - multiple working directories |
1163 |
Your process only has one current working directory, which is used by |
1164 |
all threads. This makes it hard to use relative paths (some other |
1165 |
component could call "chdir" at any time, and it is hard to control when |
1166 |
the path will be used by IO::AIO). |
1167 |
|
1168 |
One solution for this is to always use absolute paths. This usually |
1169 |
works, but can be quite slow (the kernel has to walk the whole path on |
1170 |
every access), and can also be a hassle to implement. |
1171 |
|
1172 |
Newer POSIX systems have a number of functions (openat, fdopendir, |
1173 |
futimensat and so on) that make it possible to specify working |
1174 |
directories per operation. |
1175 |
|
1176 |
For portability, and because the clowns who "designed", or shall I |
1177 |
write, perpetrated this new interface were obviously half-drunk, this |
1178 |
abstraction cannot be perfect, though. |
1179 |
|
1180 |
IO::AIO allows you to convert directory paths into a so-called |
1181 |
IO::AIO::WD object. This object stores the canonicalised, absolute |
1182 |
version of the path, and on systems that allow it, also a directory file |
1183 |
descriptor. |
1184 |
|
1185 |
Everywhere where a pathname is accepted by IO::AIO (e.g. in "aio_stat" |
1186 |
or "aio_unlink"), one can specify an array reference with an IO::AIO::WD |
1187 |
object and a pathname instead (or the IO::AIO::WD object alone, which |
1188 |
gets interpreted as "[$wd, "."]"). If the pathname is absolute, the |
1189 |
IO::AIO::WD object is ignored, otherwise the pathname is resolved |
1190 |
relative to that IO::AIO::WD object. |
1191 |
|
1192 |
For example, to get a wd object for /etc and then stat passwd inside, |
1193 |
you would write: |
1194 |
|
1195 |
aio_wd "/etc", sub { |
1196 |
my $etcdir = shift; |
1197 |
|
1198 |
# although $etcdir can be undef on error, there is generally no reason |
1199 |
# to check for errors here, as aio_stat will fail with ENOENT |
1200 |
# when $etcdir is undef. |
1201 |
|
1202 |
aio_stat [$etcdir, "passwd"], sub { |
1203 |
# yay |
1204 |
}; |
1205 |
}; |
1206 |
|
1207 |
The fact that "aio_wd" is a request and not a normal function shows that |
1208 |
creating an IO::AIO::WD object is itself a potentially blocking |
1209 |
operation, which is why it is done asynchronously. |
1210 |
|
1211 |
To stat the directory obtained with "aio_wd" above, one could write |
1212 |
either of the following three request calls: |
1213 |
|
1214 |
aio_lstat "/etc" , sub { ... # pathname as normal string |
1215 |
aio_lstat [$wd, "."], sub { ... # "." relative to $wd (i.e. $wd itself) |
1216 |
aio_lstat $wd , sub { ... # shorthand for the previous |
1217 |
|
1218 |
As with normal pathnames, IO::AIO keeps a copy of the working directory |
1219 |
object and the pathname string, so you could write the following without |
1220 |
causing any issues due to $path getting reused: |
1221 |
|
1222 |
my $path = [$wd, undef]; |
1223 |
|
1224 |
for my $name (qw(abc def ghi)) { |
1225 |
$path->[1] = $name; |
1226 |
aio_stat $path, sub { |
1227 |
# ... |
1228 |
}; |
1229 |
} |
1230 |
|
1231 |
There are some caveats: when directories get renamed (or deleted), the |
1232 |
pathname string doesn't change, so will point to the new directory (or |
1233 |
nowhere at all), while the directory fd, if available on the system, |
1234 |
will still point to the original directory. Most functions accepting a |
1235 |
pathname will use the directory fd on newer systems, and the string on |
1236 |
older systems. Some functions (such as "aio_realpath") will always rely |
1237 |
on the string form of the pathname. |
1238 |
|
1239 |
So this functionality is mainly useful to get some protection against |
1240 |
"chdir", to easily get an absolute path out of a relative path for |
1241 |
future reference, and to speed up doing many operations in the same |
1242 |
directory (e.g. when stat'ing all files in a directory). |
1243 |
|
1244 |
The following functions implement this working directory abstraction: |
1245 |
|
1246 |
aio_wd $pathname, $callback->($wd) |
1247 |
Asynchonously canonicalise the given pathname and convert it to an |
1248 |
IO::AIO::WD object representing it. If possible and supported on the |
1249 |
system, also open a directory fd to speed up pathname resolution |
1250 |
relative to this working directory. |
1251 |
|
1252 |
If something goes wrong, then "undef" is passwd to the callback |
1253 |
instead of a working directory object and $! is set appropriately. |
1254 |
Since passing "undef" as working directory component of a pathname |
1255 |
fails the request with "ENOENT", there is often no need for error |
1256 |
checking in the "aio_wd" callback, as future requests using the |
1257 |
value will fail in the expected way. |
1258 |
|
1259 |
IO::AIO::CWD |
1260 |
This is a compile time constant (object) that represents the process |
1261 |
current working directory. |
1262 |
|
1263 |
Specifying this object as working directory object for a pathname is |
1264 |
as if the pathname would be specified directly, without a directory |
1265 |
object. For example, these calls are functionally identical: |
1266 |
|
1267 |
aio_stat "somefile", sub { ... }; |
1268 |
aio_stat [IO::AIO::CWD, "somefile"], sub { ... }; |
1269 |
|
1270 |
To recover the path associated with an IO::AIO::WD object, you can use |
1271 |
"aio_realpath": |
1272 |
|
1273 |
aio_realpath $wd, sub { |
1274 |
warn "path is $_[0]\n"; |
1275 |
}; |
1276 |
|
1277 |
Currently, "aio_statvfs" always, and "aio_rename" and "aio_rmdir" |
1278 |
sometimes, fall back to using an absolue path. |
1279 |
|
1280 |
IO::AIO::REQ CLASS |
1281 |
All non-aggregate "aio_*" functions return an object of this class when |
1282 |
called in non-void context. |
1283 |
|
1284 |
cancel $req |
1285 |
Cancels the request, if possible. Has the effect of skipping |
1286 |
execution when entering the execute state and skipping calling the |
1287 |
callback when entering the the result state, but will leave the |
1288 |
request otherwise untouched (with the exception of readdir). That |
1289 |
means that requests that currently execute will not be stopped and |
1290 |
resources held by the request will not be freed prematurely. |
1291 |
|
1292 |
cb $req $callback->(...) |
1293 |
Replace (or simply set) the callback registered to the request. |
1294 |
|
1295 |
IO::AIO::GRP CLASS |
1296 |
