1 |
NAME |
2 |
IO::AIO - Asynchronous Input/Output |
3 |
|
4 |
SYNOPSIS |
5 |
use IO::AIO; |
6 |
|
7 |
aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub { |
8 |
my $fh = shift |
9 |
or die "/etc/passwd: $!"; |
10 |
... |
11 |
}; |
12 |
|
13 |
aio_unlink "/tmp/file", sub { }; |
14 |
|
15 |
aio_read $fh, 30000, 1024, $buffer, 0, sub { |
16 |
$_[0] > 0 or die "read error: $!"; |
17 |
}; |
18 |
|
19 |
# version 2+ has request and group objects |
20 |
use IO::AIO 2; |
21 |
|
22 |
aioreq_pri 4; # give next request a very high priority |
23 |
my $req = aio_unlink "/tmp/file", sub { }; |
24 |
$req->cancel; # cancel request if still in queue |
25 |
|
26 |
my $grp = aio_group sub { print "all stats done\n" }; |
27 |
add $grp aio_stat "..." for ...; |
28 |
|
29 |
DESCRIPTION |
30 |
This module implements asynchronous I/O using whatever means your |
31 |
operating system supports. It is implemented as an interface to "libeio" |
32 |
(<http://software.schmorp.de/pkg/libeio.html>). |
33 |
|
34 |
Asynchronous means that operations that can normally block your program |
35 |
(e.g. reading from disk) will be done asynchronously: the operation will |
36 |
still block, but you can do something else in the meantime. This is |
37 |
extremely useful for programs that need to stay interactive even when |
38 |
doing heavy I/O (GUI programs, high performance network servers etc.), |
39 |
but can also be used to easily do operations in parallel that are |
40 |
normally done sequentially, e.g. stat'ing many files, which is much |
41 |
faster on a RAID volume or over NFS when you do a number of stat |
42 |
operations concurrently. |
43 |
|
44 |
While most of this works on all types of file descriptors (for example |
45 |
sockets), using these functions on file descriptors that support |
46 |
nonblocking operation (again, sockets, pipes etc.) is very inefficient. |
47 |
Use an event loop for that (such as the EV module): IO::AIO will |
48 |
naturally fit into such an event loop itself. |
49 |
|
50 |
In this version, a number of threads are started that execute your |
51 |
requests and signal their completion. You don't need thread support in |
52 |
perl, and the threads created by this module will not be visible to |
53 |
perl. In the future, this module might make use of the native aio |
54 |
functions available on many operating systems. However, they are often |
55 |
not well-supported or restricted (GNU/Linux doesn't allow them on normal |
56 |
files currently, for example), and they would only support aio_read and |
57 |
aio_write, so the remaining functionality would have to be implemented |
58 |
using threads anyway. |
59 |
|
60 |
Although the module will work in the presence of other (Perl-) threads, |
61 |
it is currently not reentrant in any way, so use appropriate locking |
62 |
yourself, always call "poll_cb" from within the same thread, or never |
63 |
call "poll_cb" (or other "aio_" functions) recursively. |
64 |
|
65 |
EXAMPLE |
66 |
This is a simple example that uses the EV module and loads /etc/passwd |
67 |
asynchronously: |
68 |
|
69 |
use Fcntl; |
70 |
use EV; |
71 |
use IO::AIO; |
72 |
|
73 |
# register the IO::AIO callback with EV |
74 |
my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb; |
75 |
|
76 |
# queue the request to open /etc/passwd |
77 |
aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub { |
78 |
my $fh = shift |
79 |
or die "error while opening: $!"; |
80 |
|
81 |
# stat'ing filehandles is generally non-blocking |
82 |
my $size = -s $fh; |
83 |
|
84 |
# queue a request to read the file |
85 |
my $contents; |
86 |
aio_read $fh, 0, $size, $contents, 0, sub { |
87 |
$_[0] == $size |
88 |
or die "short read: $!"; |
89 |
|
90 |
close $fh; |
91 |
|
92 |
# file contents now in $contents |
93 |
print $contents; |
94 |
|
95 |
# exit event loop and program |
96 |
EV::unloop; |
97 |
}; |
98 |
}; |
99 |
|
100 |
# possibly queue up other requests, or open GUI windows, |
101 |
# check for sockets etc. etc. |
102 |
|
103 |
# process events as long as there are some: |
104 |
EV::loop; |
105 |
|
106 |
REQUEST ANATOMY AND LIFETIME |
107 |
Every "aio_*" function creates a request. which is a C data structure |
108 |
not directly visible to Perl. |
109 |
|
110 |
If called in non-void context, every request function returns a Perl |
111 |
object representing the request. In void context, nothing is returned, |
112 |
which saves a bit of memory. |
113 |
|
114 |
The perl object is a fairly standard ref-to-hash object. The hash |
115 |
contents are not used by IO::AIO so you are free to store anything you |
116 |
like in it. |
117 |
|
118 |
During their existance, aio requests travel through the following |
119 |
states, in order: |
120 |
|
121 |
ready |
122 |
Immediately after a request is created it is put into the ready |
123 |
state, waiting for a thread to execute it. |
124 |
|
125 |
execute |
126 |
A thread has accepted the request for processing and is currently |
127 |
executing it (e.g. blocking in read). |
128 |
|
129 |
pending |
130 |
The request has been executed and is waiting for result processing. |
131 |
|
132 |
While request submission and execution is fully asynchronous, result |
133 |
processing is not and relies on the perl interpreter calling |
134 |
"poll_cb" (or another function with the same effect). |
135 |
|
136 |
result |
137 |
The request results are processed synchronously by "poll_cb". |
138 |
|
139 |
The "poll_cb" function will process all outstanding aio requests by |
140 |
calling their callbacks, freeing memory associated with them and |
141 |
managing any groups they are contained in. |
142 |
|
143 |
done |
144 |
Request has reached the end of its lifetime and holds no resources |
145 |
anymore (except possibly for the Perl object, but its connection to |
146 |
the actual aio request is severed and calling its methods will |
147 |
either do nothing or result in a runtime error). |
148 |
|
149 |
FUNCTIONS |
150 |
QUICK OVERVIEW |
151 |
This section simply lists the prototypes of the most important functions |
152 |
for quick reference. See the following sections for function-by-function |
153 |
documentation. |
154 |
|
155 |
aio_open $pathname, $flags, $mode, $callback->($fh) |
156 |
aio_close $fh, $callback->($status) |
157 |
aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
158 |
aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
159 |
aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval) |
160 |
aio_readahead $fh,$offset,$length, $callback->($retval) |
161 |
aio_stat $fh_or_path, $callback->($status) |
162 |
aio_lstat $fh, $callback->($status) |
163 |
aio_statvfs $fh_or_path, $callback->($statvfs) |
164 |
aio_utime $fh_or_path, $atime, $mtime, $callback->($status) |
165 |
aio_chown $fh_or_path, $uid, $gid, $callback->($status) |
166 |
aio_truncate $fh_or_path, $offset, $callback->($status) |
167 |
aio_chmod $fh_or_path, $mode, $callback->($status) |
168 |
aio_unlink $pathname, $callback->($status) |
169 |
aio_mknod $path, $mode, $dev, $callback->($status) |
170 |
aio_link $srcpath, $dstpath, $callback->($status) |
171 |
aio_symlink $srcpath, $dstpath, $callback->($status) |
172 |
aio_readlink $path, $callback->($link) |
173 |
aio_rename $srcpath, $dstpath, $callback->($status) |
174 |
aio_mkdir $pathname, $mode, $callback->($status) |
175 |
aio_rmdir $pathname, $callback->($status) |
176 |
aio_readdir $pathname, $callback->($entries) |
177 |
aio_readdirx $pathname, $flags, $callback->($entries, $flags) |
178 |
IO::AIO::READDIR_DENTS IO::AIO::READDIR_DIRS_FIRST |
179 |
IO::AIO::READDIR_STAT_ORDER IO::AIO::READDIR_FOUND_UNKNOWN |
180 |
