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 $aio_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. It is implemented as an interface to "libeio" |
55 |
(<http://software.schmorp.de/pkg/libeio.html>). |
56 |
|
57 |
Asynchronous means that operations that can normally block your program |
58 |
(e.g. reading from disk) will be done asynchronously: the operation will |
59 |
still block, but you can do something else in the meantime. This is |
60 |
extremely useful for programs that need to stay interactive even when |
61 |
doing heavy I/O (GUI programs, high performance network servers etc.), |
62 |
but can also be used to easily do operations in parallel that are |
63 |
normally done sequentially, e.g. stat'ing many files, which is much |
64 |
faster on a RAID volume or over NFS when you do a number of stat |
65 |
operations concurrently. |
66 |
|
67 |
While most of this works on all types of file descriptors (for example |
68 |
sockets), using these functions on file descriptors that support |
69 |
nonblocking operation (again, sockets, pipes etc.) is very inefficient. |
70 |
Use an event loop for that (such as the EV module): IO::AIO will |
71 |
naturally fit into such an event loop itself. |
72 |
|
73 |
In this version, a number of threads are started that execute your |
74 |
requests and signal their completion. You don't need thread support in |
75 |
perl, and the threads created by this module will not be visible to |
76 |
perl. In the future, this module might make use of the native aio |
77 |
functions available on many operating systems. However, they are often |
78 |
not well-supported or restricted (GNU/Linux doesn't allow them on normal |
79 |
files currently, for example), and they would only support aio_read and |
80 |
aio_write, so the remaining functionality would have to be implemented |
81 |
using threads anyway. |
82 |
|
83 |
Although the module will work in the presence of other (Perl-) threads, |
84 |
it is currently not reentrant in any way, so use appropriate locking |
85 |
yourself, always call "poll_cb" from within the same thread, or never |
86 |
call "poll_cb" (or other "aio_" functions) recursively. |
87 |
|
88 |
EXAMPLE |
89 |
This is a simple example that uses the EV module and loads /etc/passwd |
90 |
asynchronously: |
91 |
|
92 |
use Fcntl; |
93 |
use EV; |
94 |
use IO::AIO; |
95 |
|
96 |
# register the IO::AIO callback with EV |
97 |
my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb; |
98 |
|
99 |
# queue the request to open /etc/passwd |
100 |
aio_open "/etc/passwd", O_RDONLY, 0, sub { |
101 |
my $fh = shift |
102 |
or die "error while opening: $!"; |
103 |
|
104 |
# stat'ing filehandles is generally non-blocking |
105 |
my $size = -s $fh; |
106 |
|
107 |
# queue a request to read the file |
108 |
my $contents; |
109 |
aio_read $fh, 0, $size, $contents, 0, sub { |
110 |
$_[0] == $size |
111 |
or die "short read: $!"; |
112 |
|
113 |
close $fh; |
114 |
|
115 |
# file contents now in $contents |
116 |
print $contents; |
117 |
|
118 |
# exit event loop and program |
119 |
EV::unloop; |
120 |
}; |
121 |
}; |
122 |
|
123 |
# possibly queue up other requests, or open GUI windows, |
124 |
# check for sockets etc. etc. |
125 |
|
126 |
# process events as long as there are some: |
127 |
EV::loop; |
128 |
|
129 |
REQUEST ANATOMY AND LIFETIME |
130 |
Every "aio_*" function creates a request. which is a C data structure |
131 |
not directly visible to Perl. |
132 |
|
133 |
If called in non-void context, every request function returns a Perl |
134 |
object representing the request. In void context, nothing is returned, |
135 |
which saves a bit of memory. |
136 |
|
137 |
The perl object is a fairly standard ref-to-hash object. The hash |
138 |
contents are not used by IO::AIO so you are free to store anything you |
139 |
like in it. |
140 |
|
141 |
During their existance, aio requests travel through the following |
142 |
states, in order: |
143 |
|
144 |
ready |
145 |
Immediately after a request is created it is put into the ready |
146 |
state, waiting for a thread to execute it. |
147 |
|
148 |
execute |
149 |
A thread has accepted the request for processing and is currently |
150 |
executing it (e.g. blocking in read). |
151 |
|
152 |
pending |
153 |
The request has been executed and is waiting for result processing. |
154 |
|
155 |
While request submission and execution is fully asynchronous, result |
156 |
processing is not and relies on the perl interpreter calling |
157 |
"poll_cb" (or another function with the same effect). |
158 |
|
159 |
result |
160 |
The request results are processed synchronously by "poll_cb". |
161 |
|
162 |
The "poll_cb" function will process all outstanding aio requests by |
163 |
calling their callbacks, freeing memory associated with them and |
164 |
managing any groups they are contained in. |
165 |
|
166 |
done |
167 |
Request has reached the end of its lifetime and holds no resources |
168 |
anymore (except possibly for the Perl object, but its connection to |
169 |
the actual aio request is severed and calling its methods will |
170 |
either do nothing or result in a runtime error). |
171 |
|
172 |
FUNCTIONS |
173 |
AIO REQUEST FUNCTIONS |
174 |
All the "aio_*" calls are more or less thin wrappers around the syscall |
175 |
with the same name (sans "aio_"). The arguments are similar or |
176 |
identical, and they all accept an additional (and optional) $callback |
177 |
argument which must be a code reference. This code reference will get |
178 |
called with the syscall return code (e.g. most syscalls return -1 on |
179 |
error, unlike perl, which usually delivers "false") as its sole argument |
180 |
after the given syscall has been executed asynchronously. |
181 |
|
182 |
All functions expecting a filehandle keep a copy of the filehandle |
183 |
internally until the request has finished. |
184 |
|
185 |
All functions return request objects of type IO::AIO::REQ that allow |
186 |
further manipulation of those requests while they are in-flight. |
187 |
|
188 |
The pathnames you pass to these routines *must* be absolute and encoded |
189 |
as octets. The reason for the former is that at the time the request is |
190 |
being executed, the current working directory could have changed. |
191 |
Alternatively, you can make sure that you never change the current |
192 |
working directory anywhere in the program and then use relative paths. |
193 |
|
194 |
To encode pathnames as octets, either make sure you either: a) always |
195 |
pass in filenames you got from outside (command line, readdir etc.) |
196 |
without tinkering, b) are ASCII or ISO 8859-1, c) use the Encode module |
197 |
and encode your pathnames to the locale (or other) encoding in effect in |
198 |
the user environment, d) use Glib::filename_from_unicode on unicode |
199 |
filenames or e) use something else to ensure your scalar has the correct |
200 |
contents. |
201 |
|
202 |
This works, btw. independent of the internal UTF-8 bit, which IO::AIO |
203 |
handles correctly whether it is set or not. |
204 |
|
205 |
$prev_pri = aioreq_pri [$pri] |
206 |
Returns the priority value that would be used for the next request |
207 |
and, if $pri is given, sets the priority for the next aio request. |
208 |
|
209 |
The default priority is 0, the minimum and maximum priorities are -4 |
210 |
and 4, respectively. Requests with higher priority will be serviced |
211 |
first. |
212 |
|
213 |
The priority will be reset to 0 after each call to one of the |
214 |
"aio_*" functions. |
215 |
|
216 |
Example: open a file with low priority, then read something from it |
217 |
with higher priority so the read request is serviced before other |
218 |
low priority open requests (potentially spamming the cache): |
219 |
|
220 |
aioreq_pri -3; |
221 |
aio_open ..., sub { |
222 |
return unless $_[0]; |
223 |
|
224 |
aioreq_pri -2; |
225 |
aio_read $_[0], ..., sub { |
226 |
... |
227 |
}; |
228 |
}; |
229 |
|
230 |
aioreq_nice $pri_adjust |
231 |
Similar to "aioreq_pri", but subtracts the given value from the |
232 |
current priority, so the effect is cumulative. |
233 |
|
234 |
aio_open $pathname, $flags, $mode, $callback->($fh) |
235 |
Asynchronously open or create a file and call the callback with a |
236 |
newly created filehandle for the file. |
237 |
|
238 |
The pathname passed to "aio_open" must be absolute. See API NOTES, |
239 |
above, for an explanation. |
240 |
|
241 |
The $flags argument is a bitmask. See the "Fcntl" module for a list. |
242 |
They are the same as used by "sysopen". |
243 |
|
244 |
Likewise, $mode specifies the mode of the newly created file, if it |
245 |
didn't exist and "O_CREAT" has been given, just like perl's |
246 |
"sysopen", except that it is mandatory (i.e. use 0 if you don't |
247 |
create new files, and 0666 or 0777 if you do). Note that the $mode |
248 |
will be modified by the umask in effect then the request is being |
249 |
executed, so better never change the umask. |
250 |
|
251 |
Example: |
252 |
|
253 |
aio_open "/etc/passwd", O_RDONLY, 0, sub { |
254 |
if ($_[0]) { |
255 |
print "open successful, fh is $_[0]\n"; |
256 |
... |
257 |
} else { |
258 |
die "open failed: $!\n"; |
259 |
} |
260 |
}; |
261 |
|
262 |
aio_close $fh, $callback->($status) |
263 |
Asynchronously close a file and call the callback with the result |
264 |
code. |
265 |
|
266 |
Unfortunately, you can't do this to perl. Perl *insists* very |
267 |
strongly on closing the file descriptor associated with the |
268 |
filehandle itself. |
269 |
|
270 |
Therefore, "aio_close" will not close the filehandle - instead it |
271 |
will use dup2 to overwrite the file descriptor with the write-end of |
272 |
a pipe (the pipe fd will be created on demand and will be cached). |
273 |
|
274 |
Or in other words: the file descriptor will be closed, but it will |
275 |
not be free for reuse until the perl filehandle is closed. |
276 |
|
277 |
aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
278 |
aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
279 |
Reads or writes $length bytes from or to the specified $fh and |
280 |
$offset into the scalar given by $data and offset $dataoffset and |
281 |
calls the callback without the actual number of bytes read (or -1 on |
282 |
error, just like the syscall). |
283 |
|
284 |
"aio_read" will, like "sysread", shrink or grow the $data scalar to |
285 |
offset plus the actual number of bytes read. |
286 |
|
287 |
If $offset is undefined, then the current file descriptor offset |
288 |
will be used (and updated), otherwise the file descriptor offset |
289 |
will not be changed by these calls. |
290 |
|
291 |
If $length is undefined in "aio_write", use the remaining length of |
292 |
$data. |
293 |
|
294 |
If $dataoffset is less than zero, it will be counted from the end of |
295 |
$data. |
296 |
|
297 |
The $data scalar *MUST NOT* be modified in any way while the request |
298 |
is outstanding. Modifying it can result in segfaults or World War |
299 |
III (if the necessary/optional hardware is installed). |
300 |
|
301 |
Example: Read 15 bytes at offset 7 into scalar $buffer, starting at |
302 |
offset 0 within the scalar: |
303 |
|
304 |
aio_read $fh, 7, 15, $buffer, 0, sub { |
305 |
$_[0] > 0 or die "read error: $!"; |
306 |
print "read $_[0] bytes: <$buffer>\n"; |
307 |
}; |
308 |
|
309 |
aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval) |
310 |
Tries to copy $length bytes from $in_fh to $out_fh. It starts |
311 |
reading at byte offset $in_offset, and starts writing at the current |
312 |
file offset of $out_fh. Because of that, it is not safe to issue |
313 |
more than one "aio_sendfile" per $out_fh, as they will interfere |
314 |
with each other. |
315 |
|
316 |
This call tries to make use of a native "sendfile" syscall to |
317 |
provide zero-copy operation. For this to work, $out_fh should refer |
318 |
to a socket, and $in_fh should refer to mmap'able file. |
319 |
|
320 |
If the native sendfile call fails or is not implemented, it will be |
321 |
emulated, so you can call "aio_sendfile" on any type of filehandle |
322 |
regardless of the limitations of the operating system. |
323 |
|
324 |
Please note, however, that "aio_sendfile" can read more bytes from |
325 |
$in_fh than are written, and there is no way to find out how many |
326 |
bytes have been read from "aio_sendfile" alone, as "aio_sendfile" |
327 |
only provides the number of bytes written to $out_fh. Only if the |
328 |
result value equals $length one can assume that $length bytes have |
329 |
been read. |
330 |
|
331 |
aio_readahead $fh,$offset,$length, $callback->($retval) |
332 |
"aio_readahead" populates the page cache with data from a file so |
333 |
that subsequent reads from that file will not block on disk I/O. The |
334 |
$offset argument specifies the starting point from which data is to |
335 |
be read and $length specifies the number of bytes to be read. I/O is |
336 |
performed in whole pages, so that offset is effectively rounded down |
337 |
to a page boundary and bytes are read up to the next page boundary |
338 |
greater than or equal to (off-set+length). "aio_readahead" does not |
339 |
read beyond the end of the file. The current file offset of the file |
340 |
is left unchanged. |
341 |
|
342 |
If that syscall doesn't exist (likely if your OS isn't Linux) it |
343 |
will be emulated by simply reading the data, which would have a |
344 |
similar effect. |
345 |
|
346 |
aio_stat $fh_or_path, $callback->($status) |
347 |
aio_lstat $fh, $callback->($status) |
348 |
Works like perl's "stat" or "lstat" in void context. The callback |
349 |
will be called after the stat and the results will be available |
350 |
using "stat _" or "-s _" etc... |
351 |
|
352 |
The pathname passed to "aio_stat" must be absolute. See API NOTES, |
353 |
above, for an explanation. |
354 |
|
355 |
Currently, the stats are always 64-bit-stats, i.e. instead of |
356 |
returning an error when stat'ing a large file, the results will be |
357 |
silently truncated unless perl itself is compiled with large file |
358 |
support. |
359 |
|
360 |
Example: Print the length of /etc/passwd: |
361 |
|
362 |
aio_stat "/etc/passwd", sub { |
363 |
$_[0] and die "stat failed: $!"; |
364 |
print "size is ", -s _, "\n"; |
365 |
}; |
366 |
|
367 |
aio_utime $fh_or_path, $atime, $mtime, $callback->($status) |
368 |
Works like perl's "utime" function (including the special case of |
369 |
$atime and $mtime being undef). Fractional times are supported if |
370 |
the underlying syscalls support them. |
371 |
|