This class is a subclass of IO::AIO::REQ, so all its methods apply to |
1297 |
objects of this class, too. |
1298 |
|
1299 |
A IO::AIO::GRP object is a special request that can contain multiple |
1300 |
other aio requests. |
1301 |
|
1302 |
You create one by calling the "aio_group" constructing function with a |
1303 |
callback that will be called when all contained requests have entered |
1304 |
the "done" state: |
1305 |
|
1306 |
my $grp = aio_group sub { |
1307 |
print "all requests are done\n"; |
1308 |
}; |
1309 |
|
1310 |
You add requests by calling the "add" method with one or more |
1311 |
"IO::AIO::REQ" objects: |
1312 |
|
1313 |
$grp->add (aio_unlink "..."); |
1314 |
|
1315 |
add $grp aio_stat "...", sub { |
1316 |
$_[0] or return $grp->result ("error"); |
1317 |
|
1318 |
# add another request dynamically, if first succeeded |
1319 |
add $grp aio_open "...", sub { |
1320 |
$grp->result ("ok"); |
1321 |
}; |
1322 |
}; |
1323 |
|
1324 |
This makes it very easy to create composite requests (see the source of |
1325 |
"aio_move" for an application) that work and feel like simple requests. |
1326 |
|
1327 |
* The IO::AIO::GRP objects will be cleaned up during calls to |
1328 |
"IO::AIO::poll_cb", just like any other request. |
1329 |
|
1330 |
* They can be canceled like any other request. Canceling will cancel |
1331 |
not only the request itself, but also all requests it contains. |
1332 |
|
1333 |
* They can also can also be added to other IO::AIO::GRP objects. |
1334 |
|
1335 |
* You must not add requests to a group from within the group callback |
1336 |
(or any later time). |
1337 |
|
1338 |
Their lifetime, simplified, looks like this: when they are empty, they |
1339 |
will finish very quickly. If they contain only requests that are in the |
1340 |
"done" state, they will also finish. Otherwise they will continue to |
1341 |
exist. |
1342 |
|
1343 |
That means after creating a group you have some time to add requests |
1344 |
(precisely before the callback has been invoked, which is only done |
1345 |
within the "poll_cb"). And in the callbacks of those requests, you can |
1346 |
add further requests to the group. And only when all those requests have |
1347 |
finished will the the group itself finish. |
1348 |
|
1349 |
add $grp ... |
1350 |
$grp->add (...) |
1351 |
Add one or more requests to the group. Any type of IO::AIO::REQ can |
1352 |
be added, including other groups, as long as you do not create |
1353 |
circular dependencies. |
1354 |
|
1355 |
Returns all its arguments. |
1356 |
|
1357 |
$grp->cancel_subs |
1358 |
Cancel all subrequests and clears any feeder, but not the group |
1359 |
request itself. Useful when you queued a lot of events but got a |
1360 |
result early. |
1361 |
|
1362 |
The group request will finish normally (you cannot add requests to |
1363 |
the group). |
1364 |
|
1365 |
$grp->result (...) |
1366 |
Set the result value(s) that will be passed to the group callback |
1367 |
when all subrequests have finished and set the groups errno to the |
1368 |
current value of errno (just like calling "errno" without an error |
1369 |
number). By default, no argument will be passed and errno is zero. |
1370 |
|
1371 |
$grp->errno ([$errno]) |
1372 |
Sets the group errno value to $errno, or the current value of errno |
1373 |
when the argument is missing. |
1374 |
|
1375 |
Every aio request has an associated errno value that is restored |
1376 |
when the callback is invoked. This method lets you change this value |
1377 |
from its default (0). |
1378 |
|
1379 |
Calling "result" will also set errno, so make sure you either set $! |
1380 |
before the call to "result", or call c<errno> after it. |
1381 |
|
1382 |
feed $grp $callback->($grp) |
1383 |
Sets a feeder/generator on this group: every group can have an |
1384 |
attached generator that generates requests if idle. The idea behind |
1385 |
this is that, although you could just queue as many requests as you |
1386 |
want in a group, this might starve other requests for a potentially |
1387 |
long time. For example, "aio_scandir" might generate hundreds of |
1388 |
thousands of "aio_stat" requests, delaying any later requests for a |
1389 |
long time. |
1390 |
|
1391 |
To avoid this, and allow incremental generation of requests, you can |
1392 |
instead a group and set a feeder on it that generates those |
1393 |
requests. The feed callback will be called whenever there are few |
1394 |
enough (see "limit", below) requests active in the group itself and |
1395 |
is expected to queue more requests. |
1396 |
|
1397 |
The feed callback can queue as many requests as it likes (i.e. "add" |
1398 |
does not impose any limits). |
1399 |
|
1400 |
If the feed does not queue more requests when called, it will be |
1401 |
automatically removed from the group. |
1402 |
|
1403 |
If the feed limit is 0 when this method is called, it will be set to |
1404 |
2 automatically. |
1405 |
|
1406 |
Example: |
1407 |
|
1408 |
# stat all files in @files, but only ever use four aio requests concurrently: |
1409 |
|
1410 |
my $grp = aio_group sub { print "finished\n" }; |
1411 |
limit $grp 4; |
1412 |
feed $grp sub { |
1413 |
my $file = pop @files |
1414 |
or return; |
1415 |
|
1416 |
add $grp aio_stat $file, sub { ... }; |
1417 |
}; |
1418 |
|
1419 |
limit $grp $num |
1420 |
Sets the feeder limit for the group: The feeder will be called |
1421 |
whenever the group contains less than this many requests. |
1422 |
|
1423 |
Setting the limit to 0 will pause the feeding process. |
1424 |
|
1425 |
The default value for the limit is 0, but note that setting a feeder |
1426 |
automatically bumps it up to 2. |
1427 |
|
1428 |
SUPPORT FUNCTIONS |
1429 |
EVENT PROCESSING AND EVENT LOOP INTEGRATION |
1430 |
$fileno = IO::AIO::poll_fileno |
1431 |
Return the *request result pipe file descriptor*. This filehandle |
1432 |
must be polled for reading by some mechanism outside this module |
1433 |
(e.g. EV, Glib, select and so on, see below or the SYNOPSIS). If the |
1434 |
pipe becomes readable you have to call "poll_cb" to check the |
1435 |
results. |
1436 |
|
1437 |
See "poll_cb" for an example. |
1438 |
|
1439 |
IO::AIO::poll_cb |
1440 |
Process some requests that have reached the result phase (i.e. they |
1441 |
have been executed but the results are not yet reported). You have |
1442 |
to call this "regularly" to finish outstanding requests. |
1443 |
|
1444 |
Returns 0 if all events could be processed (or there were no events |
