1 |
NAME |
2 |
IO::AIO - Asynchronous Input/Output |
3 |
|
4 |
SYNOPSIS |
5 |
use IO::AIO; |
6 |
|
7 |
aio_open "/etc/passwd", 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 |
# AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...) |
30 |
use AnyEvent::AIO; |
31 |
|
32 |
# EV integration |
33 |
my $w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb; |
34 |
|
35 |
# Event integration |
36 |
Event->io (fd => IO::AIO::poll_fileno, |
37 |
poll => 'r', |
38 |
cb => \&IO::AIO::poll_cb); |
39 |
|
40 |
# Glib/Gtk2 integration |
41 |
add_watch Glib::IO IO::AIO::poll_fileno, |
42 |
in => sub { IO::AIO::poll_cb; 1 }; |
43 |
|
44 |
# Tk integration |
45 |
Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "", |
46 |
readable => \&IO::AIO::poll_cb); |
47 |
|
48 |
# Danga::Socket integration |
49 |
Danga::Socket->AddOtherFds (IO::AIO::poll_fileno => |
50 |
\&IO::AIO::poll_cb); |
51 |
|
52 |
DESCRIPTION |
53 |
This module implements asynchronous I/O using whatever means your |
54 |
operating system supports. |
55 |
|
56 |
Asynchronous means that operations that can normally block your program |
57 |
(e.g. reading from disk) will be done asynchronously: the operation will |
58 |
still block, but you can do something else in the meantime. This is |
59 |
extremely useful for programs that need to stay interactive even when |
60 |
doing heavy I/O (GUI programs, high performance network servers etc.), |
61 |
but can also be used to easily do operations in parallel that are |
62 |
normally done sequentially, e.g. stat'ing many files, which is much |
63 |
faster on a RAID volume or over NFS when you do a number of stat |
64 |
operations concurrently. |
65 |
|
66 |
While most of this works on all types of file descriptors (for example |
67 |
sockets), using these functions on file descriptors that support |
68 |
nonblocking operation (again, sockets, pipes etc.) is very inefficient. |
69 |
Use an event loop for that (such as the Event module): IO::AIO will |
70 |
naturally fit into such an event loop itself. |
71 |
|
72 |
In this version, a number of threads are started that execute your |
73 |
requests and signal their completion. You don't need thread support in |
74 |
perl, and the threads created by this module will not be visible to |
75 |
perl. In the future, this module might make use of the native aio |
76 |
functions available on many operating systems. However, they are often |
77 |
not well-supported or restricted (GNU/Linux doesn't allow them on normal |
78 |
files currently, for example), and they would only support aio_read and |
79 |
aio_write, so the remaining functionality would have to be implemented |
80 |
using threads anyway. |
81 |
|
82 |
Although the module will work in the presence of other (Perl-) threads, |
83 |
it is currently not reentrant in any way, so use appropriate locking |
84 |
yourself, always call "poll_cb" from within the same thread, or never |
85 |
call "poll_cb" (or other "aio_" functions) recursively. |
86 |
|
87 |
EXAMPLE |
88 |
This is a simple example that uses the Event module and loads |
89 |
/etc/passwd asynchronously: |
90 |
|
91 |
use Fcntl; |
92 |
use Event; |
93 |
use IO::AIO; |
94 |
|
95 |
# register the IO::AIO callback with Event |
96 |
Event->io (fd => IO::AIO::poll_fileno, |
97 |
poll => 'r', |
98 |
cb => \&IO::AIO::poll_cb); |
99 |
|
100 |
# queue the request to open /etc/passwd |
101 |
aio_open "/etc/passwd", O_RDONLY, 0, sub { |
102 |
my $fh = shift |
103 |
or die "error while opening: $!"; |
104 |
|
105 |
# stat'ing filehandles is generally non-blocking |
106 |
my $size = -s $fh; |
107 |
|
108 |
# queue a request to read the file |
109 |
my $contents; |
110 |
aio_read $fh, 0, $size, $contents, 0, sub { |
111 |
$_[0] == $size |
112 |
or die "short read: $!"; |
113 |
|
114 |
close $fh; |
115 |
|
116 |
# file contents now in $contents |
117 |
print $contents; |
118 |
|
119 |
# exit event loop and program |
120 |
Event::unloop; |
121 |
}; |
122 |
}; |
123 |
|
124 |
# possibly queue up other requests, or open GUI windows, |
125 |
# check for sockets etc. etc. |
126 |
|
127 |
# process events as long as there are some: |
128 |
Event::loop; |
129 |
|
130 |
REQUEST ANATOMY AND LIFETIME |
131 |
Every "aio_*" function creates a request. which is a C data structure |
132 |
not directly visible to Perl. |
133 |
|
134 |
If called in non-void context, every request function returns a Perl |
135 |
object representing the request. In void context, nothing is returned, |
136 |
which saves a bit of memory. |
137 |
|
138 |
The perl object is a fairly standard ref-to-hash object. The hash |
139 |
contents are not used by IO::AIO so you are free to store anything you |
140 |
like in it. |
141 |
|
142 |
During their existance, aio requests travel through the following |
143 |
states, in order: |
144 |
|
145 |
ready |
146 |
Immediately after a request is created it is put into the ready |
147 |
state, waiting for a thread to execute it. |
148 |
|
149 |
execute |
150 |
A thread has accepted the request for processing and is currently |
151 |
executing it (e.g. blocking in read). |
152 |
|
153 |
pending |
154 |
The request has been executed and is waiting for result processing. |
155 |
|
156 |
While request submission and execution is fully asynchronous, result |
157 |
processing is not and relies on the perl interpreter calling |
158 |
"poll_cb" (or another function with the same effect). |
159 |
|
160 |
result |
161 |
The request results are processed synchronously by "poll_cb". |
162 |
|
163 |
The "poll_cb" function will process all outstanding aio requests by |
164 |
calling their callbacks, freeing memory associated with them and |
165 |
managing any groups they are contained in. |
166 |
|
167 |
done |
168 |