aio_load $path, $data, $callback->($status) |
181 |
aio_copy $srcpath, $dstpath, $callback->($status) |
182 |
aio_move $srcpath, $dstpath, $callback->($status) |
183 |
aio_scandir $path, $maxreq, $callback->($dirs, $nondirs) |
184 |
aio_rmtree $path, $callback->($status) |
185 |
aio_sync $callback->($status) |
186 |
aio_fsync $fh, $callback->($status) |
187 |
aio_fdatasync $fh, $callback->($status) |
188 |
aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status) |
189 |
aio_pathsync $path, $callback->($status) |
190 |
aio_msync $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status) |
191 |
aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status) |
192 |
aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status) |
193 |
aio_mlockall $flags, $callback->($status) |
194 |
aio_group $callback->(...) |
195 |
aio_nop $callback->() |
196 |
|
197 |
$prev_pri = aioreq_pri [$pri] |
198 |
aioreq_nice $pri_adjust |
199 |
|
200 |
IO::AIO::poll_wait |
201 |
IO::AIO::poll_cb |
202 |
IO::AIO::poll |
203 |
IO::AIO::flush |
204 |
IO::AIO::max_poll_reqs $nreqs |
205 |
IO::AIO::max_poll_time $seconds |
206 |
IO::AIO::min_parallel $nthreads |
207 |
IO::AIO::max_parallel $nthreads |
208 |
IO::AIO::max_idle $nthreads |
209 |
IO::AIO::max_outstanding $maxreqs |
210 |
IO::AIO::nreqs |
211 |
IO::AIO::nready |
212 |
IO::AIO::npending |
213 |
|
214 |
IO::AIO::sendfile $ofh, $ifh, $offset, $count |
215 |
IO::AIO::fadvise $fh, $offset, $len, $advice |
216 |
IO::AIO::madvise $scalar, $offset, $length, $advice |
217 |
IO::AIO::mprotect $scalar, $offset, $length, $protect |
218 |
IO::AIO::munlock $scalar, $offset = 0, $length = undef |
219 |
IO::AIO::munlockall |
220 |
|
221 |
AIO REQUEST FUNCTIONS |
222 |
All the "aio_*" calls are more or less thin wrappers around the syscall |
223 |
with the same name (sans "aio_"). The arguments are similar or |
224 |
identical, and they all accept an additional (and optional) $callback |
225 |
argument which must be a code reference. This code reference will get |
226 |
called with the syscall return code (e.g. most syscalls return -1 on |
227 |
error, unlike perl, which usually delivers "false") as its sole argument |
228 |
after the given syscall has been executed asynchronously. |
229 |
|
230 |
All functions expecting a filehandle keep a copy of the filehandle |
231 |
internally until the request has finished. |
232 |
|
233 |
All functions return request objects of type IO::AIO::REQ that allow |
234 |
further manipulation of those requests while they are in-flight. |
235 |
|
236 |
The pathnames you pass to these routines *must* be absolute and encoded |
237 |
as octets. The reason for the former is that at the time the request is |
238 |
being executed, the current working directory could have changed. |
239 |
Alternatively, you can make sure that you never change the current |
240 |
working directory anywhere in the program and then use relative paths. |
241 |
|
242 |
To encode pathnames as octets, either make sure you either: a) always |
243 |
pass in filenames you got from outside (command line, readdir etc.) |
244 |
without tinkering, b) are ASCII or ISO 8859-1, c) use the Encode module |
245 |
and encode your pathnames to the locale (or other) encoding in effect in |
246 |
the user environment, d) use Glib::filename_from_unicode on unicode |
247 |
filenames or e) use something else to ensure your scalar has the correct |
248 |
contents. |
249 |
|
250 |
This works, btw. independent of the internal UTF-8 bit, which IO::AIO |
251 |
handles correctly whether it is set or not. |
252 |
|
253 |
$prev_pri = aioreq_pri [$pri] |
254 |
Returns the priority value that would be used for the next request |
255 |
and, if $pri is given, sets the priority for the next aio request. |
256 |
|
257 |
The default priority is 0, the minimum and maximum priorities are -4 |
258 |
and 4, respectively. Requests with higher priority will be serviced |
259 |
first. |
260 |
|
261 |
The priority will be reset to 0 after each call to one of the |
262 |
"aio_*" functions. |
263 |
|
264 |
Example: open a file with low priority, then read something from it |
265 |
with higher priority so the read request is serviced before other |
266 |
low priority open requests (potentially spamming the cache): |
267 |
|
268 |
aioreq_pri -3; |
269 |
aio_open ..., sub { |
270 |
return unless $_[0]; |
271 |
|
272 |
aioreq_pri -2; |
273 |
aio_read $_[0], ..., sub { |
274 |
... |
275 |
}; |
276 |
}; |
277 |
|
278 |
aioreq_nice $pri_adjust |
279 |
Similar to "aioreq_pri", but subtracts the given value from the |
280 |
current priority, so the effect is cumulative. |
281 |
|
282 |
aio_open $pathname, $flags, $mode, $callback->($fh) |
283 |
Asynchronously open or create a file and call the callback with a |
284 |
newly created filehandle for the file. |
285 |
|
286 |
The pathname passed to "aio_open" must be absolute. See API NOTES, |
287 |
above, for an explanation. |
288 |
|
289 |
The $flags argument is a bitmask. See the "Fcntl" module for a list. |
290 |
They are the same as used by "sysopen". |
291 |
|
292 |
Likewise, $mode specifies the mode of the newly created file, if it |
293 |
didn't exist and "O_CREAT" has been given, just like perl's |
294 |
"sysopen", except that it is mandatory (i.e. use 0 if you don't |
295 |
create new files, and 0666 or 0777 if you do). Note that the $mode |
296 |
will be modified by the umask in effect then the request is being |
297 |
executed, so better never change the umask. |
298 |
|
299 |
Example: |
300 |
|
301 |
aio_open "/etc/passwd", IO::AIO::O_RDONLY, 0, sub { |
302 |
if ($_[0]) { |
303 |
print "open successful, fh is $_[0]\n"; |
304 |
... |
305 |
} else { |
306 |
die "open failed: $!\n"; |
307 |
} |
308 |
}; |
309 |
|
310 |
aio_close $fh, $callback->($status) |
311 |
Asynchronously close a file and call the callback with the result |
312 |
code. |
313 |
|
314 |
Unfortunately, you can't do this to perl. Perl *insists* very |
315 |
strongly on closing the file descriptor associated with the |
316 |
filehandle itself. |
317 |
|
318 |
Therefore, "aio_close" will not close the filehandle - instead it |
319 |
will use dup2 to overwrite the file descriptor with the write-end of |
320 |
a pipe (the pipe fd will be created on demand and will be cached). |
321 |
|
322 |
Or in other words: the file descriptor will be closed, but it will |
323 |
not be free for reuse until the perl filehandle is closed. |
324 |
|
325 |
aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
326 |
aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval) |
327 |
Reads or writes $length bytes from or to the specified $fh and |
328 |
$offset into the scalar given by $data and offset $dataoffset and |
329 |
calls the callback without the actual number of bytes read (or -1 on |
330 |
error, just like the syscall). |
331 |
|
332 |
"aio_read" will, like "sysread", shrink or grow the $data scalar to |
333 |
offset plus the actual number of bytes read. |
334 |
|
335 |
If $offset is undefined, then the current file descriptor offset |
336 |
will be used (and updated), otherwise the file descriptor offset |
337 |
will not be changed by these calls. |
338 |
|
339 |
If $length is undefined in "aio_write", use the remaining length of |
340 |
$data. |
341 |
|
342 |
If $dataoffset is less than zero, it will be counted from the end of |
343 |
$data. |
344 |
|
345 |
The $data scalar *MUST NOT* be modified in any way while the request |
346 |
is outstanding. Modifying it can result in segfaults or World War |
347 |
III (if the necessary/optional hardware is installed). |
348 |
|
349 |
Example: Read 15 bytes at offset 7 into scalar $buffer, starting at |