1445 |
to process), or -1 if it returned earlier for whatever reason. |
1446 |
Returns immediately when no events are outstanding. The amount of |
1447 |
events processed depends on the settings of "IO::AIO::max_poll_req", |
1448 |
"IO::AIO::max_poll_time" and "IO::AIO::max_outstanding". |
1449 |
|
1450 |
If not all requests were processed for whatever reason, the poll |
1451 |
file descriptor will still be ready when "poll_cb" returns, so |
1452 |
normally you don't have to do anything special to have it called |
1453 |
later. |
1454 |
|
1455 |
Apart from calling "IO::AIO::poll_cb" when the event filehandle |
1456 |
becomes ready, it can be beneficial to call this function from loops |
1457 |
which submit a lot of requests, to make sure the results get |
1458 |
processed when they become available and not just when the loop is |
1459 |
finished and the event loop takes over again. This function returns |
1460 |
very fast when there are no outstanding requests. |
1461 |
|
1462 |
Example: Install an Event watcher that automatically calls |
1463 |
IO::AIO::poll_cb with high priority (more examples can be found in |
1464 |
the SYNOPSIS section, at the top of this document): |
1465 |
|
1466 |
Event->io (fd => IO::AIO::poll_fileno, |
1467 |
poll => 'r', async => 1, |
1468 |
cb => \&IO::AIO::poll_cb); |
1469 |
|
1470 |
IO::AIO::poll_wait |
1471 |
Wait until either at least one request is in the result phase or no |
1472 |
requests are outstanding anymore. |
1473 |
|
1474 |
This is useful if you want to synchronously wait for some requests |
1475 |
to become ready, without actually handling them. |
1476 |
|
1477 |
See "nreqs" for an example. |
1478 |
|
1479 |
IO::AIO::poll |
1480 |
Waits until some requests have been handled. |
1481 |
|
1482 |
Returns the number of requests processed, but is otherwise strictly |
1483 |
equivalent to: |
1484 |
|
1485 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
1486 |
|
1487 |
IO::AIO::flush |
1488 |
Wait till all outstanding AIO requests have been handled. |
1489 |
|
1490 |
Strictly equivalent to: |
1491 |
|
1492 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
1493 |
while IO::AIO::nreqs; |
1494 |
|
1495 |
This function can be useful at program aborts, to make sure |
1496 |
outstanding I/O has been done ("IO::AIO" uses an "END" block which |
1497 |
already calls this function on normal exits), or when you are merely |
1498 |
using "IO::AIO" for its more advanced functions, rather than for |
1499 |
async I/O, e.g.: |
1500 |
|
1501 |
my ($dirs, $nondirs); |
1502 |
IO::AIO::aio_scandir "/tmp", 0, sub { ($dirs, $nondirs) = @_ }; |
1503 |
IO::AIO::flush; |
1504 |
# $dirs, $nondirs are now set |
1505 |
|
1506 |
IO::AIO::max_poll_reqs $nreqs |
1507 |
IO::AIO::max_poll_time $seconds |
1508 |
These set the maximum number of requests (default 0, meaning |
1509 |
infinity) that are being processed by "IO::AIO::poll_cb" in one |
1510 |
call, respectively the maximum amount of time (default 0, meaning |
1511 |
infinity) spent in "IO::AIO::poll_cb" to process requests (more |
1512 |
correctly the mininum amount of time "poll_cb" is allowed to use). |
1513 |
|
1514 |
Setting "max_poll_time" to a non-zero value creates an overhead of |
1515 |
one syscall per request processed, which is not normally a problem |
1516 |
unless your callbacks are really really fast or your OS is really |
1517 |
really slow (I am not mentioning Solaris here). Using |
1518 |
"max_poll_reqs" incurs no overhead. |
1519 |
|
1520 |
Setting these is useful if you want to ensure some level of |
1521 |
interactiveness when perl is not fast enough to process all requests |
1522 |
in time. |
1523 |
|
1524 |
For interactive programs, values such as 0.01 to 0.1 should be fine. |
1525 |
|
1526 |
Example: Install an Event watcher that automatically calls |
1527 |
IO::AIO::poll_cb with low priority, to ensure that other parts of |
1528 |
the program get the CPU sometimes even under high AIO load. |
1529 |
|
1530 |
# try not to spend much more than 0.1s in poll_cb |
1531 |
IO::AIO::max_poll_time 0.1; |
1532 |
|
1533 |
# use a low priority so other tasks have priority |
1534 |
Event->io (fd => IO::AIO::poll_fileno, |
1535 |
poll => 'r', nice => 1, |
1536 |
cb => &IO::AIO::poll_cb); |
1537 |
|
1538 |
CONTROLLING THE NUMBER OF THREADS |
1539 |
IO::AIO::min_parallel $nthreads |
1540 |
Set the minimum number of AIO threads to $nthreads. The current |
1541 |
default is 8, which means eight asynchronous operations can execute |
1542 |
concurrently at any one time (the number of outstanding requests, |
1543 |
however, is unlimited). |
1544 |
|
1545 |
IO::AIO starts threads only on demand, when an AIO request is queued |
1546 |
and no free thread exists. Please note that queueing up a hundred |
1547 |
requests can create demand for a hundred threads, even if it turns |
1548 |
out that everything is in the cache and could have been processed |
1549 |
faster by a single thread. |
1550 |
|
1551 |
It is recommended to keep the number of threads relatively low, as |
1552 |
some Linux kernel versions will scale negatively with the number of |
1553 |
threads (higher parallelity => MUCH higher latency). With current |
1554 |
Linux 2.6 versions, 4-32 threads should be fine. |
1555 |
|
1556 |
Under most circumstances you don't need to call this function, as |
1557 |
the module selects a default that is suitable for low to moderate |
1558 |
load. |
1559 |
|
1560 |
IO::AIO::max_parallel $nthreads |
1561 |
Sets the maximum number of AIO threads to $nthreads. If more than |
1562 |
the specified number of threads are currently running, this function |
1563 |
kills them. This function blocks until the limit is reached. |
1564 |
|
1565 |
While $nthreads are zero, aio requests get queued but not executed |
1566 |
until the number of threads has been increased again. |
1567 |
|
1568 |
This module automatically runs "max_parallel 0" at program end, to |
1569 |
ensure that all threads are killed and that there are no outstanding |
1570 |
requests. |
1571 |
|
1572 |
Under normal circumstances you don't need to call this function. |
1573 |
|
1574 |
IO::AIO::max_idle $nthreads |
1575 |
Limit the number of threads (default: 4) that are allowed to idle |
1576 |
(i.e., threads that did not get a request to process within the idle |
1577 |
timeout (default: 10 seconds). That means if a thread becomes idle |
1578 |
while $nthreads other threads are also idle, it will free its |
1579 |