Request has reached the end of its lifetime and holds no resources |
169 |
anymore (except possibly for the Perl object, but its connection to |
170 |
the actual aio request is severed and calling its methods will |
171 |
either do nothing or result in a runtime error). |
172 |
|
173 |
FUNCTIONS |
174 |
AIO REQUEST FUNCTIONS |
175 |
All the "aio_*" calls are more or less thin wrappers around the syscall |
176 |
with the same name (sans "aio_"). The arguments are similar or |
177 |
identical, and they all accept an additional (and optional) $callback |
178 |
argument which must be a code reference. This code reference will get |
179 |
called with the syscall return code (e.g. most syscalls return -1 on |
180 |
error, unlike perl, which usually delivers "false") as its sole argument |
181 |
after the given syscall has been executed asynchronously. |
182 |
|
183 |
All functions expecting a filehandle keep a copy of the filehandle |
184 |
internally until the request has finished. |
185 |
|
186 |
All functions return request objects of type IO::AIO::REQ that allow |
187 |
further manipulation of those requests while they are in-flight. |
188 |
|
189 |
The pathnames you pass to these routines *must* be absolute and encoded |
190 |
as octets. The reason for the former is that at the time the request is |
191 |
being executed, the current working directory could have changed. |
192 |
Alternatively, you can make sure that you never change the current |
193 |
working directory anywhere in the program and then use relative paths. |
194 |
|
195 |
To encode pathnames as octets, either make sure you either: a) always |
196 |
pass in filenames you got from outside (command line, readdir etc.) |
197 |
without tinkering, b) are ASCII or ISO 8859-1, c) use the Encode module |
198 |
and encode your pathnames to the locale (or other) encoding in effect in |
199 |
the user environment, d) use Glib::filename_from_unicode on unicode |
200 |
filenames or e) use something else to ensure your scalar has the correct |
201 |
contents. |
202 |
|
203 |
This works, btw. independent of the internal UTF-8 bit, which IO::AIO |
204 |
handles correctly whether it is set or not. |
205 |
|
206 |
$prev_pri = aioreq_pri [$pri] |
207 |
Returns the priority value that would be used for the next request |
208 |
and, if $pri is given, sets the priority for the next aio request. |
209 |
|
210 |
The default priority is 0, the minimum and maximum priorities are -4 |
211 |
and 4, respectively. Requests with higher priority will be serviced |
212 |
first. |
213 |
|
214 |
The priority will be reset to 0 after each call to one of the |
215 |
"aio_*" functions. |
216 |
|
217 |
Example: open a file with low priority, then read something from it |
218 |
with higher priority so the read request is serviced before other |
219 |
low priority open requests (potentially spamming the cache): |
220 |
|
221 |
aioreq_pri -3; |
222 |
aio_open ..., sub { |
223 |
return unless $_[0]; |
224 |
|
225 |
aioreq_pri -2; |
226 |
aio_read $_[0], ..., sub { |
227 |
... |
228 |
}; |
229 |
}; |
230 |
|
231 |
aioreq_nice $pri_adjust |
232 |
Similar to "aioreq_pri", but subtracts the given value from the |
233 |
current priority, so the effect is cumulative. |
234 |
|
235 |
aio_open $pathname, $flags, $mode, $callback->($fh) |
236 |
Asynchronously open or create a file and call the callback with a |
237 |
newly created filehandle for the file. |
238 |
|
239 |
The pathname passed to "aio_open" must be absolute. See API NOTES, |
240 |
above, for an explanation. |
241 |
|
242 |
The $flags argument is a bitmask. See the "Fcntl" module for a list. |
243 |
They are the same as used by "sysopen". |
244 |
|
245 |
Likewise, $mode specifies the mode of the newly created file, if it |
246 |
didn't exist and "O_CREAT" has been given, just like perl's |
247 |
"sysopen", except that it is mandatory (i.e. use 0 if you don't |
248 |
create new files, and 0666 or 0777 if you do). Note that the $mode |
249 |
will be modified by the umask in effect then the request is being |
250 |
executed, so better never change the umask. |
251 |
|
252 |
Example: |
253 |
|
254 |
aio_open "/etc/passwd", O_RDONLY, 0, sub { |
255 |
if ($_[0]) { |
256 |
print "open successful, fh is $_[0]\n"; |
257 |
... |
258 |
} else { |
259 |
die "open failed: $!\n"; |
260 |
} |
261 |
}; |
262 |
|
263 |
aio_close $fh, $callback->($status) |
264 |
Asynchronously close a file and call the callback with the result |
265 |
code. |
266 |
|
267 |
Unfortunately, you can't do this to perl. Perl *insists* very |
268 |
strongly on closing the file descriptor associated with the |
269 |
filehandle itself. |
270 |
|
271 |
Therefore, "aio_close" will not close the filehandle - instead it |
272 |
will use dup2 to overwrite the file descriptor with the write-end of |
273 |
a pipe (the pipe fd will be created on demand and will be cached). |
274 |
|
275 |
Or in other words: the file descriptor will be closed, but it will |
276 |
not be free for reuse until the perl filehandle is closed. |
277 |
|
278 |
aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
279 |
aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
280 |
Reads or writes $length bytes from or to the specified $fh and |
281 |
$offset into the scalar given by $data and offset $dataoffset and |
282 |
calls the callback without the actual number of bytes read (or -1 on |
283 |
error, just like the syscall). |
284 |
|
285 |
"aio_read" will, like "sysread", shrink or grow the $data scalar to |
286 |
offset plus the actual number of bytes read. |
287 |
|
288 |
If $offset is undefined, then the current file descriptor offset |
289 |
will be used (and updated), otherwise the file descriptor offset |
290 |
will not be changed by these calls. |
291 |
|
292 |
If $length is undefined in "aio_write", use the remaining length of |
293 |
$data. |
294 |
|
295 |
If $dataoffset is less than zero, it will be counted from the end of |
296 |
$data. |
297 |
|
298 |
The $data scalar *MUST NOT* be modified in any way while the request |
299 |
is outstanding. Modifying it can result in segfaults or World War |
300 |
III (if the necessary/optional hardware is installed). |
301 |
|
302 |
Example: Read 15 bytes at offset 7 into scalar $buffer, starting at |
303 |
offset 0 within the scalar: |
304 |
|
305 |
aio_read $fh, 7, 15, $buffer, 0, sub { |
306 |
$_[0] > 0 or die "read error: $!"; |
307 |
print "read $_[0] bytes: <$buffer>\n"; |
308 |
}; |
309 |
|
310 |
aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval) |
311 |
Tries to copy $length bytes from $in_fh to $out_fh. It starts |
312 |
reading at byte offset $in_offset, and starts writing at the current |
313 |
file offset of $out_fh. Because of that, it is not safe to issue |
314 |
more than one "aio_sendfile" per $out_fh, as they will interfere |
315 |
with each other. |
316 |
|
317 |
This call tries to make use of a native "sendfile" syscall to |
318 |
provide zero-copy operation. For this to work, $out_fh should refer |
319 |
to a socket, and $in_fh should refer to mmap'able file. |
320 |
|
321 |
If the native sendfile call fails or is not implemented, it will be |
322 |
emulated, so you can call "aio_sendfile" on any type of filehandle |
323 |
regardless of the limitations of the operating system. |
324 |
|
325 |
Please note, however, that "aio_sendfile" can read more bytes from |
326 |
$in_fh than are written, and there is no way to find out how many |
327 |
bytes have been read from "aio_sendfile" alone, as "aio_sendfile" |
328 |
only provides the number of bytes written to $out_fh. Only if the |
329 |
result value equals $length one can assume that $length bytes have |
330 |
been read. |
331 |
|
332 |
aio_readahead $fh,$offset,$length, $callback->($retval) |
333 |
"aio_readahead" populates the page cache with data from a file so |
334 |
that subsequent reads from that file will not block on disk I/O. The |
335 |
$offset argument specifies the starting point from which data is to |
336 |
be read and $length specifies the number of bytes to be read. I/O is |
337 |
performed in whole pages, so that offset is effectively rounded down |
338 |
to a page boundary and bytes are read up to the next page boundary |
339 |
greater than or equal to (off-set+length). "aio_readahead" does not |
340 |
read beyond the end of the file. The current file offset of the file |
341 |
is left unchanged. |
342 |
|
343 |
If that syscall doesn't exist (likely if your OS isn't Linux) it |
344 |
will be emulated by simply reading the data, which would have a |
345 |
similar effect. |
346 |
|
347 |
aio_stat $fh_or_path, $callback->($status) |
348 |
aio_lstat $fh, $callback->($status) |
349 |
Works like perl's "stat" or "lstat" in void context. The callback |
350 |
will be called after the stat and the results will be available |
351 |
using "stat _" or "-s _" etc... |
352 |
|
353 |
The pathname passed to "aio_stat" must be absolute. See API NOTES, |
354 |
above, for an explanation. |
355 |
|
356 |
Currently, the stats are always 64-bit-stats, i.e. instead of |
357 |
returning an error when stat'ing a large file, the results will be |
358 |
silently truncated unless perl itself is compiled with large file |
359 |
support. |
360 |
|
361 |
Example: Print the length of /etc/passwd: |
362 |
|
363 |
aio_stat "/etc/passwd", sub { |
364 |
$_[0] and die "stat failed: $!"; |
365 |
print "size is ", -s _, "\n"; |
366 |
}; |
367 |
|
368 |
aio_utime $fh_or_path, $atime, $mtime, $callback->($status) |
369 |
Works like perl's "utime" function (including the special case of |
370 |
$atime and $mtime being undef). Fractional times are supported if |
371 |
the underlying syscalls support them. |
372 |
|
373 |
When called with a pathname, uses utimes(2) if available, otherwise |
374 |
utime(2). If called on a file descriptor, uses futimes(2) if |
375 |
available, otherwise returns ENOSYS, so this is not portable. |
376 |
|
377 |
Examples: |
378 |
|
379 |
# set atime and mtime to current time (basically touch(1)): |
380 |
aio_utime "path", undef, undef; |
381 |
# set atime to current time and mtime to beginning of the epoch: |
382 |
aio_utime "path", time, undef; # undef==0 |
383 |
|
384 |
aio_chown $fh_or_path, $uid, $gid, $callback->($status) |
385 |
Works like perl's "chown" function, except that "undef" for either |
386 |
$uid or $gid is being interpreted as "do not change" (but -1 can |
387 |
also be used). |
388 |
|
389 |
Examples: |
390 |
|
391 |
# same as "chown root path" in the shell: |
392 |
aio_chown "path", 0, -1; |
393 |
# same as above: |
394 |
aio_chown "path", 0, undef; |
395 |
|
396 |
aio_truncate $fh_or_path, $offset, $callback->($status) |
397 |
Works like truncate(2) or ftruncate(2). |
398 |
|
399 |
aio_chmod $fh_or_path, $mode, $callback->($status) |
400 |
Works like perl's "chmod" function. |
401 |
|
402 |
aio_unlink $pathname, $callback->($status) |
403 |
Asynchronously unlink (delete) a file and call the callback with the |
404 |
result code. |
405 |
|
406 |
aio_mknod $path, $mode, $dev, $callback->($status) |
407 |
[EXPERIMENTAL] |
408 |
|
409 |
Asynchronously create a device node (or fifo). See mknod(2). |
410 |
|
411 |
The only (POSIX-) portable way of calling this function is: |
412 |
|
413 |
aio_mknod $path, IO::AIO::S_IFIFO | $mode, 0, sub { ... |
414 |
|
415 |
aio_link $srcpath, $dstpath, $callback->($status) |
416 |
Asynchronously create a new link to the existing object at $srcpath |
417 |
at the path $dstpath and call the callback with the result code. |