350 |
offset 0 within the scalar: |
351 |
|
352 |
aio_read $fh, 7, 15, $buffer, 0, sub { |
353 |
$_[0] > 0 or die "read error: $!"; |
354 |
print "read $_[0] bytes: <$buffer>\n"; |
355 |
}; |
356 |
|
357 |
aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval) |
358 |
Tries to copy $length bytes from $in_fh to $out_fh. It starts |
359 |
reading at byte offset $in_offset, and starts writing at the current |
360 |
file offset of $out_fh. Because of that, it is not safe to issue |
361 |
more than one "aio_sendfile" per $out_fh, as they will interfere |
362 |
with each other. |
363 |
|
364 |
Please note that "aio_sendfile" can read more bytes from $in_fh than |
365 |
are written, and there is no way to find out how many bytes have |
366 |
been read from "aio_sendfile" alone, as "aio_sendfile" only provides |
367 |
the number of bytes written to $out_fh. Only if the result value |
368 |
equals $length one can assume that $length bytes have been read. |
369 |
|
370 |
Unlike with other "aio_" functions, it makes a lot of sense to use |
371 |
"aio_sendfile" on non-blocking sockets, as long as one end |
372 |
(typically the $in_fh) is a file - the file I/O will then be |
373 |
asynchronous, while the socket I/O will be non-blocking. Note, |
374 |
however, that you can run into a trap where "aio_sendfile" reads |
375 |
some data with readahead, then fails to write all data, and when the |
376 |
socket is ready the next time, the data in the cache is already |
377 |
lost, forcing "aio_sendfile" to again hit the disk. Explicit |
378 |
"aio_read" + "aio_write" let's you control resource usage much |
379 |
better. |
380 |
|
381 |
This call tries to make use of a native "sendfile" syscall to |
382 |
provide zero-copy operation. For this to work, $out_fh should refer |
383 |
to a socket, and $in_fh should refer to an mmap'able file. |
384 |
|
385 |
If a native sendfile cannot be found or it fails with "ENOSYS", |
386 |
"ENOTSUP", "EOPNOTSUPP", "EAFNOSUPPORT", "EPROTOTYPE" or "ENOTSOCK", |
387 |
it will be emulated, so you can call "aio_sendfile" on any type of |
388 |
filehandle regardless of the limitations of the operating system. |
389 |
|
390 |
aio_readahead $fh,$offset,$length, $callback->($retval) |
391 |
"aio_readahead" populates the page cache with data from a file so |
392 |
that subsequent reads from that file will not block on disk I/O. The |
393 |
$offset argument specifies the starting point from which data is to |
394 |
be read and $length specifies the number of bytes to be read. I/O is |
395 |
performed in whole pages, so that offset is effectively rounded down |
396 |
to a page boundary and bytes are read up to the next page boundary |
397 |
greater than or equal to (off-set+length). "aio_readahead" does not |
398 |
read beyond the end of the file. The current file offset of the file |
399 |
is left unchanged. |
400 |
|
401 |
If that syscall doesn't exist (likely if your OS isn't Linux) it |
402 |
will be emulated by simply reading the data, which would have a |
403 |
similar effect. |
404 |
|
405 |
aio_stat $fh_or_path, $callback->($status) |
406 |
aio_lstat $fh, $callback->($status) |
407 |
Works like perl's "stat" or "lstat" in void context. The callback |
408 |
will be called after the stat and the results will be available |
409 |
using "stat _" or "-s _" etc... |
410 |
|
411 |
The pathname passed to "aio_stat" must be absolute. See API NOTES, |
412 |
above, for an explanation. |
413 |
|
414 |
Currently, the stats are always 64-bit-stats, i.e. instead of |
415 |
returning an error when stat'ing a large file, the results will be |
416 |
silently truncated unless perl itself is compiled with large file |
417 |
support. |
418 |
|
419 |
Example: Print the length of /etc/passwd: |
420 |
|
421 |
aio_stat "/etc/passwd", sub { |
422 |
$_[0] and die "stat failed: $!"; |
423 |
print "size is ", -s _, "\n"; |
424 |
}; |
425 |
|
426 |
aio_statvfs $fh_or_path, $callback->($statvfs) |
427 |
Works like the POSIX "statvfs" or "fstatvfs" syscalls, depending on |
428 |
whether a file handle or path was passed. |
429 |
|
430 |
On success, the callback is passed a hash reference with the |
431 |
following members: "bsize", "frsize", "blocks", "bfree", "bavail", |
432 |
"files", "ffree", "favail", "fsid", "flag" and "namemax". On |
433 |
failure, "undef" is passed. |
434 |
|
435 |
The following POSIX IO::AIO::ST_* constants are defined: "ST_RDONLY" |
436 |
and "ST_NOSUID". |
437 |
|
438 |
The following non-POSIX IO::AIO::ST_* flag masks are defined to |
439 |
their correct value when available, or to 0 on systems that do not |
440 |
support them: "ST_NODEV", "ST_NOEXEC", "ST_SYNCHRONOUS", |
441 |
"ST_MANDLOCK", "ST_WRITE", "ST_APPEND", "ST_IMMUTABLE", |
442 |
"ST_NOATIME", "ST_NODIRATIME" and "ST_RELATIME". |
443 |
|
444 |
Example: stat "/wd" and dump out the data if successful. |
445 |
|
446 |
aio_statvfs "/wd", sub { |
447 |
my $f = $_[0] |
448 |
or die "statvfs: $!"; |
449 |
|
450 |
use Data::Dumper; |
451 |
say Dumper $f; |
452 |
}; |
453 |
|
454 |
# result: |
455 |
{ |
456 |
bsize => 1024, |
457 |
bfree => 4333064312, |
458 |
blocks => 10253828096, |
459 |
files => 2050765568, |
460 |
flag => 4096, |
461 |
favail => 2042092649, |
462 |
bavail => 4333064312, |
463 |
ffree => 2042092649, |
464 |
namemax => 255, |
465 |
frsize => 1024, |
466 |
fsid => 1810 |
467 |
} |
468 |
|
469 |
aio_utime $fh_or_path, $atime, $mtime, $callback->($status) |
470 |
Works like perl's "utime" function (including the special case of |
471 |
$atime and $mtime being undef). Fractional times are supported if |
472 |
the underlying syscalls support them. |
473 |
|
474 |
When called with a pathname, uses utimes(2) if available, otherwise |
475 |
utime(2). If called on a file descriptor, uses futimes(2) if |
476 |
available, otherwise returns ENOSYS, so this is not portable. |
477 |
|
478 |
Examples: |
479 |
|
480 |
# set atime and mtime to current time (basically touch(1)): |
481 |
aio_utime "path", undef, undef; |
482 |
# set atime to current time and mtime to beginning of the epoch: |
483 |
aio_utime "path", time, undef; # undef==0 |
484 |
|
485 |
aio_chown $fh_or_path, $uid, $gid, $callback->($status) |
486 |
Works like perl's "chown" function, except that "undef" for either |
487 |
$uid or $gid is being interpreted as "do not change" (but -1 can |
488 |
also be used). |
489 |
|
490 |
Examples: |
491 |
|
492 |
# same as "chown root path" in the shell: |
493 |
aio_chown "path", 0, -1; |
494 |
# same as above: |
495 |
aio_chown "path", 0, undef; |
496 |
|
497 |
aio_truncate $fh_or_path, $offset, $callback->($status) |
498 |
Works like truncate(2) or ftruncate(2). |
499 |
|
500 |
aio_chmod $fh_or_path, $mode, $callback->($status) |
501 |
Works like perl's "chmod" function. |
502 |
|
503 |
aio_unlink $pathname, $callback->($status) |
504 |
Asynchronously unlink (delete) a file and call the callback with the |
505 |
result code. |
506 |
|
507 |
aio_mknod $path, $mode, $dev, $callback->($status) |
508 |
[EXPERIMENTAL] |
509 |
|
510 |
Asynchronously create a device node (or fifo). See mknod(2). |
511 |
|
512 |
The only (POSIX-) portable way of calling this function is: |
513 |
|
514 |
aio_mknod $path, IO::AIO::S_IFIFO | $mode, 0, sub { ... |
515 |
|
516 |
aio_link $srcpath, $dstpath, $callback->($status) |
517 |
Asynchronously create a new link to the existing object at $srcpath |
518 |
at the path $dstpath and call the callback with the result code. |
519 |
|
520 |
aio_symlink $srcpath, $dstpath, $callback->($status) |
521 |
Asynchronously create a new symbolic link to the existing object at |
522 |
