ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/IO-AIO/AIO.pm
(Generate patch)

Comparing IO-AIO/AIO.pm (file contents):
Revision 1.183 by root, Sun Sep 12 03:40:05 2010 UTC vs.
Revision 1.210 by root, Tue Sep 27 01:43:03 2011 UTC

168use common::sense; 168use common::sense;
169 169
170use base 'Exporter'; 170use base 'Exporter';
171 171
172BEGIN { 172BEGIN {
173 our $VERSION = '3.65'; 173 our $VERSION = '4.0';
174 174
175 our @AIO_REQ = qw(aio_sendfile aio_read aio_write aio_open aio_close 175 our @AIO_REQ = qw(aio_sendfile aio_read aio_write aio_open aio_close
176 aio_stat aio_lstat aio_unlink aio_rmdir aio_readdir aio_readdirx 176 aio_stat aio_lstat aio_unlink aio_rmdir aio_readdir aio_readdirx
177 aio_scandir aio_symlink aio_readlink aio_sync aio_fsync 177 aio_scandir aio_symlink aio_readlink aio_realpath aio_sync
178 aio_fdatasync aio_sync_file_range aio_pathsync aio_readahead 178 aio_fsync aio_syncfs aio_fdatasync aio_sync_file_range aio_fallocate
179 aio_pathsync aio_readahead
179 aio_rename aio_link aio_move aio_copy aio_group 180 aio_rename aio_link aio_move aio_copy aio_group
180 aio_nop aio_mknod aio_load aio_rmtree aio_mkdir aio_chown 181 aio_nop aio_mknod aio_load aio_rmtree aio_mkdir aio_chown
181 aio_chmod aio_utime aio_truncate 182 aio_chmod aio_utime aio_truncate
182 aio_msync aio_mtouch aio_mlock aio_mlockall 183 aio_msync aio_mtouch aio_mlock aio_mlockall
183 aio_statvfs); 184 aio_statvfs
185 aio_wd);
184 186
185 our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice)); 187 our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice));
186 our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush 188 our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
187 min_parallel max_parallel max_idle 189 min_parallel max_parallel max_idle idle_timeout
188 nreqs nready npending nthreads 190 nreqs nready npending nthreads
189 max_poll_time max_poll_reqs 191 max_poll_time max_poll_reqs
190 sendfile fadvise madvise 192 sendfile fadvise madvise
191 mmap munmap munlock munlockall); 193 mmap munmap munlock munlockall);
192 194
204 206
205This section simply lists the prototypes of the most important functions 207This section simply lists the prototypes of the most important functions
206for quick reference. See the following sections for function-by-function 208for quick reference. See the following sections for function-by-function
207documentation. 209documentation.
208 210
211 aio_wd $pathname, $callback->($wd)
209 aio_open $pathname, $flags, $mode, $callback->($fh) 212 aio_open $pathname, $flags, $mode, $callback->($fh)
210 aio_close $fh, $callback->($status) 213 aio_close $fh, $callback->($status)
211 aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval) 214 aio_read $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
212 aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval) 215 aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)
213 aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval) 216 aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)
218 aio_utime $fh_or_path, $atime, $mtime, $callback->($status) 221 aio_utime $fh_or_path, $atime, $mtime, $callback->($status)
219 aio_chown $fh_or_path, $uid, $gid, $callback->($status) 222 aio_chown $fh_or_path, $uid, $gid, $callback->($status)
220 aio_truncate $fh_or_path, $offset, $callback->($status) 223 aio_truncate $fh_or_path, $offset, $callback->($status)
221 aio_chmod $fh_or_path, $mode, $callback->($status) 224 aio_chmod $fh_or_path, $mode, $callback->($status)
222 aio_unlink $pathname, $callback->($status) 225 aio_unlink $pathname, $callback->($status)
223 aio_mknod $path, $mode, $dev, $callback->($status) 226 aio_mknod $pathname, $mode, $dev, $callback->($status)
224 aio_link $srcpath, $dstpath, $callback->($status) 227 aio_link $srcpath, $dstpath, $callback->($status)
225 aio_symlink $srcpath, $dstpath, $callback->($status) 228 aio_symlink $srcpath, $dstpath, $callback->($status)
226 aio_readlink $path, $callback->($link) 229 aio_readlink $pathname, $callback->($link)
230 aio_realpath $pathname, $callback->($link)
227 aio_rename $srcpath, $dstpath, $callback->($status) 231 aio_rename $srcpath, $dstpath, $callback->($status)
228 aio_mkdir $pathname, $mode, $callback->($status) 232 aio_mkdir $pathname, $mode, $callback->($status)
229 aio_rmdir $pathname, $callback->($status) 233 aio_rmdir $pathname, $callback->($status)
230 aio_readdir $pathname, $callback->($entries) 234 aio_readdir $pathname, $callback->($entries)
231 aio_readdirx $pathname, $flags, $callback->($entries, $flags) 235 aio_readdirx $pathname, $flags, $callback->($entries, $flags)
232 IO::AIO::READDIR_DENTS IO::AIO::READDIR_DIRS_FIRST 236 IO::AIO::READDIR_DENTS IO::AIO::READDIR_DIRS_FIRST
233 IO::AIO::READDIR_STAT_ORDER IO::AIO::READDIR_FOUND_UNKNOWN 237 IO::AIO::READDIR_STAT_ORDER IO::AIO::READDIR_FOUND_UNKNOWN
234 aio_load $path, $data, $callback->($status) 238 aio_load $pathname, $data, $callback->($status)
235 aio_copy $srcpath, $dstpath, $callback->($status) 239 aio_copy $srcpath, $dstpath, $callback->($status)
236 aio_move $srcpath, $dstpath, $callback->($status) 240 aio_move $srcpath, $dstpath, $callback->($status)
237 aio_scandir $path, $maxreq, $callback->($dirs, $nondirs) 241 aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
238 aio_rmtree $path, $callback->($status) 242 aio_rmtree $pathname, $callback->($status)
239 aio_sync $callback->($status) 243 aio_sync $callback->($status)
244 aio_syncfs $fh, $callback->($status)
240 aio_fsync $fh, $callback->($status) 245 aio_fsync $fh, $callback->($status)
241 aio_fdatasync $fh, $callback->($status) 246 aio_fdatasync $fh, $callback->($status)
242 aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status) 247 aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
243 aio_pathsync $path, $callback->($status) 248 aio_pathsync $pathname, $callback->($status)
244 aio_msync $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status) 249 aio_msync $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
245 aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status) 250 aio_mtouch $scalar, $offset = 0, $length = undef, flags = 0, $callback->($status)
246 aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status) 251 aio_mlock $scalar, $offset = 0, $length = undef, $callback->($status)
247 aio_mlockall $flags, $callback->($status) 252 aio_mlockall $flags, $callback->($status)
248 aio_group $callback->(...) 253 aio_group $callback->(...)