372 |
When called with a pathname, uses utimes(2) if available, otherwise |
373 |
utime(2). If called on a file descriptor, uses futimes(2) if |
374 |
available, otherwise returns ENOSYS, so this is not portable. |
375 |
|
376 |
Examples: |
377 |
|
378 |
# set atime and mtime to current time (basically touch(1)): |
379 |
aio_utime "path", undef, undef; |
380 |
# set atime to current time and mtime to beginning of the epoch: |
381 |
aio_utime "path", time, undef; # undef==0 |
382 |
|
383 |
aio_chown $fh_or_path, $uid, $gid, $callback->($status) |
384 |
Works like perl's "chown" function, except that "undef" for either |
385 |
$uid or $gid is being interpreted as "do not change" (but -1 can |
386 |
also be used). |
387 |
|
388 |
Examples: |
389 |
|
390 |
# same as "chown root path" in the shell: |
391 |
aio_chown "path", 0, -1; |
392 |
# same as above: |
393 |
aio_chown "path", 0, undef; |
394 |
|
395 |
aio_truncate $fh_or_path, $offset, $callback->($status) |
396 |
Works like truncate(2) or ftruncate(2). |
397 |
|
398 |
aio_chmod $fh_or_path, $mode, $callback->($status) |
399 |
Works like perl's "chmod" function. |
400 |
|
401 |
aio_unlink $pathname, $callback->($status) |
402 |
Asynchronously unlink (delete) a file and call the callback with the |
403 |
result code. |
404 |
|
405 |
aio_mknod $path, $mode, $dev, $callback->($status) |
406 |
[EXPERIMENTAL] |
407 |
|
408 |
Asynchronously create a device node (or fifo). See mknod(2). |
409 |
|
410 |
The only (POSIX-) portable way of calling this function is: |
411 |
|
412 |
aio_mknod $path, IO::AIO::S_IFIFO | $mode, 0, sub { ... |
413 |
|
414 |
aio_link $srcpath, $dstpath, $callback->($status) |
415 |
Asynchronously create a new link to the existing object at $srcpath |
416 |
at the path $dstpath and call the callback with the result code. |
417 |
|
418 |
aio_symlink $srcpath, $dstpath, $callback->($status) |
419 |
Asynchronously create a new symbolic link to the existing object at |
420 |
$srcpath at the path $dstpath and call the callback with the result |
421 |
code. |
422 |
|
423 |
aio_readlink $path, $callback->($link) |
424 |
Asynchronously read the symlink specified by $path and pass it to |
425 |
the callback. If an error occurs, nothing or undef gets passed to |
426 |
the callback. |
427 |
|
428 |
aio_rename $srcpath, $dstpath, $callback->($status) |
429 |
Asynchronously rename the object at $srcpath to $dstpath, just as |
430 |
rename(2) and call the callback with the result code. |
431 |
|
432 |
aio_mkdir $pathname, $mode, $callback->($status) |
433 |
Asynchronously mkdir (create) a directory and call the callback with |
434 |
the result code. $mode will be modified by the umask at the time the |
435 |
request is executed, so do not change your umask. |
436 |
|
437 |
aio_rmdir $pathname, $callback->($status) |
438 |
Asynchronously rmdir (delete) a directory and call the callback with |
439 |
the result code. |
440 |
|
441 |
aio_readdir $pathname, $callback->($entries) |
442 |
Unlike the POSIX call of the same name, "aio_readdir" reads an |
443 |
entire directory (i.e. opendir + readdir + closedir). The entries |
444 |
will not be sorted, and will NOT include the "." and ".." entries. |
445 |
|
446 |
The callback is passed a single argument which is either "undef" or |
447 |
an array-ref with the filenames. |
448 |
|
449 |
aio_readdirx $pathname, $flags, $callback->($entries, $flags) |
450 |
Quite similar to "aio_readdir", but the $flags argument allows to |
451 |
tune behaviour and output format. In case of an error, $entries will |
452 |
be "undef". |
453 |
|
454 |
The flags are a combination of the following constants, ORed |
455 |
together (the flags will also be passed to the callback, possibly |
456 |
modified): |
457 |
|
458 |
IO::AIO::READDIR_DENTS |
459 |
When this flag is off, then the callback gets an arrayref with |
460 |
of names only (as with "aio_readdir"), otherwise it gets an |
461 |
arrayref with "[$name, $type, $inode]" arrayrefs, each |
462 |
describing a single directory entry in more detail. |
463 |
|
464 |
$name is the name of the entry. |
465 |
|
466 |
$type is one of the "IO::AIO::DT_xxx" constants: |
467 |
|
468 |
"IO::AIO::DT_UNKNOWN", "IO::AIO::DT_FIFO", "IO::AIO::DT_CHR", |
469 |
"IO::AIO::DT_DIR", "IO::AIO::DT_BLK", "IO::AIO::DT_REG", |
470 |
"IO::AIO::DT_LNK", "IO::AIO::DT_SOCK", "IO::AIO::DT_WHT". |
471 |
|
472 |
"IO::AIO::DT_UNKNOWN" means just that: readdir does not know. If |
473 |
you need to know, you have to run stat yourself. Also, for speed |
474 |
reasons, the $type scalars are read-only: you can not modify |
475 |
them. |
476 |
|
477 |
$inode is the inode number (which might not be exact on systems |
478 |
with 64 bit inode numbers and 32 bit perls). This field has |
479 |
unspecified content on systems that do not deliver the inode |
480 |
information. |