418 |
|
419 |
aio_symlink $srcpath, $dstpath, $callback->($status) |
420 |
Asynchronously create a new symbolic link to the existing object at |
421 |
$srcpath at the path $dstpath and call the callback with the result |
422 |
code. |
423 |
|
424 |
aio_readlink $path, $callback->($link) |
425 |
Asynchronously read the symlink specified by $path and pass it to |
426 |
the callback. If an error occurs, nothing or undef gets passed to |
427 |
the callback. |
428 |
|
429 |
aio_rename $srcpath, $dstpath, $callback->($status) |
430 |
Asynchronously rename the object at $srcpath to $dstpath, just as |
431 |
rename(2) and call the callback with the result code. |
432 |
|
433 |
aio_mkdir $pathname, $mode, $callback->($status) |
434 |
Asynchronously mkdir (create) a directory and call the callback with |
435 |
the result code. $mode will be modified by the umask at the time the |
436 |
request is executed, so do not change your umask. |
437 |
|
438 |
aio_rmdir $pathname, $callback->($status) |
439 |
Asynchronously rmdir (delete) a directory and call the callback with |
440 |
the result code. |
441 |
|
442 |
aio_readdir $pathname, $callback->($entries) |
443 |
Unlike the POSIX call of the same name, "aio_readdir" reads an |
444 |
entire directory (i.e. opendir + readdir + closedir). The entries |
445 |
will not be sorted, and will NOT include the "." and ".." entries. |
446 |
|
447 |
The callback is passed a single argument which is either "undef" or |
448 |
an array-ref with the filenames. |
449 |
|
450 |
aio_readdirx $pathname, $flags, $callback->($entries, $flags) |
451 |
Quite similar to "aio_readdir", but the $flags argument allows to |
452 |
tune behaviour and output format. In case of an error, $entries will |
453 |
be "undef". |
454 |
|
455 |
The flags are a combination of the following constants, ORed |
456 |
together (the flags will also be passed to the callback, possibly |
457 |
modified): |
458 |
|
459 |
IO::AIO::READDIR_DENTS |
460 |
When this flag is off, then the callback gets an arrayref with |
461 |
of names only (as with "aio_readdir"), otherwise it gets an |
462 |
arrayref with "[$name, $type, $inode]" arrayrefs, each |
463 |
describing a single directory entry in more detail. |
464 |
|
465 |
$name is the name of the entry. |
466 |
|
467 |
$type is one of the "IO::AIO::DT_xxx" constants: |
468 |
|
469 |
"IO::AIO::DT_UNKNOWN", "IO::AIO::DT_FIFO", "IO::AIO::DT_CHR", |
470 |
"IO::AIO::DT_DIR", "IO::AIO::DT_BLK", "IO::AIO::DT_REG", |
471 |
"IO::AIO::DT_LNK", "IO::AIO::DT_SOCK", "IO::AIO::DT_WHT". |
472 |
|
473 |
"IO::AIO::DT_UNKNOWN" means just that: readdir does not know. If |
474 |
you need to know, you have to run stat yourself. Also, for speed |
475 |
reasons, the $type scalars are read-only: you can not modify |
476 |
them. |
477 |
|
478 |
$inode is the inode number (which might not be exact on systems |
479 |
with 64 bit inode numbers and 32 bit perls). On systems that do |
480 |
not deliver the inode information, this will always be zero. |
481 |
|
482 |
IO::AIO::READDIR_DIRS_FIRST |
483 |
When this flag is set, then the names will be returned in an |
484 |
order where likely directories come first. This is useful when |
485 |
you need to quickly find directories, or you want to find all |
486 |
directories while avoiding to stat() each entry. |
487 |
|
488 |
If the system returns type information in readdir, then this is |
489 |
used to find directories directly. Otherwise, likely directories |
490 |
are files beginning with ".", or otherwise files with no dots, |
491 |
of which files with short names are tried first. |
492 |
|
493 |
IO::AIO::READDIR_STAT_ORDER |
494 |
When this flag is set, then the names will be returned in an |
495 |
order suitable for stat()'ing each one. That is, when you plan |
496 |
to stat() all files in the given directory, then the returned |
497 |
order will likely be fastest. |
498 |
|
499 |
If both this flag and "IO::AIO::READDIR_DIRS_FIRST" are |
500 |
specified, then the likely dirs come first, resulting in a less |
501 |
optimal stat order. |
502 |
|
503 |
IO::AIO::READDIR_FOUND_UNKNOWN |
504 |
This flag should not be set when calling "aio_readdirx". |
505 |
Instead, it is being set by "aio_readdirx", when any of the |
506 |
$type's found were "IO::AIO::DT_UNKNOWN". The absense of this |
507 |
flag therefore indicates that all $type's are known, which can |
508 |
be used to speed up some algorithms. |
509 |
|
510 |
aio_load $path, $data, $callback->($status) |
511 |
This is a composite request that tries to fully load the given file |
512 |
into memory. Status is the same as with aio_read. |
513 |
|
514 |
aio_copy $srcpath, $dstpath, $callback->($status) |
515 |
Try to copy the *file* (directories not supported as either source |
516 |
or destination) from $srcpath to $dstpath and call the callback with |
517 |
the 0 (error) or -1 ok. |
518 |
|
519 |
This is a composite request that creates the destination file with |
520 |
mode 0200 and copies the contents of the source file into it using |
521 |
"aio_sendfile", followed by restoring atime, mtime, access mode and |
522 |
uid/gid, in that order. |
523 |
|
524 |
If an error occurs, the partial destination file will be unlinked, |
525 |
if possible, except when setting atime, mtime, access mode and |
526 |
uid/gid, where errors are being ignored. |
527 |
|
528 |
aio_move $srcpath, $dstpath, $callback->($status) |
529 |
Try to move the *file* (directories not supported as either source |
530 |
or destination) from $srcpath to $dstpath and call the callback with |
531 |
the 0 (error) or -1 ok. |
532 |
|
533 |
This is a composite request that tries to rename(2) the file first; |
534 |
if rename fails with "EXDEV", it copies the file with "aio_copy" |
535 |