258 IO::AIO::max_poll_reqs $nreqs 263 IO::AIO::max_poll_reqs $nreqs
259 IO::AIO::max_poll_time $seconds 264 IO::AIO::max_poll_time $seconds
260 IO::AIO::min_parallel $nthreads 265 IO::AIO::min_parallel $nthreads
261 IO::AIO::max_parallel $nthreads 266 IO::AIO::max_parallel $nthreads
262 IO::AIO::max_idle $nthreads 267 IO::AIO::max_idle $nthreads
268 IO::AIO::idle_timeout $seconds
263 IO::AIO::max_outstanding $maxreqs 269 IO::AIO::max_outstanding $maxreqs
264 IO::AIO::nreqs 270 IO::AIO::nreqs
265 IO::AIO::nready 271 IO::AIO::nready
266 IO::AIO::npending 272 IO::AIO::npending
267 273
268 IO::AIO::sendfile $ofh, $ifh, $offset, $count 274 IO::AIO::sendfile $ofh, $ifh, $offset, $count
269 IO::AIO::fadvise $fh, $offset, $len, $advice 275 IO::AIO::fadvise $fh, $offset, $len, $advice
276 IO::AIO::madvise $scalar, $offset, $length, $advice
277 IO::AIO::mprotect $scalar, $offset, $length, $protect
270 IO::AIO::munlock $scalar, $offset = 0, $length = undef 278 IO::AIO::munlock $scalar, $offset = 0, $length = undef
271 IO::AIO::munlockall 279 IO::AIO::munlockall
272 280
273=head2 AIO REQUEST FUNCTIONS 281=head2 AIO REQUEST FUNCTIONS
274 282
284internally until the request has finished. 292internally until the request has finished.
285 293
286All functions return request objects of type L<IO::AIO::REQ> that allow 294All functions return request objects of type L<IO::AIO::REQ> that allow
287further manipulation of those requests while they are in-flight. 295further manipulation of those requests while they are in-flight.
288 296
289The pathnames you pass to these routines I<must> be absolute and 297The pathnames you pass to these routines I<should> be absolute. The
290encoded as octets. The reason for the former is that at the time the 298reason for this is that at the time the request is being executed, the
291request is being executed, the current working directory could have 299current working directory could have changed. Alternatively, you can make
292changed. Alternatively, you can make sure that you never change the 300sure that you never change the current working directory anywhere in
293current working directory anywhere in the program and then use relative 301the program and then use relative paths. Lastly, you can take advantage
294paths. 302of IO::AIOs working directory abstraction - see the description of the
303C<IO::AIO::WD> class later in this document.
295 304
296To encode pathnames as octets, either make sure you either: a) always pass 305To encode pathnames as octets, either make sure you either: a) always pass
297in filenames you got from outside (command line, readdir etc.) without 306in filenames you got from outside (command line, readdir etc.) without
298tinkering, b) are ASCII or ISO 8859-1, c) use the Encode module and encode 307tinkering, b) are ASCII or ISO 8859-1, c) use the Encode module and encode
299your pathnames to the locale (or other) encoding in effect in the user 308your pathnames to the locale (or other) encoding in effect in the user
365 } else { 374 } else {
366 die "open failed: $!\n"; 375 die "open failed: $!\n";
367 } 376 }
368 }; 377 };
369 378
379In addition to all the common open modes/flags (C<O_RDONLY>, C<O_WRONLY>,
380C<O_RDWR>, C<O_CREAT>, C<O_TRUNC>, C<O_EXCL> and C<O_APPEND>), the
381following POSIX and non-POSIX constants are available (missing ones on
382your system are, as usual, C<0>):
383
384C<O_ASYNC>, C<O_DIRECT>, C<O_NOATIME>, C<O_CLOEXEC>, C<O_NOCTTY>, C<O_NOFOLLOW>,
385C<O_NONBLOCK>, C<O_EXEC>, C<O_SEARCH>, C<O_DIRECTORY>, C<O_DSYNC>,
386C<O_RSYNC>, C<O_SYNC> and C<O_TTY_INIT>.
387
370 388
371=item aio_close $fh, $callback->($status) 389=item aio_close $fh, $callback->($status)
372 390
373Asynchronously close a file and call the callback with the result 391Asynchronously close a file and call the callback with the result
374code. 392code.
424 442
425Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts 443Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
426reading at byte offset C<$in_offset>, and starts writing at the current 444reading at byte offset C<$in_offset>, and starts writing at the current
427file offset of C<$out_fh>. Because of that, it is not safe to issue more 445file offset of C<$out_fh>. Because of that, it is not safe to issue more
428than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each 446than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
429other. 447other. The same C<$in_fh> works fine though, as this function does not
448move or use the file offset of C<$in_fh>.
430 449
450Please note that C<aio_sendfile> can read more bytes from C<$in_fh> than
451are written, and there is no way to find out how many more bytes have been
452read from C<aio_sendfile> alone, as C<aio_sendfile> only provides the
453number of bytes written to C<$out_fh>. Only if the result value equals
454C<$length> one can assume that C<$length> bytes have been read.
