1 |
root |
1.1 |
NAME |
2 |
|
|
IO::AIO - Asynchronous Input/Output |
3 |
|
|
|
4 |
|
|
SYNOPSIS |
5 |
|
|
use IO::AIO; |
6 |
|
|
|
7 |
root |
1.5 |
aio_open "/etc/passwd", O_RDONLY, 0, sub { |
8 |
|
|
my ($fh) = @_; |
9 |
|
|
... |
10 |
|
|
}; |
11 |
|
|
|
12 |
|
|
aio_unlink "/tmp/file", sub { }; |
13 |
|
|
|
14 |
|
|
aio_read $fh, 30000, 1024, $buffer, 0, sub { |
15 |
|
|
$_[0] > 0 or die "read error: $!"; |
16 |
|
|
}; |
17 |
|
|
|
18 |
root |
1.18 |
# version 2+ has request and group objects |
19 |
|
|
use IO::AIO 2; |
20 |
|
|
|
21 |
|
|
aioreq_pri 4; # give next request a very high priority |
22 |
|
|
my $req = aio_unlink "/tmp/file", sub { }; |
23 |
|
|
$req->cancel; # cancel request if still in queue |
24 |
|
|
|
25 |
|
|
my $grp = aio_group sub { print "all stats done\n" }; |
26 |
|
|
add $grp aio_stat "..." for ...; |
27 |
|
|
|
28 |
|
|
# AnyEvent integration |
29 |
root |
1.14 |
open my $fh, "<&=" . IO::AIO::poll_fileno or die "$!"; |
30 |
|
|
my $w = AnyEvent->io (fh => $fh, poll => 'r', cb => sub { IO::AIO::poll_cb }); |
31 |
|
|
|
32 |
root |
1.18 |
# Event integration |
33 |
root |
1.5 |
Event->io (fd => IO::AIO::poll_fileno, |
34 |
|
|
poll => 'r', |
35 |
|
|
cb => \&IO::AIO::poll_cb); |
36 |
|
|
|
37 |
root |
1.18 |
# Glib/Gtk2 integration |
38 |
root |
1.5 |
add_watch Glib::IO IO::AIO::poll_fileno, |
39 |
root |
1.7 |
in => sub { IO::AIO::poll_cb; 1 }; |
40 |
root |
1.5 |
|
41 |
root |
1.18 |
# Tk integration |
42 |
root |
1.5 |
Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "", |
43 |
|
|
readable => \&IO::AIO::poll_cb); |
44 |
|
|
|
45 |
root |
1.18 |
# Danga::Socket integration |
46 |
root |
1.6 |
Danga::Socket->AddOtherFds (IO::AIO::poll_fileno => |
47 |
|
|
\&IO::AIO::poll_cb); |
48 |
|
|
|
49 |
root |
1.1 |
DESCRIPTION |
50 |
|
|
This module implements asynchronous I/O using whatever means your |
51 |
root |
1.2 |
operating system supports. |
52 |
root |
1.1 |
|
53 |
root |
1.19 |
Asynchronous means that operations that can normally block your program |
54 |
|
|
(e.g. reading from disk) will be done asynchronously: the operation will |
55 |
|
|
still block, but you can do something else in the meantime. This is |
56 |
|
|
extremely useful for programs that need to stay interactive even when |
57 |
|
|
doing heavy I/O (GUI programs, high performance network servers etc.), |
58 |
|
|
but can also be used to easily do operations in parallel that are |
59 |
|
|
normally done sequentially, e.g. stat'ing many files, which is much |
60 |
|
|
faster on a RAID volume or over NFS when you do a number of stat |
61 |
|
|
operations concurrently. |
62 |
|
|
|
63 |
root |
1.20 |
While most of this works on all types of file descriptors (for example |
64 |
|
|
sockets), using these functions on file descriptors that support |
65 |
|
|
nonblocking operation (again, sockets, pipes etc.) is very inefficient |
66 |
|
|
or might not work (aio_read fails on sockets/pipes/fifos). Use an event |
67 |
root |
1.19 |
loop for that (such as the Event module): IO::AIO will naturally fit |
68 |
|
|
into such an event loop itself. |
69 |
|
|
|
70 |
root |
1.18 |
In this version, a number of threads are started that execute your |
71 |
|
|
requests and signal their completion. You don't need thread support in |
72 |
|
|
perl, and the threads created by this module will not be visible to |
73 |
|
|
perl. In the future, this module might make use of the native aio |
74 |
|
|
functions available on many operating systems. However, they are often |
75 |
root |
1.19 |
not well-supported or restricted (GNU/Linux doesn't allow them on normal |
76 |
root |
1.18 |
files currently, for example), and they would only support aio_read and |
77 |
root |
1.2 |
aio_write, so the remaining functionality would have to be implemented |
78 |
|
|
using threads anyway. |
79 |
root |
1.1 |
|
80 |
root |
1.18 |
Although the module will work with in the presence of other (Perl-) |
81 |
|
|
threads, it is currently not reentrant in any way, so use appropriate |
82 |
|
|
locking yourself, always call "poll_cb" from within the same thread, or |
83 |
|
|
never call "poll_cb" (or other "aio_" functions) recursively. |
84 |
|
|
|
85 |
root |
1.19 |
EXAMPLE |
86 |
|
|
This is a simple example that uses the Event module and loads |
87 |
|
|
/etc/passwd asynchronously: |
88 |
|
|
|
89 |
|
|
use Fcntl; |
90 |
|
|
use Event; |
91 |
|
|
use IO::AIO; |
92 |
|
|
|
93 |
|
|
# register the IO::AIO callback with Event |
94 |
|
|
Event->io (fd => IO::AIO::poll_fileno, |
95 |
|
|
poll => 'r', |
96 |
|
|
cb => \&IO::AIO::poll_cb); |
97 |
|
|
|
98 |
|
|
# queue the request to open /etc/passwd |
99 |
|
|
aio_open "/etc/passwd", O_RDONLY, 0, sub { |
100 |
|
|
my $fh = $_[0] |
101 |
|
|
or die "error while opening: $!"; |
102 |
|
|
|
103 |
|
|
# stat'ing filehandles is generally non-blocking |
104 |
|
|
my $size = -s $fh; |
105 |
|
|
|
106 |
|
|
# queue a request to read the file |
107 |
|
|
my $contents; |
108 |
|
|
aio_read $fh, 0, $size, $contents, 0, sub { |
109 |
|
|
$_[0] == $size |
110 |
|
|
or die "short read: $!"; |
111 |
|
|
|
112 |
|
|
close $fh; |
113 |
|
|
|
114 |
|
|
# file contents now in $contents |
115 |
|
|
print $contents; |
116 |
|
|
|
117 |
|
|
# exit event loop and program |
118 |
|
|
Event::unloop; |
119 |
|
|
}; |
120 |
|
|
}; |
121 |
|
|
|
122 |
|
|