and, if that is successful, unlinks the $srcpath. |
536 |
|
537 |
aio_scandir $path, $maxreq, $callback->($dirs, $nondirs) |
538 |
Scans a directory (similar to "aio_readdir") but additionally tries |
539 |
to efficiently separate the entries of directory $path into two sets |
540 |
of names, directories you can recurse into (directories), and ones |
541 |
you cannot recurse into (everything else, including symlinks to |
542 |
directories). |
543 |
|
544 |
"aio_scandir" is a composite request that creates of many sub |
545 |
requests_ $maxreq specifies the maximum number of outstanding aio |
546 |
requests that this function generates. If it is "<= 0", then a |
547 |
suitable default will be chosen (currently 4). |
548 |
|
549 |
On error, the callback is called without arguments, otherwise it |
550 |
receives two array-refs with path-relative entry names. |
551 |
|
552 |
Example: |
553 |
|
554 |
aio_scandir $dir, 0, sub { |
555 |
my ($dirs, $nondirs) = @_; |
556 |
print "real directories: @$dirs\n"; |
557 |
print "everything else: @$nondirs\n"; |
558 |
}; |
559 |
|
560 |
Implementation notes. |
561 |
|
562 |
The "aio_readdir" cannot be avoided, but "stat()"'ing every entry |
563 |
can. |
564 |
|
565 |
If readdir returns file type information, then this is used directly |
566 |
to find directories. |
567 |
|
568 |
Otherwise, after reading the directory, the modification time, size |
569 |
etc. of the directory before and after the readdir is checked, and |
570 |
if they match (and isn't the current time), the link count will be |
571 |
used to decide how many entries are directories (if >= 2). |
572 |
Otherwise, no knowledge of the number of subdirectories will be |
573 |
assumed. |
574 |
|
575 |
Then entries will be sorted into likely directories a non-initial |
576 |
dot currently) and likely non-directories (see "aio_readdirx"). Then |
577 |
every entry plus an appended "/." will be "stat"'ed, likely |
578 |
directories first, in order of their inode numbers. If that |
579 |
succeeds, it assumes that the entry is a directory or a symlink to |
580 |
directory (which will be checked seperately). This is often faster |
581 |
than stat'ing the entry itself because filesystems might detect the |
582 |
type of the entry without reading the inode data (e.g. ext2fs |
583 |
filetype feature), even on systems that cannot return the filetype |
584 |
information on readdir. |
585 |
|
586 |
If the known number of directories (link count - 2) has been |
587 |
reached, the rest of the entries is assumed to be non-directories. |
588 |
|
589 |
This only works with certainty on POSIX (= UNIX) filesystems, which |
590 |
fortunately are the vast majority of filesystems around. |
591 |
|
592 |
It will also likely work on non-POSIX filesystems with reduced |
593 |
efficiency as those tend to return 0 or 1 as link counts, which |
594 |
disables the directory counting heuristic. |
595 |
|
596 |
aio_rmtree $path, $callback->($status) |
597 |
Delete a directory tree starting (and including) $path, return the |
598 |
status of the final "rmdir" only. This is a composite request that |
599 |
uses "aio_scandir" to recurse into and rmdir directories, and unlink |
600 |
everything else. |
601 |
|
602 |
aio_sync $callback->($status) |
603 |
Asynchronously call sync and call the callback when finished. |
604 |
|
605 |
aio_fsync $fh, $callback->($status) |
606 |
Asynchronously call fsync on the given filehandle and call the |
607 |
callback with the fsync result code. |
608 |
|
609 |
aio_fdatasync $fh, $callback->($status) |
610 |
Asynchronously call fdatasync on the given filehandle and call the |
611 |
callback with the fdatasync result code. |
612 |
|
613 |
If this call isn't available because your OS lacks it or it couldn't |
614 |
be detected, it will be emulated by calling "fsync" instead. |
615 |
|
616 |
aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status) |
617 |
Sync the data portion of the file specified by $offset and $length |
618 |
to disk (but NOT the metadata), by calling the Linux-specific |
619 |
sync_file_range call. If sync_file_range is not available or it |
620 |
returns ENOSYS, then fdatasync or fsync is being substituted. |
621 |
|
622 |
$flags can be a combination of |
623 |
"IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE", |
624 |
"IO::AIO::SYNC_FILE_RANGE_WRITE" and |
625 |
"IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER": refer to the sync_file_range |
626 |
manpage for details. |
627 |
|
628 |
aio_pathsync $path, $callback->($status) |
629 |
This request tries to open, fsync and close the given path. This is |
630 |
a composite request intended to sync directories after directory |
631 |
operations (E.g. rename). This might not work on all operating |
632 |
systems or have any specific effect, but usually it makes sure that |
633 |
directory changes get written to disc. It works for anything that |
634 |
can be opened for read-only, not just directories. |
635 |
|
636 |
Passes 0 when everything went ok, and -1 on error. |
637 |
|
638 |
aio_group $callback->(...) |
639 |
This is a very special aio request: Instead of doing something, it |
640 |
is a container for other aio requests, which is useful if you want |
641 |
to bundle many requests into a single, composite, request with a |
642 |
definite callback and the ability to cancel the whole request with |
643 |
its subrequests. |
644 |
|
645 |
Returns an object of class IO::AIO::GRP. See its documentation below |
646 |
for more info. |
647 |
|
648 |
Example: |
649 |
|
650 |
my $grp = aio_group sub { |
651 |
print "all stats done\n"; |
652 |
}; |
653 |
|
654 |
add $grp |
655 |
(aio_stat ...), |
656 |
(aio_stat ...), |
657 |
...; |
658 |
|
659 |
aio_nop $callback->() |
660 |