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 |
Future versions of this function might fall back to other methods |
637 |
when "fsync" on the directory fails (such as calling "sync"). |
638 |
|
639 |
Passes 0 when everything went ok, and -1 on error. |
640 |
|
641 |
aio_group $callback->(...) |
642 |
This is a very special aio request: Instead of doing something, it |
643 |
is a container for other aio requests, which is useful if you want |
644 |
to bundle many requests into a single, composite, request with a |
645 |
definite callback and the ability to cancel the whole request with |
646 |
its subrequests. |
647 |
|
648 |
Returns an object of class IO::AIO::GRP. See its documentation below |
649 |
for more info. |
650 |
|
651 |
Example: |
652 |
|
653 |
my $grp = aio_group sub { |
654 |
print "all stats done\n"; |
655 |
}; |
656 |
|
657 |
add $grp |
658 |
(aio_stat ...), |
659 |
(aio_stat ...), |
660 |
...; |
661 |
|
662 |
aio_nop $callback->() |
663 |
This is a special request - it does nothing in itself and is only |
664 |
used for side effects, such as when you want to add a dummy request |
665 |
to a group so that finishing the requests in the group depends on |
666 |
executing the given code. |
667 |
|
668 |
While this request does nothing, it still goes through the execution |
669 |
phase and still requires a worker thread. Thus, the callback will |
670 |
not be executed immediately but only after other requests in the |
671 |
queue have entered their execution phase. This can be used to |
672 |
measure request latency. |
673 |
|
674 |
IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED* |
675 |
Mainly used for debugging and benchmarking, this aio request puts |
676 |
one of the request workers to sleep for the given time. |
677 |
|
678 |
While it is theoretically handy to have simple I/O scheduling |
679 |
requests like sleep and file handle readable/writable, the overhead |
680 |
this creates is immense (it blocks a thread for a long time) so do |
681 |
not use this function except to put your application under |
682 |
artificial I/O pressure. |
683 |
|
684 |
IO::AIO::REQ CLASS |
685 |
All non-aggregate "aio_*" functions return an object of this class when |
686 |
called in non-void context. |
687 |
|
688 |
cancel $req |
689 |
Cancels the request, if possible. Has the effect of skipping |
690 |
execution when entering the execute state and skipping calling the |
691 |
callback when entering the the result state, but will leave the |
692 |
request otherwise untouched (with the exception of readdir). That |
693 |
means that requests that currently execute will not be stopped and |
694 |
resources held by the request will not be freed prematurely. |
695 |
|
696 |
cb $req $callback->(...) |
697 |
Replace (or simply set) the callback registered to the request. |
698 |
|
699 |
IO::AIO::GRP CLASS |
700 |
This class is a subclass of IO::AIO::REQ, so all its methods apply to |
701 |
objects of this class, too. |
702 |
|
703 |
A IO::AIO::GRP object is a special request that can contain multiple |
704 |
other aio requests. |
705 |
|
706 |
You create one by calling the "aio_group" constructing function with a |
707 |
callback that will be called when all contained requests have entered |
708 |
the "done" state: |
709 |
|
710 |
my $grp = aio_group sub { |
711 |
print "all requests are done\n"; |
712 |
}; |
713 |
|
714 |
You add requests by calling the "add" method with one or more |
715 |
"IO::AIO::REQ" objects: |
716 |
|
717 |
$grp->add (aio_unlink "..."); |
718 |
|
719 |
add $grp aio_stat "...", sub { |
720 |
$_[0] or return $grp->result ("error"); |
721 |
|
722 |
# add another request dynamically, if first succeeded |
723 |
add $grp aio_open "...", sub { |
724 |
$grp->result ("ok"); |
725 |
}; |
726 |
}; |
727 |
|
728 |
This makes it very easy to create composite requests (see the source of |
729 |
"aio_move" for an application) that work and feel like simple requests. |
730 |
|
731 |
* The IO::AIO::GRP objects will be cleaned up during calls to |
732 |
"IO::AIO::poll_cb", just like any other request. |
733 |
|
734 |
* They can be canceled like any other request. Canceling will cancel |
735 |
not only the request itself, but also all requests it contains. |
736 |
|
737 |
* They can also can also be added to other IO::AIO::GRP objects. |
738 |
|
739 |
* You must not add requests to a group from within the group callback |
740 |
(or any later time). |
741 |
|
742 |
Their lifetime, simplified, looks like this: when they are empty, they |
743 |
will finish very quickly. If they contain only requests that are in the |
744 |
"done" state, they will also finish. Otherwise they will continue to |
745 |
exist. |
746 |
|
747 |
That means after creating a group you have some time to add requests |
748 |
(precisely before the callback has been invoked, which is only done |
749 |