# possibly queue up other requests, or open GUI windows, |
123 |
|
|
# check for sockets etc. etc. |
124 |
|
|
|
125 |
|
|
# process events as long as there are some: |
126 |
|
|
Event::loop; |
127 |
|
|
|
128 |
root |
1.18 |
REQUEST ANATOMY AND LIFETIME |
129 |
|
|
Every "aio_*" function creates a request. which is a C data structure |
130 |
|
|
not directly visible to Perl. |
131 |
|
|
|
132 |
|
|
If called in non-void context, every request function returns a Perl |
133 |
|
|
object representing the request. In void context, nothing is returned, |
134 |
|
|
which saves a bit of memory. |
135 |
|
|
|
136 |
|
|
The perl object is a fairly standard ref-to-hash object. The hash |
137 |
|
|
contents are not used by IO::AIO so you are free to store anything you |
138 |
|
|
like in it. |
139 |
|
|
|
140 |
|
|
During their existance, aio requests travel through the following |
141 |
|
|
states, in order: |
142 |
|
|
|
143 |
|
|
ready |
144 |
|
|
Immediately after a request is created it is put into the ready |
145 |
|
|
state, waiting for a thread to execute it. |
146 |
|
|
|
147 |
|
|
execute |
148 |
|
|
A thread has accepted the request for processing and is currently |
149 |
|
|
executing it (e.g. blocking in read). |
150 |
|
|
|
151 |
|
|
pending |
152 |
|
|
The request has been executed and is waiting for result processing. |
153 |
|
|
|
154 |
|
|
While request submission and execution is fully asynchronous, result |
155 |
|
|
processing is not and relies on the perl interpreter calling |
156 |
|
|
"poll_cb" (or another function with the same effect). |
157 |
|
|
|
158 |
|
|
result |
159 |
|
|
The request results are processed synchronously by "poll_cb". |
160 |
|
|
|
161 |
|
|
The "poll_cb" function will process all outstanding aio requests by |
162 |
|
|
calling their callbacks, freeing memory associated with them and |
163 |
|
|
managing any groups they are contained in. |
164 |
|
|
|
165 |
|
|
done |
166 |
|
|
Request has reached the end of its lifetime and holds no resources |
167 |
|
|
anymore (except possibly for the Perl object, but its connection to |
168 |
|
|
the actual aio request is severed and calling its methods will |
169 |
|
|
either do nothing or result in a runtime error). |
170 |
root |
1.1 |
|
171 |
root |
1.4 |
FUNCTIONS |
172 |
root |
1.19 |
AIO REQUEST FUNCTIONS |
173 |
root |
1.20 |
All the "aio_*" calls are more or less thin wrappers around the syscall |
174 |
|
|
with the same name (sans "aio_"). The arguments are similar or |
175 |
|
|
identical, and they all accept an additional (and optional) $callback |
176 |
|
|
argument which must be a code reference. This code reference will get |
177 |
|
|
called with the syscall return code (e.g. most syscalls return -1 on |
178 |
|
|
error, unlike perl, which usually delivers "false") as it's sole |
179 |
|
|
argument when the given syscall has been executed asynchronously. |
180 |
|
|
|
181 |
|
|
All functions expecting a filehandle keep a copy of the filehandle |
182 |
|
|
internally until the request has finished. |
183 |
|
|
|
184 |
|
|
All functions return request objects of type IO::AIO::REQ that allow |
185 |
|
|
further manipulation of those requests while they are in-flight. |
186 |
|
|
|
187 |
|
|
The pathnames you pass to these routines *must* be absolute and encoded |
188 |
|
|
as octets. The reason for the former is that at the time the request is |
189 |
|
|
being executed, the current working directory could have changed. |
190 |
|
|
Alternatively, you can make sure that you never change the current |
191 |
|
|
working directory anywhere in the program and then use relative paths. |
192 |
|
|
|
193 |
|
|
To encode pathnames as octets, either make sure you either: a) always |
194 |
|
|
pass in filenames you got from outside (command line, readdir etc.) |
195 |
|
|
without tinkering, b) are ASCII or ISO 8859-1, c) use the Encode module |
196 |
|
|
and encode your pathnames to the locale (or other) encoding in effect in |
197 |
|
|
the user environment, d) use Glib::filename_from_unicode on unicode |
198 |
|
|
filenames or e) use something else to ensure your scalar has the correct |
199 |
|
|
contents. |
200 |
|
|
|
201 |
|
|
This works, btw. independent of the internal UTF-8 bit, which IO::AIO |
202 |
|
|
handles correctly wether it is set or not. |
203 |
|
|
|
204 |
|
|
$prev_pri = aioreq_pri [$pri] |
205 |
|
|
Returns the priority value that would be used for the next request |
206 |
|
|
and, if $pri is given, sets the priority for the next aio request. |
207 |
|
|
|
208 |
|
|
The default priority is 0, the minimum and maximum priorities are -4 |
209 |
|
|
and 4, respectively. Requests with higher priority will be serviced |
210 |
|
|
first. |
211 |
|
|
|
212 |
|
|
The priority will be reset to 0 after each call to one of the |
213 |
|
|
"aio_*" functions. |
214 |
|
|
|
215 |
|
|
Example: open a file with low priority, then read something from it |
216 |
|
|
with higher priority so the read request is serviced before other |
217 |
|
|
low priority open requests (potentially spamming the cache): |
218 |
|
|
|
219 |
|
|
aioreq_pri -3; |
220 |
|
|
aio_open ..., sub { |
221 |
|
|
return unless $_[0]; |
222 |
|
|
|
223 |
|
|
aioreq_pri -2; |
224 |
|
|
aio_read $_[0], ..., sub { |
225 |
|
|
... |
226 |
|
|
}; |
227 |
|
|
}; |
228 |
|
|
|
229 |
|
|
aioreq_nice $pri_adjust |
230 |
|
|