This is a special request - it does nothing in itself and is only |
661 |
used for side effects, such as when you want to add a dummy request |
662 |
to a group so that finishing the requests in the group depends on |
663 |
executing the given code. |
664 |
|
665 |
While this request does nothing, it still goes through the execution |
666 |
phase and still requires a worker thread. Thus, the callback will |
667 |
not be executed immediately but only after other requests in the |
668 |
queue have entered their execution phase. This can be used to |
669 |
measure request latency. |
670 |
|
671 |
IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED* |
672 |
Mainly used for debugging and benchmarking, this aio request puts |
673 |
one of the request workers to sleep for the given time. |
674 |
|
675 |
While it is theoretically handy to have simple I/O scheduling |
676 |
requests like sleep and file handle readable/writable, the overhead |
677 |
this creates is immense (it blocks a thread for a long time) so do |
678 |
not use this function except to put your application under |
679 |
artificial I/O pressure. |
680 |
|
681 |
IO::AIO::REQ CLASS |
682 |
All non-aggregate "aio_*" functions return an object of this class when |
683 |
called in non-void context. |
684 |
|
685 |
cancel $req |
686 |
Cancels the request, if possible. Has the effect of skipping |
687 |
execution when entering the execute state and skipping calling the |
688 |
callback when entering the the result state, but will leave the |
689 |
request otherwise untouched. That means that requests that currently |
690 |
execute will not be stopped and resources held by the request will |
691 |
not be freed prematurely. |
692 |
|
693 |
cb $req $callback->(...) |
694 |
Replace (or simply set) the callback registered to the request. |
695 |
|
696 |
IO::AIO::GRP CLASS |
697 |
This class is a subclass of IO::AIO::REQ, so all its methods apply to |
698 |
objects of this class, too. |
699 |
|
700 |
A IO::AIO::GRP object is a special request that can contain multiple |
701 |
other aio requests. |
702 |
|
703 |
You create one by calling the "aio_group" constructing function with a |
704 |
callback that will be called when all contained requests have entered |
705 |
the "done" state: |
706 |
|
707 |
my $grp = aio_group sub { |
708 |
print "all requests are done\n"; |
709 |
}; |
710 |
|
711 |
You add requests by calling the "add" method with one or more |
712 |
"IO::AIO::REQ" objects: |
713 |
|
714 |
$grp->add (aio_unlink "..."); |
715 |
|
716 |
add $grp aio_stat "...", sub { |
717 |
$_[0] or return $grp->result ("error"); |
718 |
|
719 |
# add another request dynamically, if first succeeded |
720 |
add $grp aio_open "...", sub { |
721 |
$grp->result ("ok"); |
722 |
}; |
723 |
}; |
724 |
|
725 |
This makes it very easy to create composite requests (see the source of |
726 |
"aio_move" for an application) that work and feel like simple requests. |
727 |
|
728 |
* The IO::AIO::GRP objects will be cleaned up during calls to |
729 |
"IO::AIO::poll_cb", just like any other request. |
730 |
|
731 |
* They can be canceled like any other request. Canceling will cancel |
732 |
not only the request itself, but also all requests it contains. |
733 |
|
734 |
* They can also can also be added to other IO::AIO::GRP objects. |
735 |
|
736 |
* You must not add requests to a group from within the group callback |
737 |
(or any later time). |
738 |
|
739 |
Their lifetime, simplified, looks like this: when they are empty, they |
740 |
will finish very quickly. If they contain only requests that are in the |
741 |
"done" state, they will also finish. Otherwise they will continue to |
742 |
exist. |
743 |
|
744 |
That means after creating a group you have some time to add requests |
745 |
(precisely before the callback has been invoked, which is only done |
746 |
within the "poll_cb"). And in the callbacks of those requests, you can |
747 |
add further requests to the group. And only when all those requests have |
748 |
finished will the the group itself finish. |
749 |
|
750 |
add $grp ... |
751 |
$grp->add (...) |
752 |
Add one or more requests to the group. Any type of IO::AIO::REQ can |
753 |
be added, including other groups, as long as you do not create |
754 |
circular dependencies. |
755 |
|
756 |
Returns all its arguments. |
757 |
|
758 |
$grp->cancel_subs |
759 |
Cancel all subrequests and clears any feeder, but not the group |
760 |
request itself. Useful when you queued a lot of events but got a |
761 |
result early. |
762 |
|
763 |
$grp->result (...) |
764 |
Set the result value(s) that will be passed to the group callback |
765 |
when all subrequests have finished and set the groups errno to the |
766 |
current value of errno (just like calling "errno" without an error |
767 |
number). By default, no argument will be passed and errno is zero. |
768 |
|
769 |
$grp->errno ([$errno]) |
770 |
Sets the group errno value to $errno, or the current value of errno |
771 |
when the argument is missing. |
772 |
|
773 |
Every aio request has an associated errno value that is restored |
774 |
when the callback is invoked. This method lets you change this value |
775 |
from its default (0). |
776 |
|
777 |
Calling "result" will also set errno, so make sure you either set $! |
778 |
before the call to "result", or call c<errno> after it. |
779 |
|
780 |
feed $grp $callback->($grp) |
781 |
Sets a feeder/generator on this group: every group can have an |
782 |
attached generator that generates requests if idle. The idea behind |
783 |
this is that, although you could just queue as many requests as you |
784 |
want in a group, this might starve other requests for a potentially |
785 |
long time. For example, "aio_scandir" might generate hundreds of |