within the "poll_cb"). And in the callbacks of those requests, you can |
750 |
add further requests to the group. And only when all those requests have |
751 |
finished will the the group itself finish. |
752 |
|
753 |
add $grp ... |
754 |
$grp->add (...) |
755 |
Add one or more requests to the group. Any type of IO::AIO::REQ can |
756 |
be added, including other groups, as long as you do not create |
757 |
circular dependencies. |
758 |
|
759 |
Returns all its arguments. |
760 |
|
761 |
$grp->cancel_subs |
762 |
Cancel all subrequests and clears any feeder, but not the group |
763 |
request itself. Useful when you queued a lot of events but got a |
764 |
result early. |
765 |
|
766 |
$grp->result (...) |
767 |
Set the result value(s) that will be passed to the group callback |
768 |
when all subrequests have finished and set the groups errno to the |
769 |
current value of errno (just like calling "errno" without an error |
770 |
number). By default, no argument will be passed and errno is zero. |
771 |
|
772 |
$grp->errno ([$errno]) |
773 |
Sets the group errno value to $errno, or the current value of errno |
774 |
when the argument is missing. |
775 |
|
776 |
Every aio request has an associated errno value that is restored |
777 |
when the callback is invoked. This method lets you change this value |
778 |
from its default (0). |
779 |
|
780 |
Calling "result" will also set errno, so make sure you either set $! |
781 |
before the call to "result", or call c<errno> after it. |
782 |
|
783 |
feed $grp $callback->($grp) |
784 |
Sets a feeder/generator on this group: every group can have an |
785 |
attached generator that generates requests if idle. The idea behind |
786 |
this is that, although you could just queue as many requests as you |
787 |
want in a group, this might starve other requests for a potentially |
788 |
long time. For example, "aio_scandir" might generate hundreds of |
789 |
thousands "aio_stat" requests, delaying any later requests for a |
790 |
long time. |
791 |
|
792 |
To avoid this, and allow incremental generation of requests, you can |
793 |
instead a group and set a feeder on it that generates those |
794 |
requests. The feed callback will be called whenever there are few |
795 |
enough (see "limit", below) requests active in the group itself and |
796 |
is expected to queue more requests. |
797 |
|
798 |
The feed callback can queue as many requests as it likes (i.e. "add" |
799 |
does not impose any limits). |
800 |
|
801 |
If the feed does not queue more requests when called, it will be |
802 |
automatically removed from the group. |
803 |
|
804 |
If the feed limit is 0 when this method is called, it will be set to |
805 |
2 automatically. |
806 |
|
807 |
Example: |
808 |
|
809 |
# stat all files in @files, but only ever use four aio requests concurrently: |
810 |
|
811 |
my $grp = aio_group sub { print "finished\n" }; |
812 |
limit $grp 4; |
813 |
feed $grp sub { |
814 |
my $file = pop @files |
815 |
or return; |
816 |
|
817 |
add $grp aio_stat $file, sub { ... }; |
818 |
}; |
819 |
|
820 |
limit $grp $num |
821 |
Sets the feeder limit for the group: The feeder will be called |
822 |
whenever the group contains less than this many requests. |
823 |
|
824 |
Setting the limit to 0 will pause the feeding process. |
825 |
|
826 |
The default value for the limit is 0, but note that setting a feeder |
827 |
automatically bumps it up to 2. |
828 |
|
829 |
SUPPORT FUNCTIONS |
830 |
EVENT PROCESSING AND EVENT LOOP INTEGRATION |
831 |
$fileno = IO::AIO::poll_fileno |
832 |
Return the *request result pipe file descriptor*. This filehandle |
833 |
must be polled for reading by some mechanism outside this module |
834 |
(e.g. EV, Glib, select and so on, see below or the SYNOPSIS). If the |
835 |
pipe becomes readable you have to call "poll_cb" to check the |
836 |
results. |
837 |
|
838 |
See "poll_cb" for an example. |
839 |
|
840 |
IO::AIO::poll_cb |
841 |
Process some outstanding events on the result pipe. You have to call |
842 |
this regularly. Returns 0 if all events could be processed, or -1 if |
843 |
it returned earlier for whatever reason. Returns immediately when no |
844 |
events are outstanding. The amount of events processed depends on |
845 |
the settings of "IO::AIO::max_poll_req" and |
846 |
"IO::AIO::max_poll_time". |
847 |
|
848 |
If not all requests were processed for whatever reason, the |
849 |
filehandle will still be ready when "poll_cb" returns, so normally |
850 |
you don't have to do anything special to have it called later. |
851 |
|
852 |
Example: Install an Event watcher that automatically calls |
853 |
IO::AIO::poll_cb with high priority (more examples can be found in |
854 |
the SYNOPSIS section, at the top of this document): |
855 |
|
856 |
Event->io (fd => IO::AIO::poll_fileno, |
857 |
poll => 'r', async => 1, |
858 |
cb => \&IO::AIO::poll_cb); |