resources and exit. |
1580 |
|
1581 |
This is useful when you allow a large number of threads (e.g. 100 or |
1582 |
1000) to allow for extremely high load situations, but want to free |
1583 |
resources under normal circumstances (1000 threads can easily |
1584 |
consume 30MB of RAM). |
1585 |
|
1586 |
The default is probably ok in most situations, especially if thread |
1587 |
creation is fast. If thread creation is very slow on your system you |
1588 |
might want to use larger values. |
1589 |
|
1590 |
IO::AIO::idle_timeout $seconds |
1591 |
Sets the minimum idle timeout (default 10) after which worker |
1592 |
threads are allowed to exit. SEe "IO::AIO::max_idle". |
1593 |
|
1594 |
IO::AIO::max_outstanding $maxreqs |
1595 |
Sets the maximum number of outstanding requests to $nreqs. If you do |
1596 |
queue up more than this number of requests, the next call to |
1597 |
"IO::AIO::poll_cb" (and other functions calling "poll_cb", such as |
1598 |
"IO::AIO::flush" or "IO::AIO::poll") will block until the limit is |
1599 |
no longer exceeded. |
1600 |
|
1601 |
In other words, this setting does not enforce a queue limit, but can |
1602 |
be used to make poll functions block if the limit is exceeded. |
1603 |
|
1604 |
This is a bad function to use in interactive programs because it |
1605 |
blocks, and a bad way to reduce concurrency because it is inexact. |
1606 |
If you need to issue many requests without being able to call a poll |
1607 |
function on demand, it is better to use an "aio_group" together with |
1608 |
a feed callback. |
1609 |
|
1610 |
Its main use is in scripts without an event loop - when you want to |
1611 |
stat a lot of files, you can write something like this: |
1612 |
|
1613 |
IO::AIO::max_outstanding 32; |
1614 |
|
1615 |
for my $path (...) { |
1616 |
aio_stat $path , ...; |
1617 |
IO::AIO::poll_cb; |
1618 |
} |
1619 |
|
1620 |
IO::AIO::flush; |
1621 |
|
1622 |
The call to "poll_cb" inside the loop will normally return |
1623 |
instantly, allowing the loop to progress, but as soon as more than |
1624 |
32 requests are in-flight, it will block until some requests have |
1625 |
been handled. This keeps the loop from pushing a large number of |
1626 |
"aio_stat" requests onto the queue (which, with many paths to stat, |
1627 |
can use up a lot of memory). |
1628 |
|
1629 |
The default value for "max_outstanding" is very large, so there is |
1630 |
no practical limit on the number of outstanding requests. |
1631 |
|
1632 |
STATISTICAL INFORMATION |
1633 |
IO::AIO::nreqs |
1634 |
Returns the number of requests currently in the ready, execute or |
1635 |
pending states (i.e. for which their callback has not been invoked |
1636 |
yet). |
1637 |
|
1638 |
Example: wait till there are no outstanding requests anymore: |
1639 |
|
1640 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
1641 |
while IO::AIO::nreqs; |
1642 |
|
1643 |
IO::AIO::nready |
1644 |
Returns the number of requests currently in the ready state (not yet |
1645 |
executed). |
1646 |
|
1647 |
IO::AIO::npending |
1648 |
Returns the number of requests currently in the pending state |
1649 |
(executed, but not yet processed by poll_cb). |
1650 |
|
1651 |
SUBSECOND STAT TIME ACCESS |
1652 |
Both "aio_stat"/"aio_lstat" and perl's "stat"/"lstat" functions can |
1653 |
generally find access/modification and change times with subsecond time |
1654 |
accuracy of the system supports it, but perl's built-in functions only |
1655 |
return the integer part. |
1656 |
|
1657 |
The following functions return the timestamps of the most recent stat |
1658 |
with subsecond precision on most systems and work both after |
1659 |
"aio_stat"/"aio_lstat" and perl's "stat"/"lstat" calls. Their return |
1660 |
value is only meaningful after a successful "stat"/"lstat" call, or |
1661 |
during/after a successful "aio_stat"/"aio_lstat" callback. |
1662 |
|
1663 |
This is similar to the Time::HiRes "stat" functions, but can return full |
1664 |
resolution without rounding and work with standard perl "stat", |
1665 |
alleviating the need to call the special "Time::HiRes" functions, which |
1666 |
do not act like their perl counterparts. |
1667 |
|
1668 |
On operating systems or file systems where subsecond time resolution is |
1669 |
not supported or could not be detected, a fractional part of 0 is |
1670 |
returned, so it is always safe to call these functions. |
1671 |
|
1672 |
$seconds = IO::AIO::st_atime, IO::AIO::st_mtime, IO::AIO::st_ctime, |
1673 |
IO::AIO::st_btime |
1674 |
Return the access, modication, change or birth time, respectively, |
1675 |
including fractional part. Due to the limited precision of floating |
1676 |
point, the accuracy on most platforms is only a bit better than |
1677 |
milliseconds for times around now - see the *nsec* function family, |
1678 |
below, for full accuracy. |
1679 |
|
1680 |
File birth time is only available when the OS and perl support it |
1681 |
(on FreeBSD and NetBSD at the time of this writing, although support |
1682 |
is adaptive, so if your OS/perl gains support, IO::AIO can take |
1683 |
advantage of it). On systems where it isn't available, 0 is |
1684 |
currently returned, but this might change to "undef" in a future |
1685 |
version. |
1686 |
|
1687 |
($atime, $mtime, $ctime, $btime, ...) = IO::AIO::st_xtime |
1688 |
Returns access, modification, change and birth time all in one go, |
1689 |
and maybe more times in the future version. |
1690 |
|
1691 |
$nanoseconds = IO::AIO::st_atimensec, IO::AIO::st_mtimensec, |
1692 |
IO::AIO::st_ctimensec, IO::AIO::st_btimensec |
1693 |
Return the fractional access, modifcation, change or birth time, in |
1694 |
nanoseconds, as an integer in the range 0 to 999999999. |
1695 |
|
1696 |
Note that no accessors are provided for access, modification and |
1697 |
change times - you need to get those from "stat _" if required ("int |
1698 |
IO::AIO::st_atime" and so on will *not* generally give you the |
1699 |
correct value). |
1700 |
|
1701 |
$seconds = IO::AIO::st_btimesec |
1702 |
The (integral) seconds part of the file birth time, if available. |
1703 |
|
1704 |
($atime, $mtime, $ctime, $btime, ...) = IO::AIO::st_xtimensec |
1705 |
Like the functions above, but returns all four times in one go (and |
1706 |
maybe more in future versions). |
1707 |
|
1708 |
$counter = IO::AIO::st_gen |
1709 |
Returns the generation counter (in practice this is just a random |
1710 |