786 |
thousands "aio_stat" requests, delaying any later requests for a |
787 |
long time. |
788 |
|
789 |
To avoid this, and allow incremental generation of requests, you can |
790 |
instead a group and set a feeder on it that generates those |
791 |
requests. The feed callback will be called whenever there are few |
792 |
enough (see "limit", below) requests active in the group itself and |
793 |
is expected to queue more requests. |
794 |
|
795 |
The feed callback can queue as many requests as it likes (i.e. "add" |
796 |
does not impose any limits). |
797 |
|
798 |
If the feed does not queue more requests when called, it will be |
799 |
automatically removed from the group. |
800 |
|
801 |
If the feed limit is 0 when this method is called, it will be set to |
802 |
2 automatically. |
803 |
|
804 |
Example: |
805 |
|
806 |
# stat all files in @files, but only ever use four aio requests concurrently: |
807 |
|
808 |
my $grp = aio_group sub { print "finished\n" }; |
809 |
limit $grp 4; |
810 |
feed $grp sub { |
811 |
my $file = pop @files |
812 |
or return; |
813 |
|
814 |
add $grp aio_stat $file, sub { ... }; |
815 |
}; |
816 |
|
817 |
limit $grp $num |
818 |
Sets the feeder limit for the group: The feeder will be called |
819 |
whenever the group contains less than this many requests. |
820 |
|
821 |
Setting the limit to 0 will pause the feeding process. |
822 |
|
823 |
The default value for the limit is 0, but note that setting a feeder |
824 |
automatically bumps it up to 2. |
825 |
|
826 |
SUPPORT FUNCTIONS |
827 |
EVENT PROCESSING AND EVENT LOOP INTEGRATION |
828 |
$fileno = IO::AIO::poll_fileno |
829 |
Return the *request result pipe file descriptor*. This filehandle |
830 |
must be polled for reading by some mechanism outside this module |
831 |
(e.g. Event or select, see below or the SYNOPSIS). If the pipe |
832 |
becomes readable you have to call "poll_cb" to check the results. |
833 |
|
834 |
See "poll_cb" for an example. |
835 |
|
836 |
IO::AIO::poll_cb |
837 |
Process some outstanding events on the result pipe. You have to call |
838 |
this regularly. Returns 0 if all events could be processed, or -1 if |
839 |
it returned earlier for whatever reason. Returns immediately when no |
840 |
events are outstanding. The amount of events processed depends on |
841 |
the settings of "IO::AIO::max_poll_req" and |
842 |
"IO::AIO::max_poll_time". |
843 |
|
844 |
If not all requests were processed for whatever reason, the |
845 |
filehandle will still be ready when "poll_cb" returns, so normally |
846 |
you don't have to do anything special to have it called later. |
847 |
|
848 |
Example: Install an Event watcher that automatically calls |
849 |
IO::AIO::poll_cb with high priority: |
850 |
|
851 |
Event->io (fd => IO::AIO::poll_fileno, |
852 |
poll => 'r', async => 1, |
853 |
cb => \&IO::AIO::poll_cb); |
854 |
|
855 |
IO::AIO::max_poll_reqs $nreqs |
856 |
IO::AIO::max_poll_time $seconds |
857 |
These set the maximum number of requests (default 0, meaning |
858 |
infinity) that are being processed by "IO::AIO::poll_cb" in one |
859 |
call, respectively the maximum amount of time (default 0, meaning |
860 |
infinity) spent in "IO::AIO::poll_cb" to process requests (more |
861 |
correctly the mininum amount of time "poll_cb" is allowed to use). |
862 |
|
863 |
Setting "max_poll_time" to a non-zero value creates an overhead of |
864 |
one syscall per request processed, which is not normally a problem |
865 |
unless your callbacks are really really fast or your OS is really |
866 |
really slow (I am not mentioning Solaris here). Using |
867 |
"max_poll_reqs" incurs no overhead. |
868 |
|
869 |
Setting these is useful if you want to ensure some level of |
870 |
interactiveness when perl is not fast enough to process all requests |
871 |
in time. |
872 |
|
873 |
For interactive programs, values such as 0.01 to 0.1 should be fine. |
874 |
|
875 |
Example: Install an Event watcher that automatically calls |
876 |
IO::AIO::poll_cb with low priority, to ensure that other parts of |
877 |
the program get the CPU sometimes even under high AIO load. |
878 |
|
879 |
# try not to spend much more than 0.1s in poll_cb |
880 |
IO::AIO::max_poll_time 0.1; |
881 |
|
882 |
# use a low priority so other tasks have priority |
883 |
Event->io (fd => IO::AIO::poll_fileno, |
884 |
poll => 'r', nice => 1, |
885 |
cb => &IO::AIO::poll_cb); |
886 |
|
887 |
IO::AIO::poll_wait |
888 |
If there are any outstanding requests and none of them in the result |
889 |
phase, wait till the result filehandle becomes ready for reading |
890 |
(simply does a "select" on the filehandle. This is useful if you |
891 |
want to synchronously wait for some requests to finish). |
892 |
|
893 |
See "nreqs" for an example. |
894 |
|
895 |
IO::AIO::poll |
896 |
Waits until some requests have been handled. |
897 |
|
898 |
Returns the number of requests processed, but is otherwise strictly |
899 |
equivalent to: |
900 |
|
901 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
902 |
|
903 |
IO::AIO::flush |
904 |
Wait till all outstanding AIO requests have been handled. |
905 |
|
906 |
Strictly equivalent to: |
907 |
|
908 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
909 |
while IO::AIO::nreqs; |
910 |
|
911 |
CONTROLLING THE NUMBER OF THREADS |
912 |
IO::AIO::min_parallel $nthreads |
913 |
Set the minimum number of AIO threads to $nthreads. The current |
914 |
default is 8, which means eight asynchronous operations can execute |
915 |
concurrently at any one time (the number of outstanding requests, |
916 |
however, is unlimited). |
917 |
|
918 |
IO::AIO starts threads only on demand, when an AIO request is queued |
919 |
and no free thread exists. Please note that queueing up a hundred |