859 |
|
860 |
IO::AIO::max_poll_reqs $nreqs |
861 |
IO::AIO::max_poll_time $seconds |
862 |
These set the maximum number of requests (default 0, meaning |
863 |
infinity) that are being processed by "IO::AIO::poll_cb" in one |
864 |
call, respectively the maximum amount of time (default 0, meaning |
865 |
infinity) spent in "IO::AIO::poll_cb" to process requests (more |
866 |
correctly the mininum amount of time "poll_cb" is allowed to use). |
867 |
|
868 |
Setting "max_poll_time" to a non-zero value creates an overhead of |
869 |
one syscall per request processed, which is not normally a problem |
870 |
unless your callbacks are really really fast or your OS is really |
871 |
really slow (I am not mentioning Solaris here). Using |
872 |
"max_poll_reqs" incurs no overhead. |
873 |
|
874 |
Setting these is useful if you want to ensure some level of |
875 |
interactiveness when perl is not fast enough to process all requests |
876 |
in time. |
877 |
|
878 |
For interactive programs, values such as 0.01 to 0.1 should be fine. |
879 |
|
880 |
Example: Install an Event watcher that automatically calls |
881 |
IO::AIO::poll_cb with low priority, to ensure that other parts of |
882 |
the program get the CPU sometimes even under high AIO load. |
883 |
|
884 |
# try not to spend much more than 0.1s in poll_cb |
885 |
IO::AIO::max_poll_time 0.1; |
886 |
|
887 |
# use a low priority so other tasks have priority |
888 |
Event->io (fd => IO::AIO::poll_fileno, |
889 |
poll => 'r', nice => 1, |
890 |
cb => &IO::AIO::poll_cb); |
891 |
|
892 |
IO::AIO::poll_wait |
893 |
If there are any outstanding requests and none of them in the result |
894 |
phase, wait till the result filehandle becomes ready for reading |
895 |
(simply does a "select" on the filehandle. This is useful if you |
896 |
want to synchronously wait for some requests to finish). |
897 |
|
898 |
See "nreqs" for an example. |
899 |
|
900 |
IO::AIO::poll |
901 |
Waits until some requests have been handled. |
902 |
|
903 |
Returns the number of requests processed, but is otherwise strictly |
904 |
equivalent to: |
905 |
|
906 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
907 |
|
908 |
IO::AIO::flush |
909 |
Wait till all outstanding AIO requests have been handled. |
910 |
|
911 |
Strictly equivalent to: |
912 |
|
913 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
914 |
while IO::AIO::nreqs; |
915 |
|
916 |
CONTROLLING THE NUMBER OF THREADS |
917 |
IO::AIO::min_parallel $nthreads |
918 |
Set the minimum number of AIO threads to $nthreads. The current |
919 |
default is 8, which means eight asynchronous operations can execute |
920 |
concurrently at any one time (the number of outstanding requests, |
921 |
however, is unlimited). |
922 |
|
923 |
IO::AIO starts threads only on demand, when an AIO request is queued |
924 |
and no free thread exists. Please note that queueing up a hundred |
925 |
requests can create demand for a hundred threads, even if it turns |
926 |
out that everything is in the cache and could have been processed |
927 |
faster by a single thread. |
928 |
|
929 |
It is recommended to keep the number of threads relatively low, as |
930 |
some Linux kernel versions will scale negatively with the number of |
931 |
threads (higher parallelity => MUCH higher latency). With current |
932 |
Linux 2.6 versions, 4-32 threads should be fine. |
933 |
|
934 |
Under most circumstances you don't need to call this function, as |
935 |
the module selects a default that is suitable for low to moderate |
936 |
load. |
937 |
|
938 |
IO::AIO::max_parallel $nthreads |
939 |
Sets the maximum number of AIO threads to $nthreads. If more than |
940 |
the specified number of threads are currently running, this function |
941 |
kills them. This function blocks until the limit is reached. |
942 |
|
943 |
While $nthreads are zero, aio requests get queued but not executed |
944 |
until the number of threads has been increased again. |
945 |
|
946 |
This module automatically runs "max_parallel 0" at program end, to |
947 |
ensure that all threads are killed and that there are no outstanding |
948 |
requests. |
949 |
|
950 |
Under normal circumstances you don't need to call this function. |
951 |
|
952 |
IO::AIO::max_idle $nthreads |
953 |
Limit the number of threads (default: 4) that are allowed to idle |
954 |
(i.e., threads that did not get a request to process within 10 |
955 |
seconds). That means if a thread becomes idle while $nthreads other |
956 |
threads are also idle, it will free its resources and exit. |
957 |
|
958 |
This is useful when you allow a large number of threads (e.g. 100 or |
959 |
1000) to allow for extremely high load situations, but want to free |
960 |
resources under normal circumstances (1000 threads can easily |
961 |
consume 30MB of RAM). |
962 |
|
963 |