number) of the file. This is only available on platforms which have |
1711 |
this member in their "struct stat" (most BSDs at the time of this |
1712 |
writing) and generally only to the root usert. If unsupported, 0 is |
1713 |
returned, but this might change to "undef" in a future version. |
1714 |
|
1715 |
Example: print the high resolution modification time of /etc, using |
1716 |
"stat", and "IO::AIO::aio_stat". |
1717 |
|
1718 |
if (stat "/etc") { |
1719 |
printf "stat(/etc) mtime: %f\n", IO::AIO::st_mtime; |
1720 |
} |
1721 |
|
1722 |
IO::AIO::aio_stat "/etc", sub { |
1723 |
$_[0] |
1724 |
and return; |
1725 |
|
1726 |
printf "aio_stat(/etc) mtime: %d.%09d\n", (stat _)[9], IO::AIO::st_mtimensec; |
1727 |
}; |
1728 |
|
1729 |
IO::AIO::flush; |
1730 |
|
1731 |
Output of the awbove on my system, showing reduced and full accuracy: |
1732 |
|
1733 |
stat(/etc) mtime: 1534043702.020808 |
1734 |
aio_stat(/etc) mtime: 1534043702.020807792 |
1735 |
|
1736 |
MISCELLANEOUS FUNCTIONS |
1737 |
IO::AIO implements some functions that are useful when you want to use |
1738 |
some "Advanced I/O" function not available to in Perl, without going the |
1739 |
"Asynchronous I/O" route. Many of these have an asynchronous "aio_*" |
1740 |
counterpart. |
1741 |
|
1742 |
$retval = IO::AIO::fexecve $fh, $argv, $envp |
1743 |
A more-or-less direct equivalent to the POSIX "fexecve" functions, |
1744 |
which allows you to specify the program to be executed via a file |
1745 |
descriptor (or handle). Returns -1 and sets errno to "ENOSYS" if not |
1746 |
available. |
1747 |
|
1748 |
$retval = IO::AIO::mount $special, $path, $fstype, $flags = 0, $data = |
1749 |
undef |
1750 |
Calls the GNU/Linux mount syscall with the given arguments. All |
1751 |
except $flags are strings, and if $data is "undef", a "NULL" will be |
1752 |
passed. |
1753 |
|
1754 |
The following values for $flags are available: |
1755 |
|
1756 |
"IO::AIO::MS_RDONLY", "IO::AIO::MS_NOSUID", "IO::AIO::MS_NODEV", |
1757 |
"IO::AIO::MS_NOEXEC", "IO::AIO::MS_SYNCHRONOUS", |
1758 |
"IO::AIO::MS_REMOUNT", "IO::AIO::MS_MANDLOCK", |
1759 |
"IO::AIO::MS_DIRSYNC", "IO::AIO::MS_NOATIME", |
1760 |
"IO::AIO::MS_NODIRATIME", "IO::AIO::MS_BIND", "IO::AIO::MS_MOVE", |
1761 |
"IO::AIO::MS_REC", "IO::AIO::MS_SILENT", "IO::AIO::MS_POSIXACL", |
1762 |
"IO::AIO::MS_UNBINDABLE", "IO::AIO::MS_PRIVATE", |
1763 |
"IO::AIO::MS_SLAVE", "IO::AIO::MS_SHARED", "IO::AIO::MS_RELATIME", |
1764 |
"IO::AIO::MS_KERNMOUNT", "IO::AIO::MS_I_VERSION", |
1765 |
"IO::AIO::MS_STRICTATIME", "IO::AIO::MS_LAZYTIME", |
1766 |
"IO::AIO::MS_ACTIVE", "IO::AIO::MS_NOUSER", "IO::AIO::MS_RMT_MASK", |
1767 |
"IO::AIO::MS_MGC_VAL" and "IO::AIO::MS_MGC_MSK". |
1768 |
|
1769 |
$retval = IO::AIO::umount $path, $flags = 0 |
1770 |
Invokes the GNU/Linux "umount" or "umount2" syscalls. Always calls |
1771 |
"umount" if $flags is 0, otherwqise always tries to call "umount2". |
1772 |
|
1773 |
The following $flags are available: |
1774 |
|
1775 |
"IO::AIO::MNT_FORCE", "IO::AIO::MNT_DETACH", "IO::AIO::MNT_EXPIRE" |
1776 |
and "IO::AIO::UMOUNT_NOFOLLOW". |
1777 |
|
1778 |
$numfd = IO::AIO::get_fdlimit |
1779 |
Tries to find the current file descriptor limit and returns it, or |
1780 |
"undef" and sets $! in case of an error. The limit is one larger |
1781 |
than the highest valid file descriptor number. |
1782 |
|
1783 |
IO::AIO::min_fdlimit [$numfd] |
1784 |
Try to increase the current file descriptor limit(s) to at least |
1785 |
$numfd by changing the soft or hard file descriptor resource limit. |
1786 |
If $numfd is missing, it will try to set a very high limit, although |
1787 |
this is not recommended when you know the actual minimum that you |
1788 |
require. |
1789 |
|
1790 |
If the limit cannot be raised enough, the function makes a |
1791 |
best-effort attempt to increase the limit as much as possible, using |
1792 |
various tricks, while still failing. You can query the resulting |
1793 |
limit using "IO::AIO::get_fdlimit". |
1794 |
|
1795 |
If an error occurs, returns "undef" and sets $!, otherwise returns |
1796 |
true. |
1797 |
|
1798 |
IO::AIO::sendfile $ofh, $ifh, $offset, $count |
1799 |
Calls the "eio_sendfile_sync" function, which is like |
1800 |
"aio_sendfile", but is blocking (this makes most sense if you know |
1801 |
the input data is likely cached already and the output filehandle is |
1802 |
set to non-blocking operations). |
1803 |
|
1804 |
Returns the number of bytes copied, or -1 on error. |
1805 |
|
1806 |
IO::AIO::fadvise $fh, $offset, $len, $advice |
1807 |
Simply calls the "posix_fadvise" function (see its manpage for |
1808 |
details). The following advice constants are available: |
1809 |
"IO::AIO::FADV_NORMAL", "IO::AIO::FADV_SEQUENTIAL", |
1810 |
"IO::AIO::FADV_RANDOM", "IO::AIO::FADV_NOREUSE", |
1811 |
"IO::AIO::FADV_WILLNEED", "IO::AIO::FADV_DONTNEED". |
1812 |
|
1813 |
On systems that do not implement "posix_fadvise", this function |
1814 |
returns ENOSYS, otherwise the return value of "posix_fadvise". |
1815 |
|
1816 |
IO::AIO::madvise $scalar, $offset, $len, $advice |
1817 |
Simply calls the "posix_madvise" function (see its manpage for |
1818 |
details). The following advice constants are available: |
1819 |
"IO::AIO::MADV_NORMAL", "IO::AIO::MADV_SEQUENTIAL", |
1820 |
"IO::AIO::MADV_RANDOM", "IO::AIO::MADV_WILLNEED", |
1821 |
"IO::AIO::MADV_DONTNEED". |
1822 |
|
1823 |
If $offset is negative, counts from the end. If $length is negative, |
1824 |
the remaining length of the $scalar is used. If possible, $length |
1825 |
will be reduced to fit into the $scalar. |
1826 |
|
1827 |
On systems that do not implement "posix_madvise", this function |
1828 |
returns ENOSYS, otherwise the return value of "posix_madvise". |
1829 |
|
1830 |
IO::AIO::mprotect $scalar, $offset, $len, $protect |
1831 |
Simply calls the "mprotect" function on the preferably AIO::mmap'ed |
1832 |
$scalar (see its manpage for details). The following protect |
1833 |
constants are available: "IO::AIO::PROT_NONE", "IO::AIO::PROT_READ", |
1834 |
"IO::AIO::PROT_WRITE", "IO::AIO::PROT_EXEC". |
1835 |
|
1836 |
If $offset is negative, counts from the end. If $length is negative, |
1837 |
the remaining length of the $scalar is used. If possible, $length |
1838 |
will be reduced to fit into the $scalar. |
1839 |
|
1840 |
On systems that do not implement "mprotect", this function returns |
1841 |
ENOSYS, otherwise the return value of "mprotect". |
1842 |
|
1843 |
IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset] |