Similar to "aioreq_pri", but subtracts the given value from the |
231 |
|
|
current priority, so the effect is cumulative. |
232 |
|
|
|
233 |
|
|
aio_open $pathname, $flags, $mode, $callback->($fh) |
234 |
|
|
Asynchronously open or create a file and call the callback with a |
235 |
|
|
newly created filehandle for the file. |
236 |
|
|
|
237 |
|
|
The pathname passed to "aio_open" must be absolute. See API NOTES, |
238 |
|
|
above, for an explanation. |
239 |
|
|
|
240 |
|
|
The $flags argument is a bitmask. See the "Fcntl" module for a list. |
241 |
|
|
They are the same as used by "sysopen". |
242 |
|
|
|
243 |
|
|
Likewise, $mode specifies the mode of the newly created file, if it |
244 |
|
|
didn't exist and "O_CREAT" has been given, just like perl's |
245 |
|
|
"sysopen", except that it is mandatory (i.e. use 0 if you don't |
246 |
|
|
create new files, and 0666 or 0777 if you do). |
247 |
|
|
|
248 |
|
|
Example: |
249 |
|
|
|
250 |
|
|
aio_open "/etc/passwd", O_RDONLY, 0, sub { |
251 |
|
|
if ($_[0]) { |
252 |
|
|
print "open successful, fh is $_[0]\n"; |
253 |
|
|
... |
254 |
|
|
} else { |
255 |
|
|
die "open failed: $!\n"; |
256 |
|
|
} |
257 |
|
|
}; |
258 |
|
|
|
259 |
|
|
aio_close $fh, $callback->($status) |
260 |
|
|
Asynchronously close a file and call the callback with the result |
261 |
|
|
code. *WARNING:* although accepted, you should not pass in a perl |
262 |
|
|
filehandle here, as perl will likely close the file descriptor |
263 |
|
|
another time when the filehandle is destroyed. Normally, you can |
264 |
|
|
safely call perls "close" or just let filehandles go out of scope. |
265 |
|
|
|
266 |
|
|
This is supposed to be a bug in the API, so that might change. It's |
267 |
|
|
therefore best to avoid this function. |
268 |
|
|
|
269 |
|
|
aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
270 |
|
|
aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
271 |
|
|
Reads or writes "length" bytes from the specified "fh" and "offset" |
272 |
|
|
into the scalar given by "data" and offset "dataoffset" and calls |
273 |
|
|
the callback without the actual number of bytes read (or -1 on |
274 |
|
|
error, just like the syscall). |
275 |
|
|
|
276 |
|
|
The $data scalar *MUST NOT* be modified in any way while the request |
277 |
|
|
is outstanding. Modifying it can result in segfaults or WW3 (if the |
278 |
|
|
necessary/optional hardware is installed). |
279 |
|
|
|
280 |
|
|
Example: Read 15 bytes at offset 7 into scalar $buffer, starting at |
281 |
|
|
offset 0 within the scalar: |
282 |
|
|
|
283 |
|
|
aio_read $fh, 7, 15, $buffer, 0, sub { |
284 |
|
|
$_[0] > 0 or die "read error: $!"; |
285 |
|
|
print "read $_[0] bytes: <$buffer>\n"; |
286 |
|
|
}; |
287 |
|
|
|
288 |
|
|
aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval) |
289 |
|
|
Tries to copy $length bytes from $in_fh to $out_fh. It starts |
290 |
|
|
reading at byte offset $in_offset, and starts writing at the current |
291 |
|
|
file offset of $out_fh. Because of that, it is not safe to issue |
292 |
|
|
more than one "aio_sendfile" per $out_fh, as they will interfere |
293 |
|
|
with each other. |
294 |
|
|
|
295 |
|
|
This call tries to make use of a native "sendfile" syscall to |
296 |
|
|
provide zero-copy operation. For this to work, $out_fh should refer |
297 |
|
|
to a socket, and $in_fh should refer to mmap'able file. |
298 |
|
|
|
299 |
|
|
If the native sendfile call fails or is not implemented, it will be |
300 |
|
|
emulated, so you can call "aio_sendfile" on any type of filehandle |
301 |
|
|
regardless of the limitations of the operating system. |
302 |
|
|
|
303 |
|
|
Please note, however, that "aio_sendfile" can read more bytes from |
304 |
|
|
$in_fh than are written, and there is no way to find out how many |
305 |
|
|
bytes have been read from "aio_sendfile" alone, as "aio_sendfile" |
306 |
|
|
only provides the number of bytes written to $out_fh. Only if the |
307 |
|
|
result value equals $length one can assume that $length bytes have |
308 |
|
|
been read. |
309 |
|
|
|
310 |
|
|
aio_readahead $fh,$offset,$length, $callback->($retval) |
311 |
|
|
"aio_readahead" populates the page cache with data from a file so |
312 |
|
|
that subsequent reads from that file will not block on disk I/O. The |
313 |
|
|
$offset argument specifies the starting point from which data is to |
314 |
|
|
be read and $length specifies the number of bytes to be read. I/O is |
315 |
|
|
performed in whole pages, so that offset is effectively rounded down |
316 |
|
|
to a page boundary and bytes are read up to the next page boundary |
317 |
|
|
greater than or equal to (off-set+length). "aio_readahead" does not |
318 |
|
|
read beyond the end of the file. The current file offset of the file |
319 |
|
|
is left unchanged. |
320 |
|
|
|
321 |
|
|
If that syscall doesn't exist (likely if your OS isn't Linux) it |
322 |
|
|
will be emulated by simply reading the data, which would have a |
323 |
|
|
similar effect. |
324 |
|
|
|
325 |
|
|
aio_stat $fh_or_path, $callback->($status) |
326 |
|
|
aio_lstat $fh, $callback->($status) |
327 |
|
|
Works like perl's "stat" or "lstat" in void context. The callback |
328 |
|
|
will be called after the stat and the results will be available |
329 |
|
|
using "stat _" or "-s _" etc... |
330 |
|
|
|
331 |
|
|
The pathname passed to "aio_stat" must be absolute. See API NOTES, |
332 |
|
|
above, for an explanation. |
333 |
|
|
|
334 |
|
|
Currently, the stats are always 64-bit-stats, i.e. instead of |
335 |
|
|
returning an error when stat'ing a large file, the results will be |
336 |
|
|
silently truncated unless perl itself is compiled with large file |
337 |
|
|
support. |
338 |
|
|
|
339 |
|
|
Example: Print the length of /etc/passwd: |
340 |
|
|
|
341 |
|
|
aio_stat "/etc/passwd", sub { |
342 |
|
|
$_[0] and die "stat failed: $!"; |
343 |
|
|
print "size is ", -s _, "\n"; |
344 |
|
|
}; |
345 |
|
|
|
346 |
|
|
aio_unlink $pathname, $callback->($status) |
347 |
|
|
Asynchronously unlink (delete) a file and call the callback with the |
348 |
|
|
result code. |
349 |
|
|
|
350 |
|
|
aio_mknod $path, $mode, $dev, $callback->($status) |
351 |
|
|
[EXPERIMENTAL] |
352 |
|
|
|
353 |
|
|
Asynchronously create a device node (or fifo). See mknod(2). |
354 |
|
|
|
355 |
|
|
The only (POSIX-) portable way of calling this function is: |
356 |
|
|
|
357 |
|
|
aio_mknod $path, IO::AIO::S_IFIFO | $mode, 0, sub { ... |
358 |
|
|
|
359 |
|
|
aio_link $srcpath, $dstpath, $callback->($status) |
360 |
|
|
Asynchronously create a new link to the existing object at $srcpath |
361 |
|
|
at the path $dstpath and call the callback with the result code. |
362 |
|
|
|
363 |
|
|
aio_symlink $srcpath, $dstpath, $callback->($status) |
364 |
|
|
Asynchronously create a new symbolic link to the existing object at |
365 |
|
|
$srcpath at the path $dstpath and call the callback with the result |
366 |
|
|
code. |
367 |
|
|
|
368 |
|
|
aio_readlink $path, $callback->($link) |
369 |
|
|
Asynchronously read the symlink specified by $path and pass it to |
370 |
|
|
the callback. If an error occurs, nothing or undef gets passed to |
371 |
|
|
the callback. |
372 |
|
|
|
373 |
|
|
aio_rename $srcpath, $dstpath, $callback->($status) |
374 |
|
|
Asynchronously rename the object at $srcpath to $dstpath, just as |
375 |
|
|
rename(2) and call the callback with the result code. |
376 |
|
|
|
377 |
|
|
aio_rmdir $pathname, $callback->($status) |
378 |
|
|
Asynchronously rmdir (delete) a directory and call the callback with |
379 |
|
|
the result code. |
380 |
|
|
|
381 |
|
|
aio_readdir $pathname, $callback->($entries) |
382 |
|
|
Unlike the POSIX call of the same name, "aio_readdir" reads an |
383 |
|
|
entire directory (i.e. opendir + readdir + closedir). The entries |
384 |
|
|
will not be sorted, and will NOT include the "." and ".." entries. |
385 |
|
|
|
386 |
|
|
The callback a single argument which is either "undef" or an |
387 |
|
|
array-ref with the filenames. |
388 |
|
|
|
389 |
|
|
aio_copy $srcpath, $dstpath, $callback->($status) |
390 |
|
|
Try to copy the *file* (directories not supported as either source |
391 |
|
|
or destination) from $srcpath to $dstpath and call the callback with |
392 |
|
|
the 0 (error) or -1 ok. |
393 |
|
|
|
394 |
|
|
This is a composite request that it creates the destination file |
395 |
|
|
with mode 0200 and copies the contents of the source file into it |
396 |
|
|
using "aio_sendfile", followed by restoring atime, mtime, access |
397 |
|
|
mode and uid/gid, in that order. |
398 |
|
|
|
399 |
|
|
If an error occurs, the partial destination file will be unlinked, |
400 |
|
|
if possible, except when setting atime, mtime, access mode and |
401 |
|
|
uid/gid, where errors are being ignored. |
402 |
|
|
|
403 |
|
|
aio_move $srcpath, $dstpath, $callback->($status) |
404 |
|
|
Try to move the *file* (directories not supported as either source |
405 |
|
|
or destination) from $srcpath to $dstpath and call the callback with |
406 |
|
|
the 0 (error) or -1 ok. |
407 |
|
|
|
408 |
|
|
This is a composite request that tries to rename(2) the file first. |
409 |
|
|
If rename files with "EXDEV", it copies the file with "aio_copy" |
410 |
|
|
and, if that is successful, unlinking the $srcpath. |
411 |
|
|
|
412 |
|
|
aio_scandir $path, $maxreq, $callback->($dirs, $nondirs) |
413 |
|
|
Scans a directory (similar to "aio_readdir") but additionally tries |
414 |
|
|
to efficiently separate the entries of directory $path into two sets |
415 |
|
|
of names, directories you can recurse into (directories), and ones |
416 |
|
|
you cannot recurse into (everything else, including symlinks to |
417 |
|
|
directories). |
418 |
|
|
|
419 |
|
|
"aio_scandir" is a composite request that creates of many sub |
420 |
|
|
requests_ $maxreq specifies the maximum number of outstanding aio |
421 |
|
|
requests that this function generates. If it is "<= 0", then a |
422 |
|
|
suitable default will be chosen (currently 4). |
423 |
|
|
|
424 |
|
|
On error, the callback is called without arguments, otherwise it |
425 |
|
|
receives two array-refs with path-relative entry names. |
426 |
|
|
|
427 |
|
|
Example: |
428 |
|
|
|
429 |
|
|
aio_scandir $dir, 0, sub { |
430 |
|
|
my ($dirs, $nondirs) = @_; |
431 |
|
|
print "real directories: @$dirs\n"; |
432 |
|
|
print "everything else: @$nondirs\n"; |
433 |
|
|
}; |
434 |
|
|
|
435 |
|
|
Implementation notes. |
436 |
|
|
|
437 |
|
|
The "aio_readdir" cannot be avoided, but "stat()"'ing every entry |
438 |
|
|
can. |
439 |
|
|
|
440 |
|
|
After reading the directory, the modification time, size etc. of the |
441 |
|
|
directory before and after the readdir is checked, and if they match |