The default is probably ok in most situations, especially if thread |
964 |
creation is fast. If thread creation is very slow on your system you |
965 |
might want to use larger values. |
966 |
|
967 |
IO::AIO::max_outstanding $maxreqs |
968 |
This is a very bad function to use in interactive programs because |
969 |
it blocks, and a bad way to reduce concurrency because it is |
970 |
inexact: Better use an "aio_group" together with a feed callback. |
971 |
|
972 |
Sets the maximum number of outstanding requests to $nreqs. If you do |
973 |
queue up more than this number of requests, the next call to the |
974 |
"poll_cb" (and "poll_some" and other functions calling "poll_cb") |
975 |
function will block until the limit is no longer exceeded. |
976 |
|
977 |
The default value is very large, so there is no practical limit on |
978 |
the number of outstanding requests. |
979 |
|
980 |
You can still queue as many requests as you want. Therefore, |
981 |
"max_outstanding" is mainly useful in simple scripts (with low |
982 |
values) or as a stop gap to shield against fatal memory overflow |
983 |
(with large values). |
984 |
|
985 |
STATISTICAL INFORMATION |
986 |
IO::AIO::nreqs |
987 |
Returns the number of requests currently in the ready, execute or |
988 |
pending states (i.e. for which their callback has not been invoked |
989 |
yet). |
990 |
|
991 |
Example: wait till there are no outstanding requests anymore: |
992 |
|
993 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
994 |
while IO::AIO::nreqs; |
995 |
|
996 |
IO::AIO::nready |
997 |
Returns the number of requests currently in the ready state (not yet |
998 |
executed). |
999 |
|
1000 |
IO::AIO::npending |
1001 |
Returns the number of requests currently in the pending state |
1002 |
(executed, but not yet processed by poll_cb). |
1003 |
|
1004 |
MISCELLANEOUS FUNCTIONS |
1005 |
IO::AIO implements some functions that might be useful, but are not |
1006 |
asynchronous. |
1007 |
|
1008 |
IO::AIO::sendfile $ofh, $ifh, $offset, $count |
1009 |
Calls the "eio_sendfile_sync" function, which is like |
1010 |
"aio_sendfile", but is blocking (this makes most sense if you know |
1011 |
the input data is likely cached already and the output filehandle is |
1012 |
set to non-blocking operations). |
1013 |
|
1014 |
Returns the number of bytes copied, or -1 on error. |
1015 |
|
1016 |
IO::AIO::fadvise $fh, $offset, $len, $advice |
1017 |
Simply calls the "posix_fadvise" function (see it's manpage for |
1018 |
details). The following advice constants are avaiable: |
1019 |
"IO::AIO::FADV_NORMAL", "IO::AIO::FADV_SEQUENTIAL", |
1020 |
"IO::AIO::FADV_RANDOM", "IO::AIO::FADV_NOREUSE", |
1021 |
"IO::AIO::FADV_WILLNEED", "IO::AIO::FADV_DONTNEED". |
1022 |
|
1023 |
On systems that do not implement "posix_fadvise", this function |
1024 |
returns ENOSYS, otherwise the return value of "posix_fadvise". |
1025 |
|
1026 |
FORK BEHAVIOUR |
1027 |
This module should do "the right thing" when the process using it forks: |
1028 |
|
1029 |
Before the fork, IO::AIO enters a quiescent state where no requests can |
1030 |
be added in other threads and no results will be processed. After the |
1031 |
fork the parent simply leaves the quiescent state and continues |
1032 |
request/result processing, while the child frees the request/result |
1033 |
queue (so that the requests started before the fork will only be handled |
1034 |
in the parent). Threads will be started on demand until the limit set in |
1035 |
the parent process has been reached again. |
1036 |
|
1037 |
In short: the parent will, after a short pause, continue as if fork had |
1038 |
not been called, while the child will act as if IO::AIO has not been |
1039 |
used yet. |
1040 |
|
1041 |
MEMORY USAGE |
1042 |
Per-request usage: |
1043 |
|
1044 |
Each aio request uses - depending on your architecture - around 100-200 |
1045 |
bytes of memory. In addition, stat requests need a stat buffer (possibly |
1046 |
a few hundred bytes), readdir requires a result buffer and so on. Perl |
1047 |
scalars and other data passed into aio requests will also be locked and |
1048 |
will consume memory till the request has entered the done state. |
1049 |
|
1050 |
This is not awfully much, so queuing lots of requests is not usually a |
1051 |
problem. |
1052 |
|
1053 |
Per-thread usage: |
1054 |
|
1055 |
In the execution phase, some aio requests require more memory for |
1056 |
temporary buffers, and each thread requires a stack and other data |
1057 |
structures (usually around 16k-128k, depending on the OS). |
1058 |
|
1059 |
KNOWN BUGS |
1060 |
Known bugs will be fixed in the next release. |
1061 |
|
1062 |
SEE ALSO |
1063 |
AnyEvent::AIO for easy integration into event loops, Coro::AIO for a |
1064 |
more natural syntax. |
1065 |
|
1066 |
AUTHOR |
1067 |
Marc Lehmann <schmorp@schmorp.de> |
1068 |
http://home.schmorp.de/ |
1069 |
|