1844 |
Memory-maps a file (or anonymous memory range) and attaches it to |
1845 |
the given $scalar, which will act like a string scalar. Returns true |
1846 |
on success, and false otherwise. |
1847 |
|
1848 |
The scalar must exist, but its contents do not matter - this means |
1849 |
you cannot use a nonexistant array or hash element. When in doubt, |
1850 |
"undef" the scalar first. |
1851 |
|
1852 |
The only operations allowed on the mmapped scalar are |
1853 |
"substr"/"vec", which don't change the string length, and most |
1854 |
read-only operations such as copying it or searching it with regexes |
1855 |
and so on. |
1856 |
|
1857 |
Anything else is unsafe and will, at best, result in memory leaks. |
1858 |
|
1859 |
The memory map associated with the $scalar is automatically removed |
1860 |
when the $scalar is undef'd or destroyed, or when the |
1861 |
"IO::AIO::mmap" or "IO::AIO::munmap" functions are called on it. |
1862 |
|
1863 |
This calls the "mmap"(2) function internally. See your system's |
1864 |
manual page for details on the $length, $prot and $flags parameters. |
1865 |
|
1866 |
The $length must be larger than zero and smaller than the actual |
1867 |
filesize. |
1868 |
|
1869 |
$prot is a combination of "IO::AIO::PROT_NONE", |
1870 |
"IO::AIO::PROT_EXEC", "IO::AIO::PROT_READ" and/or |
1871 |
"IO::AIO::PROT_WRITE", |
1872 |
|
1873 |
$flags can be a combination of "IO::AIO::MAP_SHARED" or |
1874 |
"IO::AIO::MAP_PRIVATE", or a number of system-specific flags (when |
1875 |
not available, the are 0): "IO::AIO::MAP_ANONYMOUS" (which is set to |
1876 |
"MAP_ANON" if your system only provides this constant), |
1877 |
"IO::AIO::MAP_LOCKED", "IO::AIO::MAP_NORESERVE", |
1878 |
"IO::AIO::MAP_POPULATE", "IO::AIO::MAP_NONBLOCK", |
1879 |
"IO::AIO::MAP_FIXED", "IO::AIO::MAP_GROWSDOWN", |
1880 |
"IO::AIO::MAP_32BIT", "IO::AIO::MAP_HUGETLB", "IO::AIO::MAP_STACK", |
1881 |
"IO::AIO::MAP_FIXED_NOREPLACE", "IO::AIO::MAP_SHARED_VALIDATE", |
1882 |
"IO::AIO::MAP_SYNC" or "IO::AIO::MAP_UNINITIALIZED". |
1883 |
|
1884 |
If $fh is "undef", then a file descriptor of -1 is passed. |
1885 |
|
1886 |
$offset is the offset from the start of the file - it generally must |
1887 |
be a multiple of "IO::AIO::PAGESIZE" and defaults to 0. |
1888 |
|
1889 |
Example: |
1890 |
|
1891 |
use Digest::MD5; |
1892 |
use IO::AIO; |
1893 |
|
1894 |
open my $fh, "<verybigfile" |
1895 |
or die "$!"; |
1896 |
|
1897 |
IO::AIO::mmap my $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh |
1898 |
or die "verybigfile: $!"; |
1899 |
|
1900 |
my $fast_md5 = md5 $data; |
1901 |
|
1902 |
IO::AIO::munmap $scalar |
1903 |
Removes a previous mmap and undefines the $scalar. |
1904 |
|
1905 |
IO::AIO::mremap $scalar, $new_length, $flags = MREMAP_MAYMOVE[, |
1906 |
$new_address = 0] |
1907 |
Calls the Linux-specific mremap(2) system call. The $scalar must |
1908 |
have been mapped by "IO::AIO::mmap", and $flags must currently |
1909 |
either be 0 or "IO::AIO::MREMAP_MAYMOVE". |
1910 |
|
1911 |
Returns true if successful, and false otherwise. If the underlying |
1912 |
mmapped region has changed address, then the true value has the |
1913 |
numerical value 1, otherwise it has the numerical value 0: |
1914 |
|
1915 |
my $success = IO::AIO::mremap $mmapped, 8192, IO::AIO::MREMAP_MAYMOVE |
1916 |
or die "mremap: $!"; |
1917 |
|
1918 |
if ($success*1) { |
1919 |
warn "scalar has chanegd address in memory\n"; |
1920 |
} |
1921 |
|
1922 |
"IO::AIO::MREMAP_FIXED" and the $new_address argument are currently |
1923 |
implemented, but not supported and might go away in a future |
1924 |
version. |
1925 |
|
1926 |
On systems where this call is not supported or is not emulated, this |
1927 |
call returns falls and sets $! to "ENOSYS". |
1928 |
|
1929 |
IO::AIO::mlockall $flags |
1930 |
Calls the "eio_mlockall_sync" function, which is like |
1931 |
"aio_mlockall", but is blocking. |
1932 |
|
1933 |
IO::AIO::munlock $scalar, $offset = 0, $length = undef |
1934 |
Calls the "munlock" function, undoing the effects of a previous |
1935 |
"aio_mlock" call (see its description for details). |
1936 |
|
1937 |
IO::AIO::munlockall |
1938 |
Calls the "munlockall" function. |
1939 |
|
1940 |
On systems that do not implement "munlockall", this function returns |
1941 |
ENOSYS, otherwise the return value of "munlockall". |
1942 |
|
1943 |
$fh = IO::AIO::accept4 $r_fh, $sockaddr, $sockaddr_maxlen, $flags |
1944 |
Uses the GNU/Linux accept4(2) syscall, if available, to accept a |
1945 |
socket and return the new file handle on success, or sets $! and |
1946 |
returns "undef" on error. |
1947 |
|
1948 |
The remote name of the new socket will be stored in $sockaddr, which |
1949 |
will be extended to allow for at least $sockaddr_maxlen octets. If |
1950 |
the socket name does not fit into $sockaddr_maxlen octets, this is |
1951 |
signaled by returning a longer string in $sockaddr, which might or |
1952 |
might not be truncated. |
1953 |
|
1954 |
To accept name-less sockets, use "undef" for $sockaddr and 0 for |
1955 |
$sockaddr_maxlen. |
1956 |
|
1957 |
The main reasons to use this syscall rather than portable accept(2) |
1958 |
are that you can specify "SOCK_NONBLOCK" and/or "SOCK_CLOEXEC" flags |
1959 |
and you can accept name-less sockets by specifying 0 for |
1960 |
$sockaddr_maxlen, which is sadly not possible with perl's interface |
1961 |
to "accept". |
1962 |
|
1963 |
IO::AIO::splice $r_fh, $r_off, $w_fh, $w_off, $length, $flags |
1964 |
Calls the GNU/Linux splice(2) syscall, if available. If $r_off or |
1965 |
$w_off are "undef", then "NULL" is passed for these, otherwise they |
1966 |
should be the file offset. |
1967 |
|
1968 |
$r_fh and $w_fh should not refer to the same file, as splice might |
1969 |
silently corrupt the data in this case. |
1970 |
|
1971 |
The following symbol flag values are available: |
1972 |
"IO::AIO::SPLICE_F_MOVE", "IO::AIO::SPLICE_F_NONBLOCK", |
1973 |
"IO::AIO::SPLICE_F_MORE" and "IO::AIO::SPLICE_F_GIFT". |
1974 |
|
1975 |
See the splice(2) manpage for details. |
1976 |
|
1977 |
IO::AIO::tee $r_fh, $w_fh, $length, $flags |
1978 |
Calls the GNU/Linux tee(2) syscall, see its manpage and the |
1979 |
description for "IO::AIO::splice" above for details. |
1980 |
|
1981 |
$actual_size = IO::AIO::pipesize $r_fh[, $new_size] |
1982 |
Attempts to query or change the pipe buffer size. Obviously works |
1983 |
only on pipes, and currently works only on GNU/Linux systems, and |