442 |
|
|
(and isn't the current time), the link count will be used to decide |
443 |
|
|
how many entries are directories (if >= 2). Otherwise, no knowledge |
444 |
|
|
of the number of subdirectories will be assumed. |
445 |
|
|
|
446 |
|
|
Then entries will be sorted into likely directories (everything |
447 |
|
|
without a non-initial dot currently) and likely non-directories |
448 |
|
|
(everything else). Then every entry plus an appended "/." will be |
449 |
|
|
"stat"'ed, likely directories first. If that succeeds, it assumes |
450 |
|
|
that the entry is a directory or a symlink to directory (which will |
451 |
|
|
be checked seperately). This is often faster than stat'ing the entry |
452 |
|
|
itself because filesystems might detect the type of the entry |
453 |
|
|
without reading the inode data (e.g. ext2fs filetype feature). |
454 |
|
|
|
455 |
|
|
If the known number of directories (link count - 2) has been |
456 |
|
|
reached, the rest of the entries is assumed to be non-directories. |
457 |
|
|
|
458 |
|
|
This only works with certainty on POSIX (= UNIX) filesystems, which |
459 |
|
|
fortunately are the vast majority of filesystems around. |
460 |
|
|
|
461 |
|
|
It will also likely work on non-POSIX filesystems with reduced |
462 |
|
|
efficiency as those tend to return 0 or 1 as link counts, which |
463 |
|
|
disables the directory counting heuristic. |
464 |
|
|
|
465 |
|
|
aio_fsync $fh, $callback->($status) |
466 |
|
|
Asynchronously call fsync on the given filehandle and call the |
467 |
|
|
callback with the fsync result code. |
468 |
|
|
|
469 |
|
|
aio_fdatasync $fh, $callback->($status) |
470 |
|
|
Asynchronously call fdatasync on the given filehandle and call the |
471 |
|
|
callback with the fdatasync result code. |
472 |
|
|
|
473 |
|
|
If this call isn't available because your OS lacks it or it couldn't |
474 |
|
|
be detected, it will be emulated by calling "fsync" instead. |
475 |
|
|
|
476 |
|
|
aio_group $callback->(...) |
477 |
|
|
This is a very special aio request: Instead of doing something, it |
478 |
|
|
is a container for other aio requests, which is useful if you want |
479 |
|
|
to bundle many requests into a single, composite, request with a |
480 |
|
|
definite callback and the ability to cancel the whole request with |
481 |
|
|
its subrequests. |
482 |
|
|
|
483 |
|
|
Returns an object of class IO::AIO::GRP. See its documentation below |
484 |
|
|
for more info. |
485 |
|
|
|
486 |
|
|
Example: |
487 |
|
|
|
488 |
|
|
my $grp = aio_group sub { |
489 |
|
|
print "all stats done\n"; |
490 |
|
|
}; |
491 |
|
|
|
492 |
|
|
add $grp |
493 |
|
|
(aio_stat ...), |
494 |
|
|
(aio_stat ...), |
495 |
|
|
...; |
496 |
|
|
|
497 |
|
|
aio_nop $callback->() |
498 |
|
|
This is a special request - it does nothing in itself and is only |
499 |
|
|
used for side effects, such as when you want to add a dummy request |
500 |
|
|
to a group so that finishing the requests in the group depends on |
501 |
|
|
executing the given code. |
502 |
|
|
|
503 |
|
|
While this request does nothing, it still goes through the execution |
504 |
|
|
phase and still requires a worker thread. Thus, the callback will |
505 |
|
|
not be executed immediately but only after other requests in the |
506 |
|
|
queue have entered their execution phase. This can be used to |
507 |
|
|
measure request latency. |
508 |
|
|
|
509 |
|
|
IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED* |
510 |
|
|
Mainly used for debugging and benchmarking, this aio request puts |
511 |
|
|
one of the request workers to sleep for the given time. |
512 |
|
|
|
513 |
|
|
While it is theoretically handy to have simple I/O scheduling |
514 |
|
|
requests like sleep and file handle readable/writable, the overhead |
515 |
|
|
this creates is immense (it blocks a thread for a long time) so do |
516 |
|
|
not use this function except to put your application under |
517 |
|
|
artificial I/O pressure. |
518 |
root |
1.18 |
|
519 |
|
|
IO::AIO::REQ CLASS |
520 |
root |
1.20 |
All non-aggregate "aio_*" functions return an object of this class when |
521 |
|
|
called in non-void context. |
522 |
root |
1.18 |
|
523 |
root |
1.20 |
cancel $req |
524 |
|
|
Cancels the request, if possible. Has the effect of skipping |
525 |
|
|
execution when entering the execute state and skipping calling the |
526 |
|
|
callback when entering the the result state, but will leave the |
527 |
|
|
request otherwise untouched. That means that requests that currently |
528 |
|
|
execute will not be stopped and resources held by the request will |
529 |
|
|
not be freed prematurely. |
530 |
root |
1.18 |
|
531 |
root |
1.20 |
cb $req $callback->(...) |
532 |
|
|
Replace (or simply set) the callback registered to the request. |
533 |
root |
1.18 |
|
534 |
|
|
IO::AIO::GRP CLASS |
535 |
root |
1.20 |
This class is a subclass of IO::AIO::REQ, so all its methods apply to |
536 |
|
|
objects of this class, too. |
537 |
root |
1.18 |
|
538 |
root |
1.20 |
A IO::AIO::GRP object is a special request that can contain multiple |
539 |
|
|
other aio requests. |
540 |
root |
1.18 |
|
541 |
root |
1.20 |
You create one by calling the "aio_group" constructing function with a |
542 |
|
|
callback that will be called when all contained requests have entered |
543 |
|
|
the "done" state: |
544 |