$srcpath at the path $dstpath and call the callback with the result |
523 |
code. |
524 |
|
525 |
aio_readlink $path, $callback->($link) |
526 |
Asynchronously read the symlink specified by $path and pass it to |
527 |
the callback. If an error occurs, nothing or undef gets passed to |
528 |
the callback. |
529 |
|
530 |
aio_rename $srcpath, $dstpath, $callback->($status) |
531 |
Asynchronously rename the object at $srcpath to $dstpath, just as |
532 |
rename(2) and call the callback with the result code. |
533 |
|
534 |
aio_mkdir $pathname, $mode, $callback->($status) |
535 |
Asynchronously mkdir (create) a directory and call the callback with |
536 |
the result code. $mode will be modified by the umask at the time the |
537 |
request is executed, so do not change your umask. |
538 |
|
539 |
aio_rmdir $pathname, $callback->($status) |
540 |
Asynchronously rmdir (delete) a directory and call the callback with |
541 |
the result code. |
542 |
|
543 |
aio_readdir $pathname, $callback->($entries) |
544 |
Unlike the POSIX call of the same name, "aio_readdir" reads an |
545 |
entire directory (i.e. opendir + readdir + closedir). The entries |
546 |
will not be sorted, and will NOT include the "." and ".." entries. |
547 |
|
548 |
The callback is passed a single argument which is either "undef" or |
549 |
an array-ref with the filenames. |
550 |
|
551 |
aio_readdirx $pathname, $flags, $callback->($entries, $flags) |
552 |
Quite similar to "aio_readdir", but the $flags argument allows to |
553 |
tune behaviour and output format. In case of an error, $entries will |
554 |
be "undef". |
555 |
|
556 |
The flags are a combination of the following constants, ORed |
557 |
together (the flags will also be passed to the callback, possibly |
558 |
modified): |
559 |
|
560 |
IO::AIO::READDIR_DENTS |
561 |
When this flag is off, then the callback gets an arrayref with |
562 |
of names only (as with "aio_readdir"), otherwise it gets an |
563 |
arrayref with "[$name, $type, $inode]" arrayrefs, each |
564 |
describing a single directory entry in more detail. |
565 |
|
566 |
$name is the name of the entry. |
567 |
|
568 |
$type is one of the "IO::AIO::DT_xxx" constants: |
569 |
|
570 |
"IO::AIO::DT_UNKNOWN", "IO::AIO::DT_FIFO", "IO::AIO::DT_CHR", |
571 |
"IO::AIO::DT_DIR", "IO::AIO::DT_BLK", "IO::AIO::DT_REG", |
572 |
"IO::AIO::DT_LNK", "IO::AIO::DT_SOCK", "IO::AIO::DT_WHT". |
573 |
|
574 |
"IO::AIO::DT_UNKNOWN" means just that: readdir does not know. If |
575 |
you need to know, you have to run stat yourself. Also, for speed |
576 |
reasons, the $type scalars are read-only: you can not modify |
577 |
them. |
578 |
|
579 |
$inode is the inode number (which might not be exact on systems |
580 |
with 64 bit inode numbers and 32 bit perls). This field has |
581 |
unspecified content on systems that do not deliver the inode |
582 |
information. |
583 |
|
584 |
IO::AIO::READDIR_DIRS_FIRST |
585 |
When this flag is set, then the names will be returned in an |
586 |
order where likely directories come first. This is useful when |
587 |
you need to quickly find directories, or you want to find all |
588 |
directories while avoiding to stat() each entry. |
589 |
|
590 |
If the system returns type information in readdir, then this is |
591 |
used to find directories directly. Otherwise, likely directories |
592 |
are files beginning with ".", or otherwise files with no dots, |
593 |
of which files with short names are tried first. |
594 |
|
595 |
IO::AIO::READDIR_STAT_ORDER |
596 |
When this flag is set, then the names will be returned in an |
597 |
order suitable for stat()'ing each one. That is, when you plan |
598 |
to stat() all files in the given directory, then the returned |
599 |
order will likely be fastest. |
600 |
|
601 |
If both this flag and "IO::AIO::READDIR_DIRS_FIRST" are |
602 |
specified, then the likely dirs come first, resulting in a less |
603 |
optimal stat order. |
604 |
|
605 |
IO::AIO::READDIR_FOUND_UNKNOWN |
606 |
This flag should not be set when calling "aio_readdirx". |
607 |
Instead, it is being set by "aio_readdirx", when any of the |
608 |
$type's found were "IO::AIO::DT_UNKNOWN". The absense of this |
609 |
flag therefore indicates that all $type's are known, which can |
610 |
be used to speed up some algorithms. |
611 |
|
612 |
aio_load $path, $data, $callback->($status) |
613 |
This is a composite request that tries to fully load the given file |
614 |
into memory. Status is the same as with aio_read. |
615 |
|
616 |
aio_copy $srcpath, $dstpath, $callback->($status) |
617 |
Try to copy the *file* (directories not supported as either source |
618 |
or destination) from $srcpath to $dstpath and call the callback with |
619 |
a status of 0 (ok) or -1 (error, see $!). |
620 |
|
621 |
This is a composite request that creates the destination file with |
622 |
mode 0200 and copies the contents of the source file into it using |
623 |
"aio_sendfile", followed by restoring atime, mtime, access mode and |
624 |
uid/gid, in that order. |
625 |
|
626 |
If an error occurs, the partial destination file will be unlinked, |
627 |
if possible, except when setting atime, mtime, access mode and |
628 |
uid/gid, where errors are being ignored. |
629 |
|
630 |
aio_move $srcpath, $dstpath, $callback->($status) |
631 |
Try to move the *file* (directories not supported as either source |
632 |
or destination) from $srcpath to $dstpath and call the callback with |
633 |
a status of 0 (ok) or -1 (error, see $!). |
634 |
|
635 |
This is a composite request that tries to rename(2) the file first; |
636 |
if rename fails with "EXDEV", it copies the file with "aio_copy" |
637 |
and, if that is successful, unlinks the $srcpath. |
638 |
|
639 |
aio_scandir $path, $maxreq, $callback->($dirs, $nondirs) |
640 |
Scans a directory (similar to "aio_readdir") but additionally tries |
641 |
to efficiently separate the entries of directory $path into two sets |
642 |
of names, directories you can recurse into (directories), and ones |
643 |
you cannot recurse into (everything else, including symlinks to |
644 |
directories). |
645 |
|
646 |
"aio_scandir" is a composite request that creates of many sub |
647 |
requests_ $maxreq specifies the maximum number of outstanding aio |
648 |
requests that this function generates. If it is "<= 0", then a |
649 |
suitable default will be chosen (currently 4). |
650 |
|
651 |
On error, the callback is called without arguments, otherwise it |
652 |
receives two array-refs with path-relative entry names. |
653 |
|
654 |
Example: |
655 |
|
656 |
aio_scandir $dir, 0, sub { |
657 |
my ($dirs, $nondirs) = @_; |
658 |
print "real directories: @$dirs\n"; |
659 |
print "everything else: @$nondirs\n"; |
660 |
}; |
661 |
|
662 |
Implementation notes. |
663 |
|
664 |
The "aio_readdir" cannot be avoided, but "stat()"'ing every entry |
665 |
can. |
666 |
|
667 |
If readdir returns file type information, then this is used directly |
668 |
to find directories. |
669 |
|
670 |
Otherwise, after reading the directory, the modification time, size |
671 |
etc. of the directory before and after the readdir is checked, and |
672 |
if they match (and isn't the current time), the link count will be |
673 |
used to decide how many entries are directories (if >= 2). |
674 |
Otherwise, no knowledge of the number of subdirectories will be |
675 |
assumed. |
676 |
|
677 |
Then entries will be sorted into likely directories a non-initial |
678 |
dot currently) and likely non-directories (see "aio_readdirx"). Then |
679 |
every entry plus an appended "/." will be "stat"'ed, likely |
680 |
directories first, in order of their inode numbers. If that |