1984 |
fails with -1/"ENOSYS" everywhere else. If anybody knows how to |
1985 |
influence pipe buffer size on other systems, drop me a note. |
1986 |
|
1987 |
($rfh, $wfh) = IO::AIO::pipe2 [$flags] |
1988 |
This is a direct interface to the Linux pipe2(2) system call. If |
1989 |
$flags is missing or 0, then this should be the same as a call to |
1990 |
perl's built-in "pipe" function and create a new pipe, and works on |
1991 |
systems that lack the pipe2 syscall. On win32, this case invokes |
1992 |
"_pipe (..., 4096, O_BINARY)". |
1993 |
|
1994 |
If $flags is non-zero, it tries to invoke the pipe2 system call with |
1995 |
the given flags (Linux 2.6.27, glibc 2.9). |
1996 |
|
1997 |
On success, the read and write file handles are returned. |
1998 |
|
1999 |
On error, nothing will be returned. If the pipe2 syscall is missing |
2000 |
and $flags is non-zero, fails with "ENOSYS". |
2001 |
|
2002 |
Please refer to pipe2(2) for more info on the $flags, but at the |
2003 |
time of this writing, "IO::AIO::O_CLOEXEC", "IO::AIO::O_NONBLOCK" |
2004 |
and "IO::AIO::O_DIRECT" (Linux 3.4, for packet-based pipes) were |
2005 |
supported. |
2006 |
|
2007 |
Example: create a pipe race-free w.r.t. threads and fork: |
2008 |
|
2009 |
my ($rfh, $wfh) = IO::AIO::pipe2 IO::AIO::O_CLOEXEC |
2010 |
or die "pipe2: $!\n"; |
2011 |
|
2012 |
$fh = IO::AIO::memfd_create $pathname[, $flags] |
2013 |
This is a direct interface to the Linux memfd_create(2) system call. |
2014 |
The (unhelpful) default for $flags is 0, but your default should be |
2015 |
"IO::AIO::MFD_CLOEXEC". |
2016 |
|
2017 |
On success, the new memfd filehandle is returned, otherwise returns |
2018 |
"undef". If the memfd_create syscall is missing, fails with |
2019 |
"ENOSYS". |
2020 |
|
2021 |
Please refer to memfd_create(2) for more info on this call. |
2022 |
|
2023 |
The following $flags values are available: "IO::AIO::MFD_CLOEXEC", |
2024 |
"IO::AIO::MFD_ALLOW_SEALING", "IO::AIO::MFD_HUGETLB", |
2025 |
"IO::AIO::MFD_HUGETLB_2MB" and "IO::AIO::MFD_HUGETLB_1GB". |
2026 |
|
2027 |
Example: create a new memfd. |
2028 |
|
2029 |
my $fh = IO::AIO::memfd_create "somenameforprocfd", IO::AIO::MFD_CLOEXEC |
2030 |
or die "memfd_create: $!\n"; |
2031 |
|
2032 |
$fh = IO::AIO::pidfd_open $pid[, $flags] |
2033 |
This is an interface to the Linux pidfd_open(2) system call. The |
2034 |
default for $flags is 0. |
2035 |
|
2036 |
On success, a new pidfd filehandle is returned (that is already set |
2037 |
to close-on-exec), otherwise returns "undef". If the syscall is |
2038 |
missing, fails with "ENOSYS". |
2039 |
|
2040 |
Example: open pid 6341 as pidfd. |
2041 |
|
2042 |
my $fh = IO::AIO::pidfd_open 6341 |
2043 |
or die "pidfd_open: $!\n"; |
2044 |
|
2045 |
$status = IO::AIO::pidfd_send_signal $pidfh, $signal[, $siginfo[, |
2046 |
$flags]] |
2047 |
This is an interface to the Linux pidfd_send_signal system call. The |
2048 |
default for $siginfo is "undef" and the default for $flags is 0. |
2049 |
|
2050 |
Returns the system call status. If the syscall is missing, fails |
2051 |
with "ENOSYS". |
2052 |
|
2053 |
When specified, $siginfo must be a reference to a hash with one or |
2054 |
more of the following members: |
2055 |
|
2056 |
code - the "si_code" member |
2057 |
pid - the "si_pid" member |
2058 |
uid - the "si_uid" member |
2059 |
value_int - the "si_value.sival_int" member |
2060 |
value_ptr - the "si_value.sival_ptr" member, specified as an integer |
2061 |
|
2062 |
Example: send a SIGKILL to the specified process. |
2063 |
|
2064 |
my $status = IO::AIO::pidfd_send_signal $pidfh, 9, undef |
2065 |
and die "pidfd_send_signal: $!\n"; |
2066 |
|
2067 |
Example: send a SIGKILL to the specified process with extra data. |
2068 |
|
2069 |
my $status = IO::AIO::pidfd_send_signal $pidfh, 9, { code => -1, value_int => 7 } |
2070 |
and die "pidfd_send_signal: $!\n"; |
2071 |
|
2072 |
$fh = IO::AIO::pidfd_getfd $pidfh, $targetfd[, $flags] |
2073 |
This is an interface to the Linux pidfd_getfd system call. The |
2074 |
default for $flags is 0. |
2075 |
|
2076 |
On success, returns a dup'ed copy of the target file descriptor |
2077 |
(specified as an integer) returned (that is already set to |
2078 |
close-on-exec), otherwise returns "undef". If the syscall is |
2079 |
missing, fails with "ENOSYS". |
2080 |
|
2081 |
Example: get a copy of standard error of another process and print |
2082 |
soemthing to it. |
2083 |
|
2084 |
my $errfh = IO::AIO::pidfd_getfd $pidfh, 2 |
2085 |
or die "pidfd_getfd: $!\n"; |
2086 |
print $errfh "stderr\n"; |
2087 |
|
2088 |
$fh = IO::AIO::eventfd [$initval, [$flags]] |
2089 |
This is a direct interface to the Linux eventfd(2) system call. The |
2090 |
(unhelpful) defaults for $initval and $flags are 0 for both. |
2091 |
|
2092 |
On success, the new eventfd filehandle is returned, otherwise |
2093 |
returns "undef". If the eventfd syscall is missing, fails with |
2094 |
"ENOSYS". |
2095 |
|
2096 |
Please refer to eventfd(2) for more info on this call. |
2097 |
|
2098 |
The following symbol flag values are available: |
2099 |
"IO::AIO::EFD_CLOEXEC", "IO::AIO::EFD_NONBLOCK" and |
2100 |
"IO::AIO::EFD_SEMAPHORE" (Linux 2.6.30). |
2101 |
|
2102 |
Example: create a new eventfd filehandle: |
2103 |
|
2104 |
$fh = IO::AIO::eventfd 0, IO::AIO::EFD_CLOEXEC |
2105 |
or die "eventfd: $!\n"; |
2106 |
|
2107 |
$fh = IO::AIO::timerfd_create $clockid[, $flags] |
2108 |
This is a direct interface to the Linux timerfd_create(2) system |
2109 |
call. The (unhelpful) default for $flags is 0, but your default |
2110 |
should be "IO::AIO::TFD_CLOEXEC". |
2111 |
|
2112 |
On success, the new timerfd filehandle is returned, otherwise |
2113 |
returns "undef". If the timerfd_create syscall is missing, fails |
2114 |
with "ENOSYS". |
2115 |
|
2116 |
Please refer to timerfd_create(2) for more info on this call. |
2117 |
|
2118 |
The following $clockid values are available: |
2119 |
"IO::AIO::CLOCK_REALTIME", "IO::AIO::CLOCK_MONOTONIC" |
2120 |
"IO::AIO::CLOCK_CLOCK_BOOTTIME" (Linux 3.15) |
2121 |
"IO::AIO::CLOCK_CLOCK_REALTIME_ALARM" (Linux 3.11) and |
2122 |
"IO::AIO::CLOCK_CLOCK_BOOTTIME_ALARM" (Linux 3.11). |
2123 |
|
2124 |
The following $flags values are available (Linux 2.6.27): |
2125 |
"IO::AIO::TFD_NONBLOCK" and "IO::AIO::TFD_CLOEXEC". |
2126 |
|
2127 |
Example: create a new timerfd and set it to one-second repeated |
2128 |
alarms, then wait for two alarms: |
2129 |