root |
1.18 |
|
545 |
root |
1.20 |
my $grp = aio_group sub { |
546 |
|
|
print "all requests are done\n"; |
547 |
|
|
}; |
548 |
|
|
|
549 |
|
|
You add requests by calling the "add" method with one or more |
550 |
|
|
"IO::AIO::REQ" objects: |
551 |
|
|
|
552 |
|
|
$grp->add (aio_unlink "..."); |
553 |
|
|
|
554 |
|
|
add $grp aio_stat "...", sub { |
555 |
|
|
$_[0] or return $grp->result ("error"); |
556 |
root |
1.1 |
|
557 |
root |
1.20 |
# add another request dynamically, if first succeeded |
558 |
|
|
add $grp aio_open "...", sub { |
559 |
|
|
$grp->result ("ok"); |
560 |
|
|
}; |
561 |
|
|
}; |
562 |
root |
1.18 |
|
563 |
root |
1.20 |
This makes it very easy to create composite requests (see the source of |
564 |
|
|
"aio_move" for an application) that work and feel like simple requests. |
565 |
root |
1.18 |
|
566 |
root |
1.20 |
* The IO::AIO::GRP objects will be cleaned up during calls to |
567 |
|
|
"IO::AIO::poll_cb", just like any other request. |
568 |
|
|
* They can be canceled like any other request. Canceling will cancel not |
569 |
|
|
only the request itself, but also all requests it contains. |
570 |
|
|
* They can also can also be added to other IO::AIO::GRP objects. |
571 |
|
|
* You must not add requests to a group from within the group callback |
572 |
|
|
(or any later time). |
573 |
|
|
|
574 |
|
|
Their lifetime, simplified, looks like this: when they are empty, they |
575 |
|
|
will finish very quickly. If they contain only requests that are in the |
576 |
|
|
"done" state, they will also finish. Otherwise they will continue to |
577 |
|
|
exist. |
578 |
|
|
|
579 |
|
|
That means after creating a group you have some time to add requests. |
580 |
|
|
And in the callbacks of those requests, you can add further requests to |
581 |
|
|
the group. And only when all those requests have finished will the the |
582 |
|
|
group itself finish. |
583 |
|
|
|
584 |
|
|
add $grp ... |
585 |
|
|
$grp->add (...) |
586 |
|
|
Add one or more requests to the group. Any type of IO::AIO::REQ can |
587 |
|
|
be added, including other groups, as long as you do not create |
588 |
|
|
circular dependencies. |
589 |
|
|
|
590 |
|
|
Returns all its arguments. |
591 |
|
|
|
592 |
|
|
$grp->cancel_subs |
593 |
|
|
Cancel all subrequests and clears any feeder, but not the group |
594 |
|
|
request itself. Useful when you queued a lot of events but got a |
595 |
|
|
result early. |
596 |
|
|
|
597 |
|
|
$grp->result (...) |
598 |
|
|
Set the result value(s) that will be passed to the group callback |
599 |
|
|
when all subrequests have finished and set thre groups errno to the |
600 |
|
|
current value of errno (just like calling "errno" without an error |
601 |
|
|
number). By default, no argument will be passed and errno is zero. |
602 |
|
|
|
603 |
|
|
$grp->errno ([$errno]) |
604 |
|
|
Sets the group errno value to $errno, or the current value of errno |
605 |
|
|
when the argument is missing. |
606 |
|
|
|
607 |
|
|
Every aio request has an associated errno value that is restored |
608 |
|
|
when the callback is invoked. This method lets you change this value |
609 |
|
|
from its default (0). |
610 |
|
|
|
611 |
|
|
Calling "result" will also set errno, so make sure you either set $! |
612 |
|
|
before the call to "result", or call c<errno> after it. |
613 |
|
|
|
614 |
|
|
feed $grp $callback->($grp) |
615 |
|
|
Sets a feeder/generator on this group: every group can have an |
616 |
|
|
attached generator that generates requests if idle. The idea behind |
617 |
|
|
this is that, although you could just queue as many requests as you |
618 |
|
|
want in a group, this might starve other requests for a potentially |
619 |
|
|
long time. For example, "aio_scandir" might generate hundreds of |
620 |
|
|
thousands "aio_stat" requests, delaying any later requests for a |
621 |
|
|
long time. |
622 |
|
|
|
623 |
|
|
To avoid this, and allow incremental generation of requests, you can |
624 |
|
|
instead a group and set a feeder on it that generates those |
625 |
|
|
requests. The feed callback will be called whenever there are few |
626 |
|
|
enough (see "limit", below) requests active in the group itself and |
627 |
|
|
is expected to queue more requests. |
628 |
|
|
|
629 |
|
|
The feed callback can queue as many requests as it likes (i.e. "add" |
630 |
|
|
does not impose any limits). |
631 |
|
|
|
632 |
|
|
If the feed does not queue more requests when called, it will be |
633 |
|
|
automatically removed from the group. |
634 |
|
|
|
635 |
|
|
If the feed limit is 0, it will be set to 2 automatically. |
636 |
|
|
|
637 |
|
|
Example: |
638 |
|
|
|
639 |
|
|
# stat all files in @files, but only ever use four aio requests concurrently: |
640 |
|
|
|
641 |
|
|
my $grp = aio_group sub { print "finished\n" }; |
642 |
|
|
limit $grp 4; |
643 |
|
|
feed $grp sub { |
644 |
|
|
my $file = pop @files |
645 |
|
|
or return; |
646 |
root |
1.18 |
|
647 |
root |
1.20 |
add $grp aio_stat $file, sub { ... }; |
648 |
root |
1.1 |
}; |
649 |
|
|
|
650 |
root |
1.20 |
limit $grp $num |
651 |
|
|
Sets the feeder limit for the group: The feeder will be called |
652 |
|
|
whenever the group contains less than this many requests. |
653 |
root |
1.18 |
|
654 |
root |
1.20 |
Setting the limit to 0 will pause the feeding process. |
655 |
root |
1.17 |
|
656 |
root |
1.18 |