681 |
succeeds, it assumes that the entry is a directory or a symlink to |
682 |
directory (which will be checked seperately). This is often faster |
683 |
than stat'ing the entry itself because filesystems might detect the |
684 |
type of the entry without reading the inode data (e.g. ext2fs |
685 |
filetype feature), even on systems that cannot return the filetype |
686 |
information on readdir. |
687 |
|
688 |
If the known number of directories (link count - 2) has been |
689 |
reached, the rest of the entries is assumed to be non-directories. |
690 |
|
691 |
This only works with certainty on POSIX (= UNIX) filesystems, which |
692 |
fortunately are the vast majority of filesystems around. |
693 |
|
694 |
It will also likely work on non-POSIX filesystems with reduced |
695 |
efficiency as those tend to return 0 or 1 as link counts, which |
696 |
disables the directory counting heuristic. |
697 |
|
698 |
aio_rmtree $path, $callback->($status) |
699 |
Delete a directory tree starting (and including) $path, return the |
700 |
status of the final "rmdir" only. This is a composite request that |
701 |
uses "aio_scandir" to recurse into and rmdir directories, and unlink |
702 |
everything else. |
703 |
|
704 |
aio_sync $callback->($status) |
705 |
Asynchronously call sync and call the callback when finished. |
706 |
|
707 |
aio_fsync $fh, $callback->($status) |
708 |
Asynchronously call fsync on the given filehandle and call the |
709 |
callback with the fsync result code. |
710 |
|
711 |
aio_fdatasync $fh, $callback->($status) |
712 |
Asynchronously call fdatasync on the given filehandle and call the |
713 |
callback with the fdatasync result code. |
714 |
|
715 |
If this call isn't available because your OS lacks it or it couldn't |
716 |
be detected, it will be emulated by calling "fsync" instead. |
717 |
|
718 |
aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status) |
719 |
Sync the data portion of the file specified by $offset and $length |
720 |
to disk (but NOT the metadata), by calling the Linux-specific |
721 |
sync_file_range call. If sync_file_range is not available or it |
722 |
returns ENOSYS, then fdatasync or fsync is being substituted. |
723 |
|
724 |
$flags can be a combination of |
725 |
"IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE", |
726 |
"IO::AIO::SYNC_FILE_RANGE_WRITE" and |
727 |
"IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER": refer to the sync_file_range |
728 |
manpage for details. |
729 |
|
730 |
aio_pathsync $path, $callback->($status) |
731 |
This request tries to open, fsync and close the given path. This is |
732 |
a composite request intended to sync directories after directory |
733 |
operations (E.g. rename). This might not work on all operating |
734 |
systems or have any specific effect, but usually it makes sure that |
735 |
directory changes get written to disc. It works for anything that |
736 |
can be opened for read-only, not just directories. |
737 |
|
738 |
Future versions of this function might fall back to other methods |
739 |
when "fsync" on the directory fails (such as calling "sync"). |
740 |
|
741 |
Passes 0 when everything went ok, and -1 on error. |
742 |
|
743 |
aio_msync $scalar, $offset = 0, $length = undef, flags = 0, |
744 |
$callback->($status) |
745 |
This is a rather advanced IO::AIO call, which only works on |
746 |
mmap(2)ed scalars (see the "IO::AIO::mmap" function, although it |
747 |
also works on data scalars managed by the Sys::Mmap or Mmap modules, |
748 |
note that the scalar must only be modified in-place while an aio |
749 |
operation is pending on it). |
750 |
|
751 |
It calls the "msync" function of your OS, if available, with the |
752 |
memory area starting at $offset in the string and ending $length |
753 |
bytes later. If $length is negative, counts from the end, and if |
754 |
$length is "undef", then it goes till the end of the string. The |
755 |
flags can be a combination of "IO::AIO::MS_ASYNC", |
756 |
"IO::AIO::MS_INVALIDATE" and "IO::AIO::MS_SYNC". |
757 |
|
758 |
aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, |
759 |
$callback->($status) |
760 |
This is a rather advanced IO::AIO call, which works best on |
761 |
mmap(2)ed scalars. |
762 |
|
763 |
It touches (reads or writes) all memory pages in the specified range |
764 |
inside the scalar. All caveats and parameters are the same as for |
765 |
"aio_msync", above, except for flags, which must be either 0 (which |
766 |
reads all pages and ensures they are instantiated) or |
767 |
"IO::AIO::MT_MODIFY", which modifies the memory page s(by reading |
768 |
and writing an octet from it, which dirties the page). |
769 |
|
770 |
aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status) |
771 |
This is a rather advanced IO::AIO call, which works best on |
772 |
mmap(2)ed scalars. |
773 |
|
774 |
It reads in all the pages of the underlying storage into memory (if |
775 |
any) and locks them, so they are not getting swapped/paged out or |
776 |
removed. |
777 |
|
778 |
If $length is undefined, then the scalar will be locked till the |
779 |
end. |
780 |
|
781 |
On systems that do not implement "mlock", this function returns -1 |
782 |
and sets errno to "ENOSYS". |
783 |
|
784 |
Note that the corresponding "munlock" is synchronous and is |
785 |
documented under "MISCELLANEOUS FUNCTIONS". |
786 |
|
787 |
Example: open a file, mmap and mlock it - both will be undone when |
788 |
$data gets destroyed. |
789 |
|
790 |
open my $fh, "<", $path or die "$path: $!"; |
791 |
my $data; |
792 |
IO::AIO::mmap $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh; |
793 |
aio_mlock $data; # mlock in background |
794 |
|
795 |
aio_mlockall $flags, $callback->($status) |
796 |
Calls the "mlockall" function with the given $flags (a combination |
797 |
of "IO::AIO::MCL_CURRENT" and "IO::AIO::MCL_FUTURE"). |
798 |
|
799 |
On systems that do not implement "mlockall", this function returns |
800 |
-1 and sets errno to "ENOSYS". |
801 |
|
802 |
Note that the corresponding "munlockall" is synchronous and is |
803 |
documented under "MISCELLANEOUS FUNCTIONS". |
804 |
|
805 |
Example: asynchronously lock all current and future pages into |
806 |
memory. |
807 |
|
808 |
aio_mlockall IO::AIO::MCL_FUTURE; |
809 |
|
810 |
aio_group $callback->(...) |
811 |
This is a very special aio request: Instead of doing something, it |
812 |
is a container for other aio requests, which is useful if you want |
813 |
to bundle many requests into a single, composite, request with a |
814 |
definite callback and the ability to cancel the whole request with |
815 |
its subrequests. |
816 |
|
817 |
Returns an object of class IO::AIO::GRP. See its documentation below |
818 |
for more info. |
819 |
|
820 |
Example: |
821 |
|
822 |
my $grp = aio_group sub { |
823 |
print "all stats done\n"; |
824 |
}; |
825 |
|
826 |
add $grp |
827 |
(aio_stat ...), |
828 |
(aio_stat ...), |
829 |
...; |
830 |
|
831 |
aio_nop $callback->() |
832 |
This is a special request - it does nothing in itself and is only |
833 |
used for side effects, such as when you want to add a dummy request |
834 |
to a group so that finishing the requests in the group depends on |
835 |
executing the given code. |
836 |
|
837 |
While this request does nothing, it still goes through the execution |
838 |
phase and still requires a worker thread. Thus, the callback will |
839 |
not be executed immediately but only after other requests in the |
840 |
queue have entered their execution phase. This can be used to |
841 |
measure request latency. |
842 |
|
843 |