455
456Unlike with other C<aio_> functions, it makes a lot of sense to use
457C<aio_sendfile> on non-blocking sockets, as long as one end (typically
458the C<$in_fh>) is a file - the file I/O will then be asynchronous, while
459the socket I/O will be non-blocking. Note, however, that you can run
460into a trap where C<aio_sendfile> reads some data with readahead, then
461fails to write all data, and when the socket is ready the next time, the
462data in the cache is already lost, forcing C<aio_sendfile> to again hit
463the disk. Explicit C<aio_read> + C<aio_write> let's you better control
464resource usage.
465
431This call tries to make use of a native C<sendfile> syscall to provide 466This call tries to make use of a native C<sendfile>-like syscall to
432zero-copy operation. For this to work, C<$out_fh> should refer to a 467provide zero-copy operation. For this to work, C<$out_fh> should refer to
433socket, and C<$in_fh> should refer to an mmap'able file. 468a socket, and C<$in_fh> should refer to an mmap'able file.
434 469
435If a native sendfile cannot be found or it fails with C<ENOSYS>, 470If a native sendfile cannot be found or it fails with C<ENOSYS>,
436C<ENOTSUP>, C<EOPNOTSUPP>, C<EAFNOSUPPORT>, C<EPROTOTYPE> or C<ENOTSOCK>, 471C<EINVAL>, C<ENOTSUP>, C<EOPNOTSUPP>, C<EAFNOSUPPORT>, C<EPROTOTYPE> or
437it will be emulated, so you can call C<aio_sendfile> on any type of 472C<ENOTSOCK>, it will be emulated, so you can call C<aio_sendfile> on any
438filehandle regardless of the limitations of the operating system. 473type of filehandle regardless of the limitations of the operating system.
439 474
440Please note, however, that C<aio_sendfile> can read more bytes from 475As native sendfile syscalls (as practically any non-POSIX interface hacked
441C<$in_fh> than are written, and there is no way to find out how many 476together in a hurry to improve benchmark numbers) tend to be rather buggy
442bytes have been read from C<aio_sendfile> alone, as C<aio_sendfile> only 477on many systems, this implementation tries to work around some known bugs
443provides the number of bytes written to C<$out_fh>. Only if the result 478in Linux and FreeBSD kernels (probably others, too), but that might fail,
444value equals C<$length> one can assume that C<$length> bytes have been 479so you really really should check the return value of C<aio_sendfile> -
445read. 480fewre bytes than expected might have been transferred.
446 481
447 482
448=item aio_readahead $fh,$offset,$length, $callback->($retval) 483=item aio_readahead $fh,$offset,$length, $callback->($retval)
449 484
450C<aio_readahead> populates the page cache with data from a file so that 485C<aio_readahead> populates the page cache with data from a file so that
472for an explanation. 507for an explanation.
473 508
474Currently, the stats are always 64-bit-stats, i.e. instead of returning an 509Currently, the stats are always 64-bit-stats, i.e. instead of returning an
475error when stat'ing a large file, the results will be silently truncated 510error when stat'ing a large file, the results will be silently truncated
476unless perl itself is compiled with large file support. 511unless perl itself is compiled with large file support.
512
513To help interpret the mode and dev/rdev stat values, IO::AIO offers the
514following constants and functions (if not implemented, the constants will
515be C<0> and the functions will either C<croak> or fall back on traditional
516behaviour).
517
518C<S_IFMT>, C<S_IFIFO>, C<S_IFCHR>, C<S_IFBLK>, C<S_IFLNK>, C<S_IFREG>,
519C<S_IFDIR>, C<S_IFWHT>, C<S_IFSOCK>, C<IO::AIO::major $dev_t>,
520C<IO::AIO::minor $dev_t>, C<IO::AIO::makedev $major, $minor>.
477 521
478Example: Print the length of F</etc/passwd>: 522Example: Print the length of F</etc/passwd>:
479 523
480 aio_stat "/etc/passwd", sub { 524 aio_stat "/etc/passwd", sub {
481 $_[0] and die "stat failed: $!"; 525 $_[0] and die "stat failed: $!";
573 617
574Asynchronously unlink (delete) a file and call the callback with the 618Asynchronously unlink (delete) a file and call the callback with the
575result code. 619result code.
576 620
577 621
578=item aio_mknod $path, $mode, $dev, $callback->($status) 622=item aio_mknod $pathname, $mode, $dev, $callback->($status)
579 623
580[EXPERIMENTAL] 624[EXPERIMENTAL]
581 625
582Asynchronously create a device node (or fifo). See mknod(2). 626Asynchronously create a device node (or fifo). See mknod(2).
583 627
584The only (POSIX-) portable way of calling this function is: 628The only (POSIX-) portable way of calling this function is:
585 629
586 aio_mknod $path, IO::AIO::S_IFIFO | $mode, 0, sub { ... 630 aio_mknod $pathname, IO::AIO::S_IFIFO | $mode, 0, sub { ...
587 631
632See C<aio_stat> for info about some potentially helpful extra constants
633and functions.
588 634
589=item aio_link $srcpath, $dstpath, $callback->($status) 635=item aio_link $srcpath, $dstpath, $callback->($status)
590 636
591Asynchronously create a new link to the existing object at C<$srcpath> at 637Asynchronously create a new link to the existing object at C<$srcpath> at
592the path C<$dstpath> and call the callback with the result code. 638the path C<$dstpath> and call the callback with the result code.
596 642
597Asynchronously create a new symbolic link to the existing object at C<$srcpath> at 643Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
598the path C<$dstpath> and call the callback with the result code. 644the path C<$dstpath> and call the callback with the result code.
599 645
600 646
601=item aio_readlink $path, $callback->($link) 647=item aio_readlink $pathname, $callback->($link)
602 648
603Asynchronously read the symlink specified by C<$path> and pass it to 649Asynchronously read the symlink specified by C<$path> and pass it to
604the callback. If an error occurs, nothing or undef gets passed to the 650the callback. If an error occurs, nothing or undef gets passed to the
605callback. 651callback.
606 652
607 653
654=item aio_realpath $pathname, $callback->($path)
655
656Asynchronously make the path absolute and resolve any symlinks in
657C<$path>. The resulting path only consists of directories (Same as
658L<Cwd::realpath>).
659
660This request can be used to get the absolute path of the current working
661directory by passing it a path of F<.> (a single dot).
662
663
608=item aio_rename $srcpath, $dstpath, $callback->($status) 664=item aio_rename $srcpath, $dstpath, $callback->($status)
609 665
610Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as 666Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
611rename(2) and call the callback with the result code. 667rename(2) and call the callback with the result code.
612 668
634array-ref with the filenames. 690array-ref with the filenames.
635 691
636 692
637=item aio_readdirx $pathname, $flags, $callback->($entries, $flags) 693=item aio_readdirx $pathname, $flags, $callback->($entries, $flags)
638 694
639Quite similar to C<aio_readdir>, but the C<$flags> argument allows to tune 695Quite similar to C<aio_readdir>, but the C<$flags> argument allows one to
640behaviour and output format. In case of an error, C<$entries> will be 696tune behaviour and output format. In case of an error, C<$entries> will be
641C<undef>. 697C<undef>.
642 698
643The flags are a combination of the following constants, ORed together (the 699The flags are a combination of the following constants, ORed together (the
644flags will also be passed to the callback, possibly modified): 700flags will also be passed to the callback, possibly modified):
645 701
646=over 4 702=over 4
647 703
648=item IO::AIO::READDIR_DENTS 704=item IO::AIO::READDIR_DENTS
649 705
650When this flag is off, then the callback gets an arrayref with of names 706When this flag is off, then the callback gets an arrayref consisting of
651only (as with C<aio_readdir>), otherwise it gets an arrayref with 707names only (as with C<aio_readdir>), otherwise it gets an arrayref with
652C<[$name, $type, $inode]> arrayrefs, each describing a single directory 708C<[$name, $type, $inode]> arrayrefs, each describing a single directory
653entry in more detail. 709entry in more detail.
654 710
655C<$name> is the name of the entry. 711C<$name> is the name of the entry.
656 712
669systems that do not deliver the inode information. 725systems that do not deliver the inode information.
670 726
671=item IO::AIO::READDIR_DIRS_FIRST 727=item IO::AIO::READDIR_DIRS_FIRST
672 728
673When this flag is set, then the names will be returned in an order where 729When this flag is set, then the names will be returned in an order where
674likely directories come first. This is useful when you need to quickly 730likely directories come first, in optimal stat order. This is useful when
675find directories, or you want to find all directories while avoiding to 731you need to quickly find directories, or you want to find all directories
676stat() each entry. 732while avoiding to stat() each entry.
677 733
678If the system returns type information in readdir, then this is used 734If the system returns type information in readdir, then this is used
679to find directories directly. Otherwise, likely directories are files 735to find directories directly. Otherwise, likely directories are names
680beginning with ".", or otherwise files with no dots, of which files with 736beginning with ".", or otherwise names with no dots, of which names with
681short names are tried first. 737short names are tried first.
682 738
683=item IO::AIO::READDIR_STAT_ORDER 739=item IO::AIO::READDIR_STAT_ORDER
684 740
685When this flag is set, then the names will be returned in an order 741When this flag is set, then the names will be returned in an order
692 748
693=item IO::AIO::READDIR_FOUND_UNKNOWN 749=item IO::AIO::READDIR_FOUND_UNKNOWN
694 750
695This flag should not be set when calling C<aio_readdirx>. Instead, it 751This flag should not be set when calling C<aio_readdirx>. Instead, it
696is being set by C<aio_readdirx>, when any of the C<$type>'s found were 752is being set by C<aio_readdirx>, when any of the C<$type>'s found were
697C<IO::AIO::DT_UNKNOWN>. The absense of this flag therefore indicates that all 753C<IO::AIO::DT_UNKNOWN>. The absence of this flag therefore indicates that all
698C<$type>'s are known, which can be used to speed up some algorithms. 754C<$type>'s are known, which can be used to speed up some algorithms.
699 755
700=back 756=back
701 757
702 758
703=item aio_load $path, $data, $callback->($status) 759=item aio_load $pathname, $data, $callback->($status)
704 760
705This is a composite request that tries to fully load the given file into 761This is a composite request that tries to fully load the given file into
706memory. Status is the same as with aio_read. 762memory. Status is the same as with aio_read.
707 763
708=cut 764=cut
830 if ($_[0] && $! == EXDEV) { 886 if ($_[0] && $! == EXDEV) {
831 aioreq_pri $pri; 887 aioreq_pri $pri;
832 add $grp aio_copy $src, $dst, sub { 888 add $grp aio_copy $src, $dst, sub {
833 $grp->result ($_[0]); 889 $grp->result ($_[0]);
834 890
835 if (!$_[0]) { 891 unless ($_[0]) {
836 aioreq_pri $pri; 892 aioreq_pri $pri;
837 add $grp aio_unlink $src; 893 add $grp aio_unlink $src;
838 } 894 }
839 }; 895 };
840 } else { 896 } else {
843 }; 899 };
844 900
845 $grp 901 $grp
846} 902}
847 903
848=item aio_scandir $path, $maxreq, $callback->($dirs, $nondirs) 904=item aio_scandir $pathname, $maxreq, $callback->($dirs, $nondirs)
849 905
850Scans a directory (similar to C<aio_readdir>) but additionally tries to 906Scans a directory (similar to C<aio_readdir>) but additionally tries to
851efficiently separate the entries of directory C<$path> into two sets of 907efficiently separate the entries of directory C<$path> into two sets of
852names, directories you can recurse into (directories), and ones you cannot 908names, directories you can recurse into (directories), and ones you cannot
853recurse into (everything else, including symlinks to directories). 909recurse into (everything else, including symlinks to directories).
884Then entries will be sorted into likely directories a non-initial dot 940Then entries will be sorted into likely directories a non-initial dot
885currently) and likely non-directories (see C<aio_readdirx>). Then every 941currently) and likely non-directories (see C<aio_readdirx>). Then every
886entry plus an appended C</.> will be C<stat>'ed, likely directories first, 942entry plus an appended C</.> will be C<stat>'ed, likely directories first,
887in order of their inode numbers. If that succeeds, it assumes that the 943in order of their inode numbers. If that succeeds, it assumes that the
888entry is a directory or a symlink to directory (which will be checked 944entry is a directory or a symlink to directory (which will be checked
889seperately). This is often faster than stat'ing the entry itself because 945separately). This is often faster than stat'ing the entry itself because
890filesystems might detect the type of the entry without reading the inode 946filesystems might detect the type of the entry without reading the inode
891data (e.g. ext2fs filetype feature), even on systems that cannot return 947data (e.g. ext2fs filetype feature), even on systems that cannot return
892the filetype information on readdir. 948the filetype information on readdir.