SUPPORT FUNCTIONS |
657 |
root |
1.19 |
EVENT PROCESSING AND EVENT LOOP INTEGRATION |
658 |
root |
1.20 |
$fileno = IO::AIO::poll_fileno |
659 |
|
|
Return the *request result pipe file descriptor*. This filehandle |
660 |
|
|
must be polled for reading by some mechanism outside this module |
661 |
|
|
(e.g. Event or select, see below or the SYNOPSIS). If the pipe |
662 |
|
|
becomes readable you have to call "poll_cb" to check the results. |
663 |
|
|
|
664 |
|
|
See "poll_cb" for an example. |
665 |
|
|
|
666 |
|
|
IO::AIO::poll_cb |
667 |
|
|
Process some outstanding events on the result pipe. You have to call |
668 |
|
|
this regularly. Returns the number of events processed. Returns |
669 |
|
|
immediately when no events are outstanding. The amount of events |
670 |
|
|
processed depends on the settings of "IO::AIO::max_poll_req" and |
671 |
|
|
"IO::AIO::max_poll_time". |
672 |
|
|
|
673 |
|
|
If not all requests were processed for whatever reason, the |
674 |
|
|
filehandle will still be ready when "poll_cb" returns. |
675 |
|
|
|
676 |
|
|
Example: Install an Event watcher that automatically calls |
677 |
|
|
IO::AIO::poll_cb with high priority: |
678 |
|
|
|
679 |
|
|
Event->io (fd => IO::AIO::poll_fileno, |
680 |
|
|
poll => 'r', async => 1, |
681 |
|
|
cb => \&IO::AIO::poll_cb); |
682 |
|
|
|
683 |
|
|
IO::AIO::max_poll_reqs $nreqs |
684 |
|
|
IO::AIO::max_poll_time $seconds |
685 |
|
|
These set the maximum number of requests (default 0, meaning |
686 |
|
|
infinity) that are being processed by "IO::AIO::poll_cb" in one |
687 |
|
|
call, respectively the maximum amount of time (default 0, meaning |
688 |
|
|
infinity) spent in "IO::AIO::poll_cb" to process requests (more |
689 |
|
|
correctly the mininum amount of time "poll_cb" is allowed to use). |
690 |
|
|
|
691 |
|
|
Setting "max_poll_time" to a non-zero value creates an overhead of |
692 |
|
|
one syscall per request processed, which is not normally a problem |
693 |
|
|
unless your callbacks are really really fast or your OS is really |
694 |
|
|
really slow (I am not mentioning Solaris here). Using |
695 |
|
|
"max_poll_reqs" incurs no overhead. |
696 |
|
|
|
697 |
|
|
Setting these is useful if you want to ensure some level of |
698 |
|
|
interactiveness when perl is not fast enough to process all requests |
699 |
|
|
in time. |
700 |
|
|
|
701 |
|
|
For interactive programs, values such as 0.01 to 0.1 should be fine. |
702 |
root |
1.4 |
|
703 |
root |
1.20 |
Example: Install an Event watcher that automatically calls |
704 |
|
|
IO::AIO::poll_cb with low priority, to ensure that other parts of |
705 |
|
|
the program get the CPU sometimes even under high AIO load. |
706 |
root |
1.4 |
|
707 |
root |
1.20 |
# try not to spend much more than 0.1s in poll_cb |
708 |
|
|
IO::AIO::max_poll_time 0.1; |
709 |
root |
1.4 |
|
710 |
root |
1.20 |
# use a low priority so other tasks have priority |
711 |
|
|
Event->io (fd => IO::AIO::poll_fileno, |
712 |
|
|
poll => 'r', nice => 1, |
713 |
|
|
cb => &IO::AIO::poll_cb); |
714 |
|
|
|
715 |
|
|
IO::AIO::poll_wait |
716 |
|
|
Wait till the result filehandle becomes ready for reading (simply |
717 |
|
|
does a "select" on the filehandle. This is useful if you want to |
718 |
|
|
synchronously wait for some requests to finish). |
719 |
|
|
|
720 |
|
|
See "nreqs" for an example. |
721 |
|
|
|
722 |
|
|
IO::AIO::poll |
723 |
|
|
Waits until some requests have been handled. |
724 |
|
|
|
725 |
|
|
Strictly equivalent to: |
726 |
|
|
|
727 |
|
|
IO::AIO::poll_wait, IO::AIO::poll_cb |
728 |
|
|
if IO::AIO::nreqs; |
729 |
|
|
|
730 |
|
|
IO::AIO::flush |
731 |
|
|
Wait till all outstanding AIO requests have been handled. |
732 |
|
|
|
733 |
|
|
Strictly equivalent to: |
734 |
|
|
|
735 |
|
|
IO::AIO::poll_wait, IO::AIO::poll_cb |
736 |
|
|
while IO::AIO::nreqs; |
737 |
root |
1.6 |
|
738 |
root |
1.19 |
CONTROLLING THE NUMBER OF THREADS |
739 |
root |
1.20 |
IO::AIO::min_parallel $nthreads |
740 |
|
|
Set the minimum number of AIO threads to $nthreads. The current |
741 |
|
|
default is 8, which means eight asynchronous operations can execute |
742 |
|
|
concurrently at any one time (the number of outstanding requests, |
743 |
|
|
however, is unlimited). |
744 |
|
|
|
745 |
|
|
IO::AIO starts threads only on demand, when an AIO request is queued |
746 |
|
|
and no free thread exists. Please note that queueing up a hundred |
747 |
|
|
requests can create demand for a hundred threads, even if it turns |
748 |
|
|
out that everything is in the cache and could have been processed |
749 |
|
|
faster by a single thread. |
750 |
|
|
|
751 |
|
|
It is recommended to keep the number of threads relatively low, as |
752 |
|
|
some Linux kernel versions will scale negatively with the number of |
753 |
|
|
threads (higher parallelity => MUCH higher latency). With current |
754 |
|
|
Linux 2.6 versions, 4-32 threads should be fine. |
755 |
|
|
|
756 |
|
|
Under most circumstances you don't need to call this function, as |
757 |
|
|
the module selects a default that is suitable for low to moderate |
758 |
|
|
load. |
759 |
|
|
|
760 |
|
|
IO::AIO::max_parallel $nthreads |
761 |
|
|
Sets the maximum number of AIO threads to $nthreads. If more than |
762 |
|
|
the specified number of threads are currently running, this function |
763 |
|
|