IO::AIO::aio_busy $fractional_seconds, $callback->() *NOT EXPORTED* |
844 |
Mainly used for debugging and benchmarking, this aio request puts |
845 |
one of the request workers to sleep for the given time. |
846 |
|
847 |
While it is theoretically handy to have simple I/O scheduling |
848 |
requests like sleep and file handle readable/writable, the overhead |
849 |
this creates is immense (it blocks a thread for a long time) so do |
850 |
not use this function except to put your application under |
851 |
artificial I/O pressure. |
852 |
|
853 |
IO::AIO::REQ CLASS |
854 |
All non-aggregate "aio_*" functions return an object of this class when |
855 |
called in non-void context. |
856 |
|
857 |
cancel $req |
858 |
Cancels the request, if possible. Has the effect of skipping |
859 |
execution when entering the execute state and skipping calling the |
860 |
callback when entering the the result state, but will leave the |
861 |
request otherwise untouched (with the exception of readdir). That |
862 |
means that requests that currently execute will not be stopped and |
863 |
resources held by the request will not be freed prematurely. |
864 |
|
865 |
cb $req $callback->(...) |
866 |
Replace (or simply set) the callback registered to the request. |
867 |
|
868 |
IO::AIO::GRP CLASS |
869 |
This class is a subclass of IO::AIO::REQ, so all its methods apply to |
870 |
objects of this class, too. |
871 |
|
872 |
A IO::AIO::GRP object is a special request that can contain multiple |
873 |
other aio requests. |
874 |
|
875 |
You create one by calling the "aio_group" constructing function with a |
876 |
callback that will be called when all contained requests have entered |
877 |
the "done" state: |
878 |
|
879 |
my $grp = aio_group sub { |
880 |
print "all requests are done\n"; |
881 |
}; |
882 |
|
883 |
You add requests by calling the "add" method with one or more |
884 |
"IO::AIO::REQ" objects: |
885 |
|
886 |
$grp->add (aio_unlink "..."); |
887 |
|
888 |
add $grp aio_stat "...", sub { |
889 |
$_[0] or return $grp->result ("error"); |
890 |
|
891 |
# add another request dynamically, if first succeeded |
892 |
add $grp aio_open "...", sub { |
893 |
$grp->result ("ok"); |
894 |
}; |
895 |
}; |
896 |
|
897 |
This makes it very easy to create composite requests (see the source of |
898 |
"aio_move" for an application) that work and feel like simple requests. |
899 |
|
900 |
* The IO::AIO::GRP objects will be cleaned up during calls to |
901 |
"IO::AIO::poll_cb", just like any other request. |
902 |
|
903 |
* They can be canceled like any other request. Canceling will cancel |
904 |
not only the request itself, but also all requests it contains. |
905 |
|
906 |
* They can also can also be added to other IO::AIO::GRP objects. |
907 |
|
908 |
* You must not add requests to a group from within the group callback |
909 |
(or any later time). |
910 |
|
911 |
Their lifetime, simplified, looks like this: when they are empty, they |
912 |
will finish very quickly. If they contain only requests that are in the |
913 |
"done" state, they will also finish. Otherwise they will continue to |
914 |
exist. |
915 |
|
916 |
That means after creating a group you have some time to add requests |
917 |
(precisely before the callback has been invoked, which is only done |
918 |
within the "poll_cb"). And in the callbacks of those requests, you can |
919 |
add further requests to the group. And only when all those requests have |
920 |
finished will the the group itself finish. |
921 |
|
922 |
add $grp ... |
923 |
$grp->add (...) |
924 |
Add one or more requests to the group. Any type of IO::AIO::REQ can |
925 |
be added, including other groups, as long as you do not create |
926 |
circular dependencies. |
927 |
|
928 |
Returns all its arguments. |
929 |
|
930 |
$grp->cancel_subs |
931 |
Cancel all subrequests and clears any feeder, but not the group |
932 |
request itself. Useful when you queued a lot of events but got a |
933 |
result early. |
934 |
|
935 |
The group request will finish normally (you cannot add requests to |
936 |
the group). |
937 |
|
938 |
$grp->result (...) |
939 |
Set the result value(s) that will be passed to the group callback |
940 |
when all subrequests have finished and set the groups errno to the |
941 |
current value of errno (just like calling "errno" without an error |
942 |
number). By default, no argument will be passed and errno is zero. |
943 |
|
944 |
$grp->errno ([$errno]) |
945 |
Sets the group errno value to $errno, or the current value of errno |
946 |
when the argument is missing. |
947 |
|
948 |
Every aio request has an associated errno value that is restored |
949 |
when the callback is invoked. This method lets you change this value |
950 |
from its default (0). |
951 |
|
952 |
Calling "result" will also set errno, so make sure you either set $! |
953 |
before the call to "result", or call c<errno> after it. |
954 |
|
955 |
feed $grp $callback->($grp) |
956 |
Sets a feeder/generator on this group: every group can have an |
957 |
attached generator that generates requests if idle. The idea behind |
958 |
this is that, although you could just queue as many requests as you |
959 |
want in a group, this might starve other requests for a potentially |
960 |
long time. For example, "aio_scandir" might generate hundreds of |
961 |
thousands "aio_stat" requests, delaying any later requests for a |
962 |
long time. |
963 |
|
964 |
To avoid this, and allow incremental generation of requests, you can |
965 |
instead a group and set a feeder on it that generates those |
966 |
requests. The feed callback will be called whenever there are few |
967 |
enough (see "limit", below) requests active in the group itself and |
968 |
is expected to queue more requests. |
969 |
|
970 |
The feed callback can queue as many requests as it likes (i.e. "add" |
971 |
does not impose any limits). |
972 |
|
973 |
If the feed does not queue more requests when called, it will be |
974 |
automatically removed from the group. |
975 |
|
976 |
If the feed limit is 0 when this method is called, it will be set to |
977 |
2 automatically. |
978 |
|
979 |
Example: |
980 |
|
981 |
# stat all files in @files, but only ever use four aio requests concurrently: |
982 |
|
983 |
my $grp = aio_group sub { print "finished\n" }; |
984 |
limit $grp 4; |
985 |
feed $grp sub { |
986 |
my $file = pop @files |
987 |
or return; |
988 |
|
989 |
add $grp aio_stat $file, sub { ... }; |
990 |
}; |
991 |
|
992 |
limit $grp $num |
993 |
Sets the feeder limit for the group: The feeder will be called |
994 |
whenever the group contains less than this many requests. |
995 |
|
996 |
Setting the limit to 0 will pause the feeding process. |
997 |
|
998 |
The default value for the limit is 0, but note that setting a feeder |
999 |
automatically bumps it up to 2. |
1000 |
|
1001 |
SUPPORT FUNCTIONS |
1002 |
EVENT PROCESSING AND EVENT LOOP INTEGRATION |
1003 |
$fileno = IO::AIO::poll_fileno |
1004 |
Return the *request result pipe file descriptor*. This filehandle |
1005 |
must be polled for reading by some mechanism outside this module |
1006 |
(e.g. EV, Glib, select and so on, see below or the SYNOPSIS). If the |
1007 |
pipe becomes readable you have to call "poll_cb" to check the |
1008 |
results. |
1009 |
|
1010 |
See "poll_cb" for an example. |
1011 |
|
1012 |
IO::AIO::poll_cb |
1013 |
Process some outstanding events on the result pipe. You have to call |
1014 |
this regularly. Returns 0 if all events could be processed, or -1 if |
1015 |
it returned earlier for whatever reason. Returns immediately when no |
1016 |
events are outstanding. The amount of events processed depends on |