893 949
894If the known number of directories (link count - 2) has been reached, the 950If the known number of directories (link count - 2) has been reached, the
910 966
911 my $grp = aio_group $cb; 967 my $grp = aio_group $cb;
912 968
913 $maxreq = 4 if $maxreq <= 0; 969 $maxreq = 4 if $maxreq <= 0;
914 970
915 # stat once 971 # get a wd object
972
916 aioreq_pri $pri; 973 aioreq_pri $pri;
917 add $grp aio_stat $path, sub { 974 add $grp aio_wd $path, sub {
918 return $grp->result () if $_[0]; 975 my $wd = [shift, "."];
919 my $now = time;
920 my $hash1 = join ":", (stat _)[0,1,3,7,9];
921 976
922 # read the directory entries 977 # stat once
923 aioreq_pri $pri; 978 aioreq_pri $pri;
924 add $grp aio_readdirx $path, READDIR_DIRS_FIRST, sub { 979 add $grp aio_stat $wd, sub {
925 my $entries = shift
926 or return $grp->result (); 980 return $grp->result () if $_[0];
981 my $now = time;
982 my $hash1 = join ":", (stat _)[0,1,3,7,9];
927 983
928 # stat the dir another time 984 # read the directory entries
929 aioreq_pri $pri; 985 aioreq_pri $pri;
986 add $grp aio_readdirx $wd, READDIR_DIRS_FIRST, sub {
987 my $entries = shift
988 or return $grp->result ();
989
990 # stat the dir another time
991 aioreq_pri $pri;
930 add $grp aio_stat $path, sub { 992 add $grp aio_stat $wd, sub {
931 my $hash2 = join ":", (stat _)[0,1,3,7,9]; 993 my $hash2 = join ":", (stat _)[0,1,3,7,9];
932 994
933 my $ndirs; 995 my $ndirs;
934 996
935 # take the slow route if anything looks fishy 997 # take the slow route if anything looks fishy
936 if ($hash1 ne $hash2 or (stat _)[9] == $now) { 998 if ($hash1 ne $hash2 or (stat _)[9] == $now) {
937 $ndirs = -1; 999 $ndirs = -1;
938 } else { 1000 } else {
939 # if nlink == 2, we are finished 1001 # if nlink == 2, we are finished
940 # for non-posix-fs's, we rely on nlink < 2 1002 # for non-posix-fs's, we rely on nlink < 2
941 $ndirs = (stat _)[3] - 2 1003 $ndirs = (stat _)[3] - 2
942 or return $grp->result ([], $entries); 1004 or return $grp->result ([], $entries);
943 } 1005 }
944 1006
945 my (@dirs, @nondirs); 1007 my (@dirs, @nondirs);
946 1008
947 my $statgrp = add $grp aio_group sub { 1009 my $statgrp = add $grp aio_group sub {
948 $grp->result (\@dirs, \@nondirs); 1010 $grp->result (\@dirs, \@nondirs);
949 }; 1011 };
950 1012
951 limit $statgrp $maxreq; 1013 limit $statgrp $maxreq;
952 feed $statgrp sub { 1014 feed $statgrp sub {
953 return unless @$entries; 1015 return unless @$entries;
954 my $entry = shift @$entries; 1016 my $entry = shift @$entries;
955 1017
956 aioreq_pri $pri; 1018 aioreq_pri $pri;
1019 $wd->[1] = "$entry/.";
957 add $statgrp aio_stat "$path/$entry/.", sub { 1020 add $statgrp aio_stat $wd, sub {
958 if ($_[0] < 0) { 1021 if ($_[0] < 0) {
959 push @nondirs, $entry; 1022 push @nondirs, $entry;
960 } else { 1023 } else {
961 # need to check for real directory 1024 # need to check for real directory
962 aioreq_pri $pri; 1025 aioreq_pri $pri;
1026 $wd->[1] = $entry;
963 add $statgrp aio_lstat "$path/$entry", sub { 1027 add $statgrp aio_lstat $wd, sub {
964 if (-d _) { 1028 if (-d _) {
965 push @dirs, $entry; 1029 push @dirs, $entry;
966 1030
967 unless (--$ndirs) { 1031 unless (--$ndirs) {
968 push @nondirs, @$entries; 1032 push @nondirs, @$entries;
969 feed $statgrp; 1033 feed $statgrp;
1034 }
1035 } else {
1036 push @nondirs, $entry;
970 } 1037 }
971 } else {
972 push @nondirs, $entry;
973 } 1038 }
974 } 1039 }
975 } 1040 };
976 }; 1041 };
977 }; 1042 };
978 }; 1043 };
979 }; 1044 };
980 }; 1045 };
981 1046
982 $grp 1047 $grp
983} 1048}
984 1049
985=item aio_rmtree $path, $callback->($status) 1050=item aio_rmtree $pathname, $callback->($status)
986 1051
987Delete a directory tree starting (and including) C<$path>, return the 1052Delete a directory tree starting (and including) C<$path>, return the
988status of the final C<rmdir> only. This is a composite request that 1053status of the final C<rmdir> only. This is a composite request that
989uses C<aio_scandir> to recurse into and rmdir directories, and unlink 1054uses C<aio_scandir> to recurse into and rmdir directories, and unlink
990everything else. 1055everything else.
1032callback with the fdatasync result code. 1097callback with the fdatasync result code.
1033 1098
1034If this call isn't available because your OS lacks it or it couldn't be 1099If this call isn't available because your OS lacks it or it couldn't be
1035detected, it will be emulated by calling C<fsync> instead. 1100detected, it will be emulated by calling C<fsync> instead.