920 |
requests can create demand for a hundred threads, even if it turns |
921 |
out that everything is in the cache and could have been processed |
922 |
faster by a single thread. |
923 |
|
924 |
It is recommended to keep the number of threads relatively low, as |
925 |
some Linux kernel versions will scale negatively with the number of |
926 |
threads (higher parallelity => MUCH higher latency). With current |
927 |
Linux 2.6 versions, 4-32 threads should be fine. |
928 |
|
929 |
Under most circumstances you don't need to call this function, as |
930 |
the module selects a default that is suitable for low to moderate |
931 |
load. |
932 |
|
933 |
IO::AIO::max_parallel $nthreads |
934 |
Sets the maximum number of AIO threads to $nthreads. If more than |
935 |
the specified number of threads are currently running, this function |
936 |
kills them. This function blocks until the limit is reached. |
937 |
|
938 |
While $nthreads are zero, aio requests get queued but not executed |
939 |
until the number of threads has been increased again. |
940 |
|
941 |
This module automatically runs "max_parallel 0" at program end, to |
942 |
ensure that all threads are killed and that there are no outstanding |
943 |
requests. |
944 |
|
945 |
Under normal circumstances you don't need to call this function. |
946 |
|
947 |
IO::AIO::max_idle $nthreads |
948 |
Limit the number of threads (default: 4) that are allowed to idle |
949 |
(i.e., threads that did not get a request to process within 10 |
950 |
seconds). That means if a thread becomes idle while $nthreads other |
951 |
threads are also idle, it will free its resources and exit. |
952 |
|
953 |
This is useful when you allow a large number of threads (e.g. 100 or |
954 |
1000) to allow for extremely high load situations, but want to free |
955 |
resources under normal circumstances (1000 threads can easily |
956 |
consume 30MB of RAM). |
957 |
|
958 |
The default is probably ok in most situations, especially if thread |
959 |
creation is fast. If thread creation is very slow on your system you |
960 |
might want to use larger values. |
961 |
|
962 |
IO::AIO::max_outstanding $maxreqs |
963 |
This is a very bad function to use in interactive programs because |
964 |
it blocks, and a bad way to reduce concurrency because it is |
965 |
inexact: Better use an "aio_group" together with a feed callback. |
966 |
|
967 |
Sets the maximum number of outstanding requests to $nreqs. If you do |
968 |
queue up more than this number of requests, the next call to the |
969 |
"poll_cb" (and "poll_some" and other functions calling "poll_cb") |
970 |
function will block until the limit is no longer exceeded. |
971 |
|
972 |
The default value is very large, so there is no practical limit on |
973 |
the number of outstanding requests. |
974 |
|
975 |
You can still queue as many requests as you want. Therefore, |
976 |
"max_outstanding" is mainly useful in simple scripts (with low |
977 |
values) or as a stop gap to shield against fatal memory overflow |
978 |
(with large values). |
979 |
|
980 |
STATISTICAL INFORMATION |
981 |
IO::AIO::nreqs |
982 |
Returns the number of requests currently in the ready, execute or |
983 |
pending states (i.e. for which their callback has not been invoked |
984 |
yet). |
985 |
|
986 |
Example: wait till there are no outstanding requests anymore: |
987 |
|
988 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
989 |
while IO::AIO::nreqs; |
990 |
|
991 |
IO::AIO::nready |
992 |
Returns the number of requests currently in the ready state (not yet |
993 |
executed). |
994 |
|
995 |
IO::AIO::npending |
996 |
Returns the number of requests currently in the pending state |
997 |
(executed, but not yet processed by poll_cb). |
998 |
|
999 |
FORK BEHAVIOUR |
1000 |
This module should do "the right thing" when the process using it forks: |
1001 |
|
1002 |
Before the fork, IO::AIO enters a quiescent state where no requests can |
1003 |
be added in other threads and no results will be processed. After the |
1004 |
fork the parent simply leaves the quiescent state and continues |
1005 |
request/result processing, while the child frees the request/result |
1006 |
queue (so that the requests started before the fork will only be handled |
1007 |
in the parent). Threads will be started on demand until the limit set in |
1008 |
the parent process has been reached again. |
1009 |
|
1010 |
In short: the parent will, after a short pause, continue as if fork had |
1011 |
not been called, while the child will act as if IO::AIO has not been |
1012 |
used yet. |
1013 |
|
1014 |
MEMORY USAGE |
1015 |
Per-request usage: |
1016 |
|
1017 |
Each aio request uses - depending on your architecture - around 100-200 |
1018 |
bytes of memory. In addition, stat requests need a stat buffer (possibly |
1019 |
a few hundred bytes), readdir requires a result buffer and so on. Perl |
1020 |
scalars and other data passed into aio requests will also be locked and |
1021 |
will consume memory till the request has entered the done state. |
1022 |
|
1023 |
This is not awfully much, so queuing lots of requests is not usually a |
1024 |
problem. |
1025 |
|
1026 |
Per-thread usage: |
1027 |
|
1028 |
In the execution phase, some aio requests require more memory for |
1029 |
temporary buffers, and each thread requires a stack and other data |
1030 |
structures (usually around 16k-128k, depending on the OS). |
1031 |
|
1032 |
KNOWN BUGS |
1033 |
Known bugs will be fixed in the next release. |
1034 |
|
1035 |
SEE ALSO |
1036 |
AnyEvent::AIO for easy integration into event loops, Coro::AIO for a |
1037 |
more natural syntax. |
1038 |
|
1039 |
AUTHOR |
1040 |
Marc Lehmann <schmorp@schmorp.de> |
1041 |
http://home.schmorp.de/ |
1042 |
|