|
2130 |
my $fh = IO::AIO::timerfd_create IO::AIO::CLOCK_BOOTTIME, IO::AIO::TFD_CLOEXEC |
2131 |
or die "timerfd_create: $!\n"; |
2132 |
|
2133 |
defined IO::AIO::timerfd_settime $fh, 0, 1, 1 |
2134 |
or die "timerfd_settime: $!\n"; |
2135 |
|
2136 |
for (1..2) { |
2137 |
8 == sysread $fh, my $buf, 8 |
2138 |
or die "timerfd read failure\n"; |
2139 |
|
2140 |
printf "number of expirations (likely 1): %d\n", |
2141 |
unpack "Q", $buf; |
2142 |
} |
2143 |
|
2144 |
($cur_interval, $cur_value) = IO::AIO::timerfd_settime $fh, $flags, |
2145 |
$new_interval, $nbw_value |
2146 |
This is a direct interface to the Linux timerfd_settime(2) system |
2147 |
call. Please refer to its manpage for more info on this call. |
2148 |
|
2149 |
The new itimerspec is specified using two (possibly fractional) |
2150 |
second values, $new_interval and $new_value). |
2151 |
|
2152 |
On success, the current interval and value are returned (as per |
2153 |
"timerfd_gettime"). On failure, the empty list is returned. |
2154 |
|
2155 |
The following $flags values are available: |
2156 |
"IO::AIO::TFD_TIMER_ABSTIME" and "IO::AIO::TFD_TIMER_CANCEL_ON_SET". |
2157 |
|
2158 |
See "IO::AIO::timerfd_create" for a full example. |
2159 |
|
2160 |
($cur_interval, $cur_value) = IO::AIO::timerfd_gettime $fh |
2161 |
This is a direct interface to the Linux timerfd_gettime(2) system |
2162 |
call. Please refer to its manpage for more info on this call. |
2163 |
|
2164 |
On success, returns the current values of interval and value for the |
2165 |
given timerfd (as potentially fractional second values). On failure, |
2166 |
the empty list is returned. |
2167 |
|
2168 |
EVENT LOOP INTEGRATION |
2169 |
It is recommended to use AnyEvent::AIO to integrate IO::AIO |
2170 |
automatically into many event loops: |
2171 |
|
2172 |
# AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...) |
2173 |
use AnyEvent::AIO; |
2174 |
|
2175 |
You can also integrate IO::AIO manually into many event loops, here are |
2176 |
some examples of how to do this: |
2177 |
|
2178 |
# EV integration |
2179 |
my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb; |
2180 |
|
2181 |
# Event integration |
2182 |
Event->io (fd => IO::AIO::poll_fileno, |
2183 |
poll => 'r', |
2184 |
cb => \&IO::AIO::poll_cb); |
2185 |
|
2186 |
# Glib/Gtk2 integration |
2187 |
add_watch Glib::IO IO::AIO::poll_fileno, |
2188 |
in => sub { IO::AIO::poll_cb; 1 }; |
2189 |
|
2190 |
# Tk integration |
2191 |
Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "", |
2192 |
readable => \&IO::AIO::poll_cb); |
2193 |
|
2194 |
# Danga::Socket integration |
2195 |
Danga::Socket->AddOtherFds (IO::AIO::poll_fileno => |
2196 |
\&IO::AIO::poll_cb); |
2197 |
|
2198 |
FORK BEHAVIOUR |
2199 |
Usage of pthreads in a program changes the semantics of fork |
2200 |
considerably. Specifically, only async-safe functions can be called |
2201 |
after fork. Perl doesn't know about this, so in general, you cannot call |
2202 |
fork with defined behaviour in perl if pthreads are involved. IO::AIO |
2203 |
uses pthreads, so this applies, but many other extensions and (for |
2204 |
inexplicable reasons) perl itself often is linked against pthreads, so |
2205 |
this limitation applies to quite a lot of perls. |
2206 |
|
2207 |
This module no longer tries to fight your OS, or POSIX. That means |
2208 |
IO::AIO only works in the process that loaded it. Forking is fully |
2209 |
supported, but using IO::AIO in the child is not. |
2210 |
|
2211 |
You might get around by not *using* IO::AIO before (or after) forking. |
2212 |
You could also try to call the IO::AIO::reinit function in the child: |
2213 |
|
2214 |
IO::AIO::reinit |
2215 |
Abandons all current requests and I/O threads and simply |
2216 |
reinitialises all data structures. This is not an operation |
2217 |
supported by any standards, but happens to work on GNU/Linux and |
2218 |
some newer BSD systems. |
2219 |
|
2220 |
The only reasonable use for this function is to call it after |
2221 |
forking, if "IO::AIO" was used in the parent. Calling it while |
2222 |
IO::AIO is active in the process will result in undefined behaviour. |
2223 |
Calling it at any time will also result in any undefined (by POSIX) |
2224 |
behaviour. |
2225 |
|
2226 |
LINUX-SPECIFIC CALLS |
2227 |
When a call is documented as "linux-specific" then this means it |
2228 |
originated on GNU/Linux. "IO::AIO" will usually try to autodetect the |
2229 |
availability and compatibility of such calls regardless of the platform |
2230 |
it is compiled on, so platforms such as FreeBSD which often implement |
2231 |
these calls will work. When in doubt, call them and see if they fail wth |
2232 |
"ENOSYS". |
2233 |
|
2234 |
MEMORY USAGE |
2235 |
Per-request usage: |
2236 |
|
2237 |
Each aio request uses - depending on your architecture - around 100-200 |
2238 |
bytes of memory. In addition, stat requests need a stat buffer (possibly |
2239 |
a few hundred bytes), readdir requires a result buffer and so on. Perl |
2240 |
scalars and other data passed into aio requests will also be locked and |
2241 |
will consume memory till the request has entered the done state. |
2242 |
|
2243 |
This is not awfully much, so queuing lots of requests is not usually a |
2244 |
problem. |
2245 |
|
2246 |
Per-thread usage: |
2247 |
|
2248 |
In the execution phase, some aio requests require more memory for |
2249 |
temporary buffers, and each thread requires a stack and other data |
2250 |
structures (usually around 16k-128k, depending on the OS). |
2251 |
|
2252 |
KNOWN BUGS |
2253 |
Known bugs will be fixed in the next release :) |
2254 |
|
2255 |
KNOWN ISSUES |
2256 |
Calls that try to "import" foreign memory areas (such as "IO::AIO::mmap" |
2257 |
or "IO::AIO::aio_slurp") do not work with generic lvalues, such as |
2258 |
non-created hash slots or other scalars I didn't think of. It's best to |
2259 |
avoid such and either use scalar variables or making sure that the |
2260 |
scalar exists (e.g. by storing "undef") and isn't "funny" (e.g. tied). |
2261 |
|
2262 |
I am not sure anything can be done about this, so this is considered a |
2263 |
known issue, rather than a bug. |
2264 |
|
2265 |
SEE ALSO |
2266 |
AnyEvent::AIO for easy integration into event loops, Coro::AIO for a |
2267 |
more natural syntax and IO::FDPass for file descriptor passing. |
2268 |
|
2269 |
AUTHOR |
2270 |
Marc Lehmann <schmorp@schmorp.de> |
2271 |
http://home.schmorp.de/ |
2272 |
|