1017 |
the settings of "IO::AIO::max_poll_req" and |
1018 |
"IO::AIO::max_poll_time". |
1019 |
|
1020 |
If not all requests were processed for whatever reason, the |
1021 |
filehandle will still be ready when "poll_cb" returns, so normally |
1022 |
you don't have to do anything special to have it called later. |
1023 |
|
1024 |
Example: Install an Event watcher that automatically calls |
1025 |
IO::AIO::poll_cb with high priority (more examples can be found in |
1026 |
the SYNOPSIS section, at the top of this document): |
1027 |
|
1028 |
Event->io (fd => IO::AIO::poll_fileno, |
1029 |
poll => 'r', async => 1, |
1030 |
cb => \&IO::AIO::poll_cb); |
1031 |
|
1032 |
IO::AIO::poll_wait |
1033 |
If there are any outstanding requests and none of them in the result |
1034 |
phase, wait till the result filehandle becomes ready for reading |
1035 |
(simply does a "select" on the filehandle. This is useful if you |
1036 |
want to synchronously wait for some requests to finish). |
1037 |
|
1038 |
See "nreqs" for an example. |
1039 |
|
1040 |
IO::AIO::poll |
1041 |
Waits until some requests have been handled. |
1042 |
|
1043 |
Returns the number of requests processed, but is otherwise strictly |
1044 |
equivalent to: |
1045 |
|
1046 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
1047 |
|
1048 |
IO::AIO::flush |
1049 |
Wait till all outstanding AIO requests have been handled. |
1050 |
|
1051 |
Strictly equivalent to: |
1052 |
|
1053 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
1054 |
while IO::AIO::nreqs; |
1055 |
|
1056 |
IO::AIO::max_poll_reqs $nreqs |
1057 |
IO::AIO::max_poll_time $seconds |
1058 |
These set the maximum number of requests (default 0, meaning |
1059 |
infinity) that are being processed by "IO::AIO::poll_cb" in one |
1060 |
call, respectively the maximum amount of time (default 0, meaning |
1061 |
infinity) spent in "IO::AIO::poll_cb" to process requests (more |
1062 |
correctly the mininum amount of time "poll_cb" is allowed to use). |
1063 |
|
1064 |
Setting "max_poll_time" to a non-zero value creates an overhead of |
1065 |
one syscall per request processed, which is not normally a problem |
1066 |
unless your callbacks are really really fast or your OS is really |
1067 |
really slow (I am not mentioning Solaris here). Using |
1068 |
"max_poll_reqs" incurs no overhead. |
1069 |
|
1070 |
Setting these is useful if you want to ensure some level of |
1071 |
interactiveness when perl is not fast enough to process all requests |
1072 |
in time. |
1073 |
|
1074 |
For interactive programs, values such as 0.01 to 0.1 should be fine. |
1075 |
|
1076 |
Example: Install an Event watcher that automatically calls |
1077 |
IO::AIO::poll_cb with low priority, to ensure that other parts of |
1078 |
the program get the CPU sometimes even under high AIO load. |
1079 |
|
1080 |
# try not to spend much more than 0.1s in poll_cb |
1081 |
IO::AIO::max_poll_time 0.1; |
1082 |
|
1083 |
# use a low priority so other tasks have priority |
1084 |
Event->io (fd => IO::AIO::poll_fileno, |
1085 |
poll => 'r', nice => 1, |
1086 |
cb => &IO::AIO::poll_cb); |
1087 |
|
1088 |
CONTROLLING THE NUMBER OF THREADS |
1089 |
IO::AIO::min_parallel $nthreads |
1090 |
Set the minimum number of AIO threads to $nthreads. The current |
1091 |
default is 8, which means eight asynchronous operations can execute |
1092 |
concurrently at any one time (the number of outstanding requests, |
1093 |
however, is unlimited). |
1094 |
|
1095 |
IO::AIO starts threads only on demand, when an AIO request is queued |
1096 |
and no free thread exists. Please note that queueing up a hundred |
1097 |
requests can create demand for a hundred threads, even if it turns |
1098 |
out that everything is in the cache and could have been processed |
1099 |
faster by a single thread. |
1100 |
|
1101 |
It is recommended to keep the number of threads relatively low, as |
1102 |
some Linux kernel versions will scale negatively with the number of |
1103 |
threads (higher parallelity => MUCH higher latency). With current |
1104 |
Linux 2.6 versions, 4-32 threads should be fine. |
1105 |
|
1106 |
Under most circumstances you don't need to call this function, as |
1107 |
the module selects a default that is suitable for low to moderate |
1108 |
load. |
1109 |
|
1110 |
IO::AIO::max_parallel $nthreads |
1111 |
Sets the maximum number of AIO threads to $nthreads. If more than |
1112 |
the specified number of threads are currently running, this function |
1113 |
kills them. This function blocks until the limit is reached. |
1114 |
|
1115 |
While $nthreads are zero, aio requests get queued but not executed |
1116 |
until the number of threads has been increased again. |
1117 |
|
1118 |
This module automatically runs "max_parallel 0" at program end, to |
1119 |
ensure that all threads are killed and that there are no outstanding |
1120 |
requests. |
1121 |
|
1122 |
Under normal circumstances you don't need to call this function. |
1123 |
|
1124 |
IO::AIO::max_idle $nthreads |
1125 |
Limit the number of threads (default: 4) that are allowed to idle |
1126 |
(i.e., threads that did not get a request to process within 10 |
1127 |
seconds). That means if a thread becomes idle while $nthreads other |
1128 |
threads are also idle, it will free its resources and exit. |
1129 |
|
1130 |
This is useful when you allow a large number of threads (e.g. 100 or |
1131 |
1000) to allow for extremely high load situations, but want to free |
1132 |
resources under normal circumstances (1000 threads can easily |
1133 |
consume 30MB of RAM). |
1134 |
|
1135 |
The default is probably ok in most situations, especially if thread |
1136 |
creation is fast. If thread creation is very slow on your system you |
1137 |
might want to use larger values. |
1138 |
|
1139 |
IO::AIO::max_outstanding $maxreqs |
1140 |
This is a very bad function to use in interactive programs because |
1141 |
it blocks, and a bad way to reduce concurrency because it is |
1142 |
inexact: Better use an "aio_group" together with a feed callback. |
1143 |
|
1144 |
Sets the maximum number of outstanding requests to $nreqs. If you do |
1145 |
queue up more than this number of requests, the next call to the |
1146 |
"poll_cb" (and "poll_some" and other functions calling "poll_cb") |
1147 |
function will block until the limit is no longer exceeded. |
1148 |
|
1149 |
The default value is very large, so there is no practical limit on |
1150 |
the number of outstanding requests. |
1151 |
|
1152 |
You can still queue as many requests as you want. Therefore, |
1153 |
"max_outstanding" is mainly useful in simple scripts (with low |
1154 |
values) or as a stop gap to shield against fatal memory overflow |
1155 |
(with large values). |
1156 |
|
1157 |
STATISTICAL INFORMATION |
1158 |
IO::AIO::nreqs |
1159 |
Returns the number of requests currently in the ready, execute or |
1160 |
pending states (i.e. for which their callback has not been invoked |
1161 |
yet). |
1162 |
|
1163 |
Example: wait till there are no outstanding requests anymore: |
1164 |
|
1165 |
IO::AIO::poll_wait, IO::AIO::poll_cb |
1166 |
while IO::AIO::nreqs; |
1167 |
|
1168 |
IO::AIO::nready |
1169 |
Returns the number of requests currently in the ready state (not yet |
1170 |
executed). |
1171 |
|
1172 |
IO::AIO::npending |
1173 |
Returns the number of requests currently in the pending state |
1174 |
(executed, but not yet processed by poll_cb). |
1175 |
|
1176 |
MISCELLANEOUS FUNCTIONS |
1177 |
IO::AIO implements some functions that might be useful, but are not |
1178 |
asynchronous. |
1179 |
|
1180 |
IO::AIO::sendfile $ofh, $ifh, $offset, $count |
1181 |