kills them. This function blocks until the limit is reached. |
764 |
|
|
|
765 |
|
|
While $nthreads are zero, aio requests get queued but not executed |
766 |
|
|
until the number of threads has been increased again. |
767 |
|
|
|
768 |
|
|
This module automatically runs "max_parallel 0" at program end, to |
769 |
|
|
ensure that all threads are killed and that there are no outstanding |
770 |
|
|
requests. |
771 |
|
|
|
772 |
|
|
Under normal circumstances you don't need to call this function. |
773 |
|
|
|
774 |
|
|
IO::AIO::max_idle $nthreads |
775 |
|
|
Limit the number of threads (default: 4) that are allowed to idle |
776 |
|
|
(i.e., threads that did not get a request to process within 10 |
777 |
|
|
seconds). That means if a thread becomes idle while $nthreads other |
778 |
|
|
threads are also idle, it will free its resources and exit. |
779 |
|
|
|
780 |
|
|
This is useful when you allow a large number of threads (e.g. 100 or |
781 |
|
|
1000) to allow for extremely high load situations, but want to free |
782 |
|
|
resources under normal circumstances (1000 threads can easily |
783 |
|
|
consume 30MB of RAM). |
784 |
|
|
|
785 |
|
|
The default is probably ok in most situations, especially if thread |
786 |
|
|
creation is fast. If thread creation is very slow on your system you |
787 |
|
|
might want to use larger values. |
788 |
|
|
|
789 |
|
|
$oldmaxreqs = IO::AIO::max_outstanding $maxreqs |
790 |
|
|
This is a very bad function to use in interactive programs because |
791 |
|
|
it blocks, and a bad way to reduce concurrency because it is |
792 |
|
|
inexact: Better use an "aio_group" together with a feed callback. |
793 |
|
|
|
794 |
|
|
Sets the maximum number of outstanding requests to $nreqs. If you to |
795 |
|
|
queue up more than this number of requests, the next call to the |
796 |
|
|
"poll_cb" (and "poll_some" and other functions calling "poll_cb") |
797 |
|
|
function will block until the limit is no longer exceeded. |
798 |
|
|
|
799 |
|
|
The default value is very large, so there is no practical limit on |
800 |
|
|
the number of outstanding requests. |
801 |
|
|
|
802 |
|
|
You can still queue as many requests as you want. Therefore, |
803 |
|
|
"max_oustsanding" is mainly useful in simple scripts (with low |
804 |
|
|
values) or as a stop gap to shield against fatal memory overflow |
805 |
|
|
(with large values). |
806 |
root |
1.1 |
|
807 |
root |
1.19 |
STATISTICAL INFORMATION |
808 |
root |
1.20 |
IO::AIO::nreqs |
809 |
|
|
Returns the number of requests currently in the ready, execute or |
810 |
|
|
pending states (i.e. for which their callback has not been invoked |
811 |
|
|
yet). |
812 |
|
|
|
813 |
|
|
Example: wait till there are no outstanding requests anymore: |
814 |
|
|
|
815 |
|
|
IO::AIO::poll_wait, IO::AIO::poll_cb |
816 |
|
|
while IO::AIO::nreqs; |
817 |
|
|
|
818 |
|
|
IO::AIO::nready |
819 |
|
|
Returns the number of requests currently in the ready state (not yet |
820 |
|
|
executed). |
821 |
|
|
|
822 |
|
|
IO::AIO::npending |
823 |
|
|
Returns the number of requests currently in the pending state |
824 |
|
|
(executed, but not yet processed by poll_cb). |
825 |
root |
1.19 |
|
826 |
root |
1.9 |
FORK BEHAVIOUR |
827 |
root |
1.20 |
This module should do "the right thing" when the process using it forks: |
828 |
root |
1.18 |
|
829 |
root |
1.20 |
Before the fork, IO::AIO enters a quiescent state where no requests can |
830 |
|
|
be added in other threads and no results will be processed. After the |
831 |
|
|
fork the parent simply leaves the quiescent state and continues |
832 |
|
|
request/result processing, while the child frees the request/result |
833 |
|
|
queue (so that the requests started before the fork will only be handled |
834 |
|
|
in the parent). Threads will be started on demand until the limit set in |
835 |
|
|
the parent process has been reached again. |
836 |
|
|
|
837 |
|
|
In short: the parent will, after a short pause, continue as if fork had |
838 |
|
|
not been called, while the child will act as if IO::AIO has not been |
839 |
|
|
used yet. |
840 |
root |
1.18 |
|
841 |
|
|
MEMORY USAGE |
842 |
root |
1.20 |
Per-request usage: |
843 |
root |
1.18 |
|
844 |
root |
1.20 |
Each aio request uses - depending on your architecture - around 100-200 |
845 |
|
|
bytes of memory. In addition, stat requests need a stat buffer (possibly |
846 |
|
|
a few hundred bytes), readdir requires a result buffer and so on. Perl |
847 |
|
|
scalars and other data passed into aio requests will also be locked and |
848 |
|
|
will consume memory till the request has entered the done state. |
849 |
|
|
|
850 |
|
|
This is now awfully much, so queuing lots of requests is not usually a |
851 |
|
|
problem. |
852 |
|
|
|
853 |
|
|
Per-thread usage: |
854 |
|
|
|
855 |
|
|
In the execution phase, some aio requests require more memory for |
856 |
|
|
temporary buffers, and each thread requires a stack and other data |
857 |
|
|
structures (usually around 16k-128k, depending on the OS). |
858 |
root |
1.18 |
|
859 |
|
|
KNOWN BUGS |
860 |
root |
1.20 |
Known bugs will be fixed in the next release. |
861 |
root |
1.9 |
|
862 |
root |
1.1 |
SEE ALSO |
863 |
root |
1.20 |
Coro::AIO. |
864 |
root |
1.1 |
|
865 |
|
|
AUTHOR |
866 |
root |
1.20 |
Marc Lehmann <schmorp@schmorp.de> |
867 |
|
|
http://home.schmorp.de/ |
868 |
root |
1.1 |
|