1036 1101
1102=item aio_syncfs $fh, $callback->($status)
1103
1104Asynchronously call the syncfs syscall to sync the filesystem associated
1105to the given filehandle and call the callback with the syncfs result
1106code. If syncfs is not available, calls sync(), but returns C<-1> and sets
1107errno to C<ENOSYS> nevertheless.
1108
1037=item aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status) 1109=item aio_sync_file_range $fh, $offset, $nbytes, $flags, $callback->($status)
1038 1110
1039Sync the data portion of the file specified by C<$offset> and C<$length> 1111Sync the data portion of the file specified by C<$offset> and C<$length>
1040to disk (but NOT the metadata), by calling the Linux-specific 1112to disk (but NOT the metadata), by calling the Linux-specific
1041sync_file_range call. If sync_file_range is not available or it returns 1113sync_file_range call. If sync_file_range is not available or it returns
1044C<$flags> can be a combination of C<IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE>, 1116C<$flags> can be a combination of C<IO::AIO::SYNC_FILE_RANGE_WAIT_BEFORE>,
1045C<IO::AIO::SYNC_FILE_RANGE_WRITE> and 1117C<IO::AIO::SYNC_FILE_RANGE_WRITE> and
1046C<IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER>: refer to the sync_file_range 1118C<IO::AIO::SYNC_FILE_RANGE_WAIT_AFTER>: refer to the sync_file_range
1047manpage for details. 1119manpage for details.
1048 1120
1049=item aio_pathsync $path, $callback->($status) 1121=item aio_pathsync $pathname, $callback->($status)
1050 1122
1051This request tries to open, fsync and close the given path. This is a 1123This request tries to open, fsync and close the given path. This is a
1052composite request intended to sync directories after directory operations 1124composite request intended to sync directories after directory operations
1053(E.g. rename). This might not work on all operating systems or have any 1125(E.g. rename). This might not work on all operating systems or have any
1054specific effect, but usually it makes sure that directory changes get 1126specific effect, but usually it makes sure that directory changes get
1197immense (it blocks a thread for a long time) so do not use this function 1269immense (it blocks a thread for a long time) so do not use this function
1198except to put your application under artificial I/O pressure. 1270except to put your application under artificial I/O pressure.
1199 1271
1200=back 1272=back
1201 1273
1274
1275=head2 IO::AIO::WD - multiple working directories
1276
1277Your process only has one current working directory, which is used by all
1278threads. This makes it hard to use relative paths (some other component
1279could call C<chdir> at any time, and it is hard to control when the path
1280will be used by IO::AIO).
1281
1282One solution for this is to always use absolute paths. This usually works,
1283but can be quite slow (the kernel has to walk the whole path on every
1284access), and can also be a hassle to implement.
1285
1286Newer POSIX systems have a number of functions (openat, fdopendir,
1287futimensat and so on) that make it possible to specify working directories
1288per operation.
1289
1290For portability, and because the clowns who "designed", or shall I write,
1291perpetrated this new interface were obviously half-drunk, this abstraction
1292cannot be perfect, though.
1293
1294IO::AIO allows you to convert directory paths into a so-called IO::AIO::WD
1295object. This object stores the canonicalised, absolute version of the
1296path, and on systems that allow it, also a directory file descriptor.
1297
1298Everywhere where a pathname is accepted by IO::AIO (e.g. in C<aio_stat>
1299or C<aio_unlink>), one can specify an array reference with an IO::AIO::WD
1300object and a pathname instead. If the pathname is absolute, the
1301IO::AIO::WD objetc is ignored, otherwise the pathname is resolved relative
1302to that IO::AIO::WD object.
1303
1304For example, to get a wd object for F</etc> and then stat F<passwd>
1305inside, you would write:
1306
1307 aio_wd "/etc", sub {
1308 my $etcdir = shift;
1309
1310 # although $etcdir can be undef on error, there is generally no reason
1311 # to check for errors here, as aio_stat will fail with ENOENT
1312 # when $etcdir is undef.
1313
1314 aio_stat [$etcdir, "passwd"], sub {
1315 # yay
1316 };
1317 };
1318
1319This shows that creating an IO::AIO::WD object is itself a potentially
1320blocking operation, which is why it is done asynchronously.
1321
1322As with normal pathnames, IO::AIO keeps a copy of the working directory
1323object and the pathname string, so you could write the following without
1324causing any issues due to C<$path> getting reused:
1325
1326 my $path = [$wd, undef];
1327
1328 for my $name (qw(abc def ghi)) {
1329 $path->[1] = $name;
1330 aio_stat $path, sub {
1331 # ...
1332 };
1333 }
1334
1335There are some caveats: when directories get renamed (or deleted), the
1336pathname string doesn't change, so will point to the new directory (or
1337nowhere at all), while the directory fd, if available on the system,
1338will still point to the original directory. Most functions accepting a
1339pathname will use the directory fd on newer systems, and the string on
1340older systems. Some functions (such as realpath) will always rely on the
1341string form of the pathname.
1342
1343So this fucntionality is mainly useful to get some protection against
1344C<chdir>, to easily get an absolute path out of a relative path for future
1345reference, and to speed up doing many operations in the same directory
1346(e.g. when stat'ing all files in a directory).
1347
1348The following functions implement this working directory abstraction:
1349
1350=over 4
1351
1352=item aio_wd $pathname, $callback->($wd)
1353
1354Asynchonously canonicalise the given pathname and convert it to an
1355IO::AIO::WD object representing it. If possible and supported on the
1356system, also open a directory fd to speed up pathname resolution relative
1357to this working directory.
1358
1359If something goes wrong, then C<undef> is passwd to the callback instead
1360of a working directory object and C<$!> is set appropriately. Since
1361passing C<undef> as working directory component of a pathname fails the
1362request with C<ENOENT>, there is often no need for error checking in the
1363C<aio_wd> callback, as future requests using the value will fail in the
1364expected way.
1365
1366If this call isn't available because your OS lacks it or it couldn't be
1367detected, it will be emulated by calling C<fsync> instead.
1368
1369=item IO::AIO::CWD
1370
1371This is a compiletime constant (object) that represents the process
1372current working directory.
1373
1374Specifying this object as working directory object for a pathname is as
1375if the pathname would be specified directly, without a directory object,
1376e.g., these calls are functionally identical:
1377
1378 aio_stat "somefile", sub { ... };
1379 aio_stat [IO::AIO::CWD, "somefile"], sub { ... };
1380
1381=back
1382
1383
1202=head2 IO::AIO::REQ CLASS 1384=head2 IO::AIO::REQ CLASS
1203 1385
1204All non-aggregate C<aio_*> functions return an object of this class when 1386All non-aggregate C<aio_*> functions return an object of this class when
1205called in non-void context. 1387called in non-void context.
1206 1388
1383 1565
1384See C<poll_cb> for an example. 1566See C<poll_cb> for an example.
1385 1567
1386=item IO::AIO::poll_cb 1568=item IO::AIO::poll_cb
1387 1569
1388Process some outstanding events on the result pipe. You have to call this 1570Process some outstanding events on the result pipe. You have to call
1389regularly. Returns C<0> if all events could be processed, or C<-1> if it 1571this regularly. Returns C<0> if all events could be processed (or there
1390returned earlier for whatever reason. Returns immediately when no events 1572were no events to process), or C<-1> if it returned earlier for whatever
1391are outstanding. The amount of events processed depends on the settings of 1573reason. Returns immediately when no events are outstanding. The amount of
1392C<IO::AIO::max_poll_req> and C<IO::AIO::max_poll_time>. 1574events processed depends on the settings of C<IO::AIO::max_poll_req> and
1575C<IO::AIO::max_poll_time>.
1393 1576
1394If not all requests were processed for whatever reason, the filehandle 1577If not all requests were processed for whatever reason, the filehandle
1395will still be ready when C<poll_cb> returns, so normally you don't have to 1578will still be ready when C<poll_cb> returns, so normally you don't have to
1396do anything special to have it called later. 1579do anything special to have it called later.
1580
1581Apart from calling C<IO::AIO::poll_cb> when the event filehandle becomes
1582ready, it can be beneficial to call this function from loops which submit
1583a lot of requests, to make sure the results get processed when they become
1584available and not just when the loop is finished and the event loop takes
1585over again. This function returns very fast when there are no outstanding
1586requests.
1397 1587
1398Example: Install an Event watcher that automatically calls 1588Example: Install an Event watcher that automatically calls
1399IO::AIO::poll_cb with high priority (more examples can be found in the 1589IO::AIO::poll_cb with high priority (more examples can be found in the
1400SYNOPSIS section, at the top of this document): 1590SYNOPSIS section, at the top of this document):
1401 1591
1503 1693
1504Under normal circumstances you don't need to call this function. 1694Under normal circumstances you don't need to call this function.
1505 1695
1506=item IO::AIO::max_idle $nthreads 1696=item IO::AIO::max_idle $nthreads
1507 1697
1508Limit the number of threads (default: 4) that are allowed to idle (i.e., 1698Limit the number of threads (default: 4) that are allowed to idle
1509threads that did not get a request to process within 10 seconds). That 1699(i.e., threads that did not get a request to process within the idle
1510means if a thread becomes idle while C<$nthreads> other threads are also 1700timeout (default: 10 seconds). That means if a thread becomes idle while
1511idle, it will free its resources and exit. 1701C<$nthreads> other threads are also idle, it will free its resources and
1702exit.
1512 1703
1513This is useful when you allow a large number of threads (e.g. 100 or 1000) 1704This is useful when you allow a large number of threads (e.g. 100 or 1000)
1514to allow for extremely high load situations, but want to free resources 1705to allow for extremely high load situations, but want to free resources
1515under normal circumstances (1000 threads can easily consume 30MB of RAM). 1706under normal circumstances (1000 threads can easily consume 30MB of RAM).
1516 1707
1517The default is probably ok in most situations, especially if thread 1708The default is probably ok in most situations, especially if thread
1518creation is fast. If thread creation is very slow on your system you might 1709creation is fast. If thread creation is very slow on your system you might
1519want to use larger values. 1710want to use larger values.
1520 1711
1712=item IO::AIO::idle_timeout $seconds
1713
1714Sets the minimum idle timeout (default 10) after which worker threads are
1715allowed to exit. SEe C<IO::AIO::max_idle>.
1716
1521=item IO::AIO::max_outstanding $maxreqs 1717=item IO::AIO::max_outstanding $maxreqs
1718
1719Sets the maximum number of outstanding requests to C<$nreqs>. If
1720you do queue up more than this number of requests, the next call to
1721C<IO::AIO::poll_cb> (and other functions calling C<poll_cb>, such as
1722C<IO::AIO::flush> or C<IO::AIO::poll>) will block until the limit is no
1723longer exceeded.
1724
1725In other words, this setting does not enforce a queue limit, but can be
1726used to make poll functions block if the limit is exceeded.
1522 1727
1523This is a very bad function to use in interactive programs because it 1728This is a very bad function to use in interactive programs because it
1524blocks, and a bad way to reduce concurrency because it is inexact: Better 1729blocks, and a bad way to reduce concurrency because it is inexact: Better
1525use an C<aio_group> together with a feed callback. 1730use an C<aio_group> together with a feed callback.
1526 1731
1527Sets the maximum number of outstanding requests to C<$nreqs>. If you 1732It's main use is in scripts without an event loop - when you want to stat
1528do queue up more than this number of requests, the next call to the 1733a lot of files, you can write somehting like this:
1529C<poll_cb> (and C<poll_some> and other functions calling C<poll_cb>)
1530function will block until the limit is no longer exceeded.
1531 1734
1532The default value is very large, so there is no practical limit on the 1735 IO::AIO::max_outstanding 32;
1533number of outstanding requests.
1534 1736
1535You can still queue as many requests as you want. Therefore, 1737 for my $path (...) {
1536C<max_outstanding> is mainly useful in simple scripts (with low values) or 1738 aio_stat $path , ...;
1537as a stop gap to shield against fatal memory overflow (with large values). 1739 IO::AIO::poll_cb;
1740 }
1741
1742 IO::AIO::flush;
1743
1744The call to C<poll_cb> inside the loop will normally return instantly, but
1745as soon as more thna C<32> reqeusts are in-flight, it will block until
1746some requests have been handled. This keeps the loop from pushing a large
1747number of C<aio_stat> requests onto the queue.
1748
1749The default value for C<max_outstanding> is very large, so there is no
1750practical limit on the number of outstanding requests.
1538 1751
1539=back 1752=back
1540 1753
1541=head3 STATISTICAL INFORMATION 1754=head3 STATISTICAL INFORMATION
1542 1755
1580 1793
1581Returns the number of bytes copied, or C<-1> on error. 1794Returns the number of bytes copied, or C<-1> on error.
1582 1795
1583=item IO::AIO::fadvise $fh, $offset, $len, $advice 1796=item IO::AIO::fadvise $fh, $offset, $len, $advice
1584 1797
1585Simply calls the C<posix_fadvise> function (see it's 1798Simply calls the C<posix_fadvise> function (see its
1586manpage for details). The following advice constants are 1799manpage for details). The following advice constants are
1587avaiable: C<IO::AIO::FADV_NORMAL>, C<IO::AIO::FADV_SEQUENTIAL>, 1800available: C<IO::AIO::FADV_NORMAL>, C<IO::AIO::FADV_SEQUENTIAL>,
1588C<IO::AIO::FADV_RANDOM>, C<IO::AIO::FADV_NOREUSE>, 1801C<IO::AIO::FADV_RANDOM>, C<IO::AIO::FADV_NOREUSE>,
1589C<IO::AIO::FADV_WILLNEED>, C<IO::AIO::FADV_DONTNEED>. 1802C<IO::AIO::FADV_WILLNEED>, C<IO::AIO::FADV_DONTNEED>.
1590 1803
1591On systems that do not implement C<posix_fadvise>, this function returns 1804On systems that do not implement C<posix_fadvise>, this function returns
1592ENOSYS, otherwise the return value of C<posix_fadvise>. 1805ENOSYS, otherwise the return value of C<posix_fadvise>.
1806
1807=item IO::AIO::madvise $scalar, $offset, $len, $advice
1808
1809Simply calls the C<posix_madvise> function (see its
1810manpage for details). The following advice constants are
1811available: C<IO::AIO::MADV_NORMAL>, C<IO::AIO::MADV_SEQUENTIAL>,
1812C<IO::AIO::MADV_RANDOM>, C<IO::AIO::MADV_WILLNEED>, C<IO::AIO::MADV_DONTNEED>.
1813
1814On systems that do not implement C<posix_madvise>, this function returns
1815ENOSYS, otherwise the return value of C<posix_madvise>.
1816
1817=item IO::AIO::mprotect $scalar, $offset, $len, $protect
1818
1819Simply calls the C<mprotect> function on the preferably AIO::mmap'ed
1820$scalar (see its manpage for details). The following protect
1821constants are available: C<IO::AIO::PROT_NONE>, C<IO::AIO::PROT_READ>,
1822C<IO::AIO::PROT_WRITE>, C<IO::AIO::PROT_EXEC>.
1823
1824On systems that do not implement C<mprotect>, this function returns
1825ENOSYS, otherwise the return value of C<mprotect>.
1593 1826
1594=item IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset] 1827=item IO::AIO::mmap $scalar, $length, $prot, $flags, $fh[, $offset]
1595 1828
1596Memory-maps a file (or anonymous memory range) and attaches it to the 1829Memory-maps a file (or anonymous memory range) and attaches it to the
1597given C<$scalar>, which will act like a string scalar. 1830given C<$scalar>, which will act like a string scalar.
1698 Danga::Socket->AddOtherFds (IO::AIO::poll_fileno => 1931 Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
1699 \&IO::AIO::poll_cb); 1932 \&IO::AIO::poll_cb);
1700 1933
1701=head2 FORK BEHAVIOUR 1934=head2 FORK BEHAVIOUR
1702 1935
1703This module should do "the right thing" when the process using it forks: 1936Usage of pthreads in a program changes the semantics of fork
1937considerably. Specifically, only async-safe functions can be called after
1938fork. Perl doesn't know about this, so in general, you cannot call fork
1939with defined behaviour in perl if pthreads are involved. IO::AIO uses
1940pthreads, so this applies, but many other extensions and (for inexplicable
1941reasons) perl itself often is linked against pthreads, so this limitation
1942applies to quite a lot of perls.
1704 1943
1705Before the fork, IO::AIO enters a quiescent state where no requests 1944This module no longer tries to fight your OS, or POSIX. That means IO::AIO
1706can be added in other threads and no results will be processed. After 1945only works in the process that loaded it. Forking is fully supported, but
1707the fork the parent simply leaves the quiescent state and continues 1946using IO::AIO in the child is not.
1708request/result processing, while the child frees the request/result queue
1709(so that the requests started before the fork will only be handled in the
1710parent). Threads will be started on demand until the limit set in the
1711parent process has been reached again.
1712 1947
1713In short: the parent will, after a short pause, continue as if fork had 1948You might get around by not I<using> IO::AIO before (or after)
1714not been called, while the child will act as if IO::AIO has not been used 1949forking. You could also try to call the L<IO::AIO::reinit> function in the
1715yet. 1950child:
1951
1952=over 4
1953
1954=item IO::AIO::reinit
1955
1956Abandons all current requests and I/O threads and simply reinitialises all
1957data structures. This is not an operation supported by any standards, but
1958happens to work on GNU/Linux and some newer BSD systems.
1959
1960The only reasonable use for this function is to call it after forking, if
1961C<IO::AIO> was used in the parent. Calling it while IO::AIO is active in
1962the process will result in undefined behaviour. Calling it at any time
1963will also result in any undefined (by POSIX) behaviour.
1964
1965=back
1716 1966
1717=head2 MEMORY USAGE 1967=head2 MEMORY USAGE
1718 1968
1719Per-request usage: 1969Per-request usage:
1720 1970

Diff Legend

Removed lines
+ Added lines
< Changed lines
> Changed lines