Calls the "eio_sendfile_sync" function, which is like |
1182 |
"aio_sendfile", but is blocking (this makes most sense if you know |
1183 |
the input data is likely cached already and the output filehandle is |
1184 |
set to non-blocking operations). |
1185 |
|
1186 |
Returns the number of bytes copied, or -1 on error. |
1187 |
|
1188 |
IO::AIO::fadvise $fh, $offset, $len, $advice |
1189 |
Simply calls the "posix_fadvise" function (see its manpage for |
1190 |
details). The following advice constants are avaiable: |
1191 |
"IO::AIO::FADV_NORMAL", "IO::AIO::FADV_SEQUENTIAL", |
1192 |
"IO::AIO::FADV_RANDOM", "IO::AIO::FADV_NOREUSE", |
1193 |
"IO::AIO::FADV_WILLNEED", "IO::AIO::FADV_DONTNEED". |
1194 |
|
1195 |
On systems that do not implement "posix_fadvise", this function |
1196 |
returns ENOSYS, otherwise the return value of "posix_fadvise". |
1197 |
|
1198 |
IO::AIO::madvise $scalar, $offset, $len, $advice |
1199 |
Simply calls the "posix_madvise" function (see its manpage for |
1200 |
details). The following advice constants are avaiable: |
1201 |
"IO::AIO::MADV_NORMAL", "IO::AIO::MADV_SEQUENTIAL", |
1202 |
"IO::AIO::MADV_RANDOM", "IO::AIO::MADV_WILLNEED", |
1203 |
"IO::AIO::MADV_DONTNEED". |
1204 |
|
1205 |
On systems that do not implement "posix_madvise", this function |
1206 |
returns ENOSYS, otherwise the return value of "posix_madvise". |
1207 |
|
1208 |
IO::AIO::mprotect $scalar, $offset, $len, $protect |
1209 |
Simply calls the "mprotect" function on the preferably AIO::mmap'ed |
1210 |
$scalar (see its manpage for details). The following protect |
1211 |
constants are avaiable: "IO::AIO::PROT_NONE", "IO::AIO::PROT_READ", |
1212 |
"IO::AIO::PROT_WRITE", "IO::AIO::PROT_EXEC". |
1213 |
|
1214 |
On systems that do not implement "mprotect", this function returns |
1215 |
ENOSYS, otherwise the return value of "mprotect". |
1216 |
|
1217 |
IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset] |
1218 |
Memory-maps a file (or anonymous memory range) and attaches it to |
1219 |
the given $scalar, which will act like a string scalar. |
1220 |
|
1221 |
The only operations allowed on the scalar are "substr"/"vec" that |
1222 |
don't change the string length, and most read-only operations such |
1223 |
as copying it or searching it with regexes and so on. |
1224 |
|
1225 |
Anything else is unsafe and will, at best, result in memory leaks. |
1226 |
|
1227 |
The memory map associated with the $scalar is automatically removed |
1228 |
when the $scalar is destroyed, or when the "IO::AIO::mmap" or |
1229 |
"IO::AIO::munmap" functions are called. |
1230 |
|
1231 |
This calls the "mmap"(2) function internally. See your system's |
1232 |
manual page for details on the $length, $prot and $flags parameters. |
1233 |
|
1234 |
The $length must be larger than zero and smaller than the actual |
1235 |
filesize. |
1236 |
|
1237 |
$prot is a combination of "IO::AIO::PROT_NONE", |
1238 |
"IO::AIO::PROT_EXEC", "IO::AIO::PROT_READ" and/or |
1239 |
"IO::AIO::PROT_WRITE", |
1240 |
|
1241 |
$flags can be a combination of "IO::AIO::MAP_SHARED" or |
1242 |
"IO::AIO::MAP_PRIVATE", or a number of system-specific flags (when |
1243 |
not available, the are defined as 0): "IO::AIO::MAP_ANONYMOUS" |
1244 |
(which is set to "MAP_ANON" if your system only provides this |
1245 |
constant), "IO::AIO::MAP_HUGETLB", "IO::AIO::MAP_LOCKED", |
1246 |
"IO::AIO::MAP_NORESERVE", "IO::AIO::MAP_POPULATE" or |
1247 |
"IO::AIO::MAP_NONBLOCK" |
1248 |
|
1249 |
If $fh is "undef", then a file descriptor of -1 is passed. |
1250 |
|
1251 |
$offset is the offset from the start of the file - it generally must |
1252 |
be a multiple of "IO::AIO::PAGESIZE" and defaults to 0. |
1253 |
|
1254 |
Example: |
1255 |
|
1256 |
use Digest::MD5; |
1257 |
use IO::AIO; |
1258 |
|
1259 |
open my $fh, "<verybigfile" |
1260 |
or die "$!"; |
1261 |
|
1262 |
IO::AIO::mmap my $data, -s $fh, IO::AIO::PROT_READ, IO::AIO::MAP_SHARED, $fh |
1263 |
or die "verybigfile: $!"; |
1264 |
|
1265 |
my $fast_md5 = md5 $data; |
1266 |
|
1267 |
IO::AIO::munmap $scalar |
1268 |
Removes a previous mmap and undefines the $scalar. |
1269 |
|
1270 |
IO::AIO::munlock $scalar, $offset = 0, $length = undef |
1271 |
Calls the "munlock" function, undoing the effects of a previous |
1272 |
"aio_mlock" call (see its description for details). |
1273 |
|
1274 |
IO::AIO::munlockall |
1275 |
Calls the "munlockall" function. |
1276 |
|
1277 |
On systems that do not implement "munlockall", this function returns |
1278 |
ENOSYS, otherwise the return value of "munlockall". |
1279 |
|
1280 |
EVENT LOOP INTEGRATION |
1281 |
It is recommended to use AnyEvent::AIO to integrate IO::AIO |
1282 |
automatically into many event loops: |
1283 |
|
1284 |
# AnyEvent integration (EV, Event, Glib, Tk, POE, urxvt, pureperl...) |
1285 |
use AnyEvent::AIO; |
1286 |
|
1287 |
You can also integrate IO::AIO manually into many event loops, here are |
1288 |
some examples of how to do this: |
1289 |
|
1290 |
# EV integration |
1291 |
my $aio_w = EV::io IO::AIO::poll_fileno, EV::READ, \&IO::AIO::poll_cb; |
1292 |
|
1293 |
# Event integration |
1294 |
Event->io (fd => IO::AIO::poll_fileno, |
1295 |
poll => 'r', |
1296 |
cb => \&IO::AIO::poll_cb); |
1297 |
|
1298 |
# Glib/Gtk2 integration |
1299 |
add_watch Glib::IO IO::AIO::poll_fileno, |
1300 |
in => sub { IO::AIO::poll_cb; 1 }; |
1301 |
|
1302 |
# Tk integration |
1303 |
Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "", |
1304 |
readable => \&IO::AIO::poll_cb); |
1305 |
|
1306 |
# Danga::Socket integration |
1307 |
Danga::Socket->AddOtherFds (IO::AIO::poll_fileno => |
1308 |
\&IO::AIO::poll_cb); |
1309 |
|
1310 |
FORK BEHAVIOUR |
1311 |
This module should do "the right thing" when the process using it forks: |
1312 |
|
1313 |
Before the fork, IO::AIO enters a quiescent state where no requests can |
1314 |
be added in other threads and no results will be processed. After the |
1315 |
fork the parent simply leaves the quiescent state and continues |
1316 |
request/result processing, while the child frees the request/result |
1317 |
queue (so that the requests started before the fork will only be handled |
1318 |
in the parent). Threads will be started on demand until the limit set in |
1319 |
the parent process has been reached again. |
1320 |
|
1321 |
In short: the parent will, after a short pause, continue as if fork had |
1322 |
not been called, while the child will act as if IO::AIO has not been |
1323 |
used yet. |
1324 |
|
1325 |
MEMORY USAGE |
1326 |
Per-request usage: |
1327 |
|
1328 |
Each aio request uses - depending on your architecture - around 100-200 |
1329 |
bytes of memory. In addition, stat requests need a stat buffer (possibly |
1330 |
a few hundred bytes), readdir requires a result buffer and so on. Perl |
1331 |
scalars and other data passed into aio requests will also be locked and |
1332 |
will consume memory till the request has entered the done state. |
1333 |
|
1334 |
This is not awfully much, so queuing lots of requests is not usually a |
1335 |
problem. |
1336 |
|
1337 |
Per-thread usage: |
1338 |
|
1339 |
In the execution phase, some aio requests require more memory for |
1340 |
temporary buffers, and each thread requires a stack and other data |
1341 |
structures (usually around 16k-128k, depending on the OS). |
1342 |
|
1343 |
KNOWN BUGS |
1344 |
Known bugs will be fixed in the next release. |
1345 |
|
1346 |
SEE ALSO |
1347 |
AnyEvent::AIO for easy integration into event loops, Coro::AIO for a |
1348 |
more natural syntax. |
1349 |
|
1350 |
AUTHOR |
1351 |
Marc Lehmann <schmorp@schmorp.de> |
1352 |
http://home.schmorp.de/ |
1353 |
|