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 1.33 by root, Wed Aug 17 06:12:10 2005 UTC vs.
Revision 1.34 by root, Thu Aug 18 16:32:10 2005 UTC

… …
63use base 'Exporter'; 63use base 'Exporter';
64 64
65use Fcntl (); 65use Fcntl ();
66 66
67BEGIN { 67BEGIN {
68 $VERSION = 1.4; 68 $VERSION = 1.5;
69 69
70 @EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink 70 @EXPORT = qw(aio_read aio_write aio_open aio_close aio_stat aio_lstat aio_unlink
71 aio_rmdir aio_symlink aio_fsync aio_fdatasync aio_readahead); 71 aio_rmdir aio_symlink aio_fsync aio_fdatasync aio_readahead);
72 @EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel max_outstanding nreqs); 72 @EXPORT_OK = qw(poll_fileno poll_cb min_parallel max_parallel max_outstanding nreqs);
73 73
… …
287 IO::AIO::poll_wait, IO::AIO::poll_cb 287 IO::AIO::poll_wait, IO::AIO::poll_cb
288 if IO::AIO::nreqs; 288 if IO::AIO::nreqs;
289 289
290=item IO::AIO::min_parallel $nthreads 290=item IO::AIO::min_parallel $nthreads
291 291
292Set the minimum number of AIO threads to C<$nthreads>. The default is 292Set the minimum number of AIO threads to C<$nthreads>. The current default
293C<1>, which means a single asynchronous operation can be done at one time 293is C<4>, which means four asynchronous operations can be done at one time
294(the number of outstanding operations, however, is unlimited). 294(the number of outstanding operations, however, is unlimited).
295
296IO::AIO starts threads only on demand, when an AIO request is queued and
297no free thread exists.
295 298
296It is recommended to keep the number of threads low, as some Linux 299It is recommended to keep the number of threads low, as some Linux
297kernel versions will scale negatively with the number of threads (higher 300kernel versions will scale negatively with the number of threads (higher
298parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32 301parallelity => MUCH higher latency). With current Linux 2.6 versions, 4-32
299threads should be fine. 302threads should be fine.
300 303
301Under normal circumstances you don't need to call this function, as this 304Under most circumstances you don't need to call this function, as the
302module automatically starts some threads (the exact number might change, 305module selects a default that is suitable for low to moderate load.
303and is currently 4).
304 306
305=item IO::AIO::max_parallel $nthreads 307=item IO::AIO::max_parallel $nthreads
306 308
307Sets the maximum number of AIO threads to C<$nthreads>. If more than 309Sets the maximum number of AIO threads to C<$nthreads>. If more than the
308the specified number of threads are currently running, kill them. This 310specified number of threads are currently running, this function kills
309function blocks until the limit is reached. 311them. This function blocks until the limit is reached.
312
313While C<$nthreads> are zero, aio requests get queued but not executed
314until the number of threads has been increased again.
310 315
311This module automatically runs C<max_parallel 0> at program end, to ensure 316This module automatically runs C<max_parallel 0> at program end, to ensure
312that all threads are killed and that there are no outstanding requests. 317that all threads are killed and that there are no outstanding requests.
313 318
314Under normal circumstances you don't need to call this function. 319Under normal circumstances you don't need to call this function.
… …
318Sets the maximum number of outstanding requests to C<$nreqs>. If you 323Sets the maximum number of outstanding requests to C<$nreqs>. If you
319try to queue up more than this number of requests, the caller will block until 324try to queue up more than this number of requests, the caller will block until
320some requests have been handled. 325some requests have been handled.
321 326
322The default is very large, so normally there is no practical limit. If you 327The default is very large, so normally there is no practical limit. If you
323queue up many requests in a loop it it often improves speed if you set 328queue up many requests in a loop it often improves speed if you set
324this to a relatively low number, such as C<100>. 329this to a relatively low number, such as C<100>.
325 330
326Under normal circumstances you don't need to call this function. 331Under normal circumstances you don't need to call this function.
327 332
328=back 333=back
… …
353 358
3541; 3591;
355 360
356=head2 FORK BEHAVIOUR 361=head2 FORK BEHAVIOUR
357 362
358Before the fork IO::AIO enters a quiescent state where no requests can be 363Before the fork, IO::AIO enters a quiescent state where no requests
359added in other threads and no results will be processed. After the fork 364can be added in other threads and no results will be processed. After
360the parent simply leaves the quiescent state and continues request/result 365the fork the parent simply leaves the quiescent state and continues
361processing, while the child clears the request/result queue and starts the 366request/result processing, while the child clears the request/result
362same number of threads as were in use by the parent. 367queue (so the requests started before the fork will only be handled in
368the parent). Threats will be started on demand until the limit ste in the
369parent process has been reached again.
363 370
364=head1 SEE ALSO 371=head1 SEE ALSO
365 372
366L<Coro>, L<Linux::AIO>. 373L<Coro>, L<Linux::AIO>.
367 374

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines