IO-AIO/AIO.pm

=head1 NAME

IO::AIO - Asynchronous Input/Output

=head1 SYNOPSIS

 use IO::AIO;

 aio_open "/etc/passwd", O_RDONLY, 0, sub {
    my ($fh) = @_;
    ...
 };

 aio_unlink "/tmp/file", sub { };

 aio_read $fh, 30000, 1024, $buffer, 0, sub {
    $_[0] > 0 or die "read error: $!";
 };

 # version 2+ has request and group objects
 use IO::AIO 2;

 aioreq_pri 4; # give next request a very high priority
 my $req = aio_unlink "/tmp/file", sub { };
 $req->cancel; # cancel request if still in queue

 my $grp = aio_group sub { print "all stats done\n" };
 add $grp aio_stat "..." for ...;

 # AnyEvent integration
 open my $fh, "<&=" . IO::AIO::poll_fileno or die "$!";
 my $w = AnyEvent->io (fh => $fh, poll => 'r', cb => sub { IO::AIO::poll_cb });

 # Event integration
 Event->io (fd => IO::AIO::poll_fileno,
            poll => 'r',
            cb => \&IO::AIO::poll_cb);

 # Glib/Gtk2 integration
 add_watch Glib::IO IO::AIO::poll_fileno,
           in => sub { IO::AIO::poll_cb; 1 };

 # Tk integration
 Tk::Event::IO->fileevent (IO::AIO::poll_fileno, "",
                           readable => \&IO::AIO::poll_cb);

 # Danga::Socket integration
 Danga::Socket->AddOtherFds (IO::AIO::poll_fileno =>
                             \&IO::AIO::poll_cb);

=head1 DESCRIPTION

This module implements asynchronous I/O using whatever means your
operating system supports.

In this version, a number of threads are started that execute your
requests and signal their completion. You don't need thread support
in perl, and the threads created by this module will not be visible
to perl. In the future, this module might make use of the native aio
functions available on many operating systems. However, they are often
not well-supported or restricted (Linux doesn't allow them on normal
files currently, for example), and they would only support aio_read and
aio_write, so the remaining functionality would have to be implemented
using threads anyway.

Although the module will work with in the presence of other (Perl-)
threads, it is currently not reentrant in any way, so use appropriate
locking yourself, always call C<poll_cb> from within the same thread, or
never call C<poll_cb> (or other C<aio_> functions) recursively.

=head1 REQUEST ANATOMY AND LIFETIME

Every C<aio_*> function creates a request. which is a C data structure not
directly visible to Perl.

If called in non-void context, every request function returns a Perl
object representing the request. In void context, nothing is returned,
which saves a bit of memory.

The perl object is a fairly standard ref-to-hash object. The hash contents
are not used by IO::AIO so you are free to store anything you like in it.

During their existance, aio requests travel through the following states,
in order:

=over 4

=item ready

Immediately after a request is created it is put into the ready state,
waiting for a thread to execute it.

=item execute

A thread has accepted the request for processing and is currently
executing it (e.g. blocking in read).

=item pending

The request has been executed and is waiting for result processing.

While request submission and execution is fully asynchronous, result
processing is not and relies on the perl interpreter calling C<poll_cb>
(or another function with the same effect).

=item result

The request results are processed synchronously by C<poll_cb>.

The C<poll_cb> function will process all outstanding aio requests by
calling their callbacks, freeing memory associated with them and managing
any groups they are contained in.

=item done

Request has reached the end of its lifetime and holds no resources anymore
(except possibly for the Perl object, but its connection to the actual
aio request is severed and calling its methods will either do nothing or
result in a runtime error).

=cut

package IO::AIO;

no warnings;
use strict 'vars';

use base 'Exporter';

BEGIN {
   our $VERSION = '2.0';

   our @AIO_REQ = qw(aio_sendfile aio_read aio_write aio_open aio_close aio_stat
                     aio_lstat aio_unlink aio_rmdir aio_readdir aio_scandir aio_symlink
                     aio_fsync aio_fdatasync aio_readahead aio_rename aio_link aio_move
                     aio_group aio_nop);
   our @EXPORT = (@AIO_REQ, qw(aioreq_pri aioreq_nice));
   our @EXPORT_OK = qw(poll_fileno poll_cb poll_wait flush
                       min_parallel max_parallel nreqs);

   @IO::AIO::GRP::ISA = 'IO::AIO::REQ';

   require XSLoader;
   XSLoader::load ("IO::AIO", $VERSION);
}

=head1 FUNCTIONS

=head2 AIO FUNCTIONS

All the C<aio_*> calls are more or less thin wrappers around the syscall
with the same name (sans C<aio_>). The arguments are similar or identical,
and they all accept an additional (and optional) C<$callback> argument
which must be a code reference. This code reference will get called with
the syscall return code (e.g. most syscalls return C<-1> on error, unlike
perl, which usually delivers "false") as it's sole argument when the given
syscall has been executed asynchronously.

All functions expecting a filehandle keep a copy of the filehandle
internally until the request has finished.

All requests return objects of type L<IO::AIO::REQ> that allow further
manipulation of those requests while they are in-flight.

The pathnames you pass to these routines I<must> be absolute and
encoded in byte form. The reason for the former is that at the time the
request is being executed, the current working directory could have
changed. Alternatively, you can make sure that you never change the
current working directory.

To encode pathnames to byte form, either make sure you either: a)
always pass in filenames you got from outside (command line, readdir
etc.), b) are ASCII or ISO 8859-1, c) use the Encode module and encode
your pathnames to the locale (or other) encoding in effect in the user
environment, d) use Glib::filename_from_unicode on unicode filenames or e)
use something else.

=over 4

=item aioreq_pri $pri

Sets the priority for the next aio request. The default priority
is C<0>, the minimum and maximum priorities are C<-4> and C<4>,
respectively. Requests with higher priority will be serviced first.

The priority will be reset to C<0> after each call to one of the C<aio_>
functions.

Example: open a file with low priority, then read something from it with
higher priority so the read request is serviced before other low priority
open requests (potentially spamming the cache):

   aioreq_pri -3;
   aio_open ..., sub {
      return unless $_[0];

      aioreq_pri -2;
      aio_read $_[0], ..., sub {
         ...
      };
   };

=item aioreq_nice $pri_adjust

Similar to C<aioreq_pri>, but subtracts the given value from the current
priority, so effects are cumulative.

=item aio_open $pathname, $flags, $mode, $callback->($fh)

Asynchronously open or create a file and call the callback with a newly
created filehandle for the file.

The pathname passed to C<aio_open> must be absolute. See API NOTES, above,
for an explanation.

The C<$flags> argument is a bitmask. See the C<Fcntl> module for a
list. They are the same as used by C<sysopen>.

Likewise, C<$mode> specifies the mode of the newly created file, if it
didn't exist and C<O_CREAT> has been given, just like perl's C<sysopen>,
except that it is mandatory (i.e. use C<0> if you don't create new files,
and C<0666> or C<0777> if you do).

Example:

   aio_open "/etc/passwd", O_RDONLY, 0, sub {
      if ($_[0]) {
         print "open successful, fh is $_[0]\n";
         ...
      } else {
         die "open failed: $!\n";
      }
   };

=item aio_close $fh, $callback->($status)

Asynchronously close a file and call the callback with the result
code. I<WARNING:> although accepted, you should not pass in a perl
filehandle here, as perl will likely close the file descriptor another
time when the filehandle is destroyed. Normally, you can safely call perls
C<close> or just let filehandles go out of scope.

This is supposed to be a bug in the API, so that might change. It's
therefore best to avoid this function.

=item aio_read  $fh,$offset,$length, $data,$dataoffset, $callback->($retval)

=item aio_write $fh,$offset,$length, $data,$dataoffset, $callback->($retval)

Reads or writes C<length> bytes from the specified C<fh> and C<offset>
into the scalar given by C<data> and offset C<dataoffset> and calls the
callback without the actual number of bytes read (or -1 on error, just
like the syscall).

The C<$data> scalar I<MUST NOT> be modified in any way while the request
is outstanding. Modifying it can result in segfaults or WW3 (if the
necessary/optional hardware is installed).

Example: Read 15 bytes at offset 7 into scalar C<$buffer>, starting at
offset C<0> within the scalar:

   aio_read $fh, 7, 15, $buffer, 0, sub {
      $_[0] > 0 or die "read error: $!";
      print "read $_[0] bytes: <$buffer>\n";
   };

=item aio_move $srcpath, $dstpath, $callback->($status)

Try to move the I<file> (directories not supported as either source or
destination) from C<$srcpath> to C<$dstpath> and call the callback with
the C<0> (error) or C<-1> ok.

This is a composite request that tries to rename(2) the file first. If
rename files with C<EXDEV>, it creates the destination file with mode 0200
and copies the contents of the source file into it using C<aio_sendfile>,
followed by restoring atime, mtime, access mode and uid/gid, in that
order, and unlinking the C<$srcpath>.

If an error occurs, the partial destination file will be unlinked, if
possible, except when setting atime, mtime, access mode and uid/gid, where
errors are being ignored.

=cut

sub aio_move($$$) {
   my ($src, $dst, $cb) = @_;

   my $grp = aio_group $cb;

   add $grp aio_rename $src, $dst, sub {
      if ($_[0] && $! == EXDEV) {
         add $grp aio_open $src, O_RDONLY, 0, sub {
            if (my $src_fh = $_[0]) {
               my @stat = stat $src_fh;

               add $grp aio_open $dst, O_WRONLY, 0200, sub {
                  if (my $dst_fh = $_[0]) {
                     add $grp aio_sendfile $dst_fh, $src_fh, 0, $stat[7], sub {
                        close $src_fh;

                        if ($_[0] == $stat[7]) {
                           utime $stat[8], $stat[9], $dst;
                           chmod $stat[2] & 07777, $dst_fh;
                           chown $stat[4], $stat[5], $dst_fh;
                           close $dst_fh;

                           add $grp aio_unlink $src, sub {
                              $grp->result ($_[0]);
                           };
                        } else {
                           my $errno = $!;
                           add $grp aio_unlink $dst, sub {
                              $! = $errno;
                              $grp->result (-1);
                           };
                        }
                     };
                  } else {
                     $grp->result (-1);
                  }
               },

            } else {
               $grp->result (-1);
            }
         };
      } else {
         $grp->result ($_[0]);
      }
   };

   $grp
}

=item aio_sendfile $out_fh, $in_fh, $in_offset, $length, $callback->($retval)

Tries to copy C<$length> bytes from C<$in_fh> to C<$out_fh>. It starts
reading at byte offset C<$in_offset>, and starts writing at the current
file offset of C<$out_fh>. Because of that, it is not safe to issue more
than one C<aio_sendfile> per C<$out_fh>, as they will interfere with each
other.

This call tries to make use of a native C<sendfile> syscall to provide
zero-copy operation. For this to work, C<$out_fh> should refer to a
socket, and C<$in_fh> should refer to mmap'able file.

If the native sendfile call fails or is not implemented, it will be
emulated, so you can call C<aio_sendfile> on any type of filehandle
regardless of the limitations of the operating system.

Please note, however, that C<aio_sendfile> can read more bytes from
C<$in_fh> than are written, and there is no way to find out how many
bytes have been read from C<aio_sendfile> alone, as C<aio_sendfile> only
provides the number of bytes written to C<$out_fh>. Only if the result
value equals C<$length> one can assume that C<$length> bytes have been
read.

=item aio_readahead $fh,$offset,$length, $callback->($retval)

C<aio_readahead> populates the page cache with data from a file so that
subsequent reads from that file will not block on disk I/O. The C<$offset>
argument specifies the starting point from which data is to be read and
C<$length> specifies the number of bytes to be read. I/O is performed in
whole pages, so that offset is effectively rounded down to a page boundary
and bytes are read up to the next page boundary greater than or equal to
(off-set+length). C<aio_readahead> does not read beyond the end of the
file. The current file offset of the file is left unchanged.

If that syscall doesn't exist (likely if your OS isn't Linux) it will be
emulated by simply reading the data, which would have a similar effect.

=item aio_stat  $fh_or_path, $callback->($status)

=item aio_lstat $fh, $callback->($status)

Works like perl's C<stat> or C<lstat> in void context. The callback will
be called after the stat and the results will be available using C<stat _>
or C<-s _> etc...

The pathname passed to C<aio_stat> must be absolute. See API NOTES, above,
for an explanation.

Currently, the stats are always 64-bit-stats, i.e. instead of returning an
error when stat'ing a large file, the results will be silently truncated
unless perl itself is compiled with large file support.

Example: Print the length of F</etc/passwd>:

   aio_stat "/etc/passwd", sub {
      $_[0] and die "stat failed: $!";
      print "size is ", -s _, "\n";
   };

=item aio_unlink $pathname, $callback->($status)

Asynchronously unlink (delete) a file and call the callback with the
result code.

=item aio_link $srcpath, $dstpath, $callback->($status)

Asynchronously create a new link to the existing object at C<$srcpath> at
the path C<$dstpath> and call the callback with the result code.

=item aio_symlink $srcpath, $dstpath, $callback->($status)

Asynchronously create a new symbolic link to the existing object at C<$srcpath> at
the path C<$dstpath> and call the callback with the result code.

=item aio_rename $srcpath, $dstpath, $callback->($status)

Asynchronously rename the object at C<$srcpath> to C<$dstpath>, just as
rename(2) and call the callback with the result code.

=item aio_rmdir $pathname, $callback->($status)

Asynchronously rmdir (delete) a directory and call the callback with the
result code.

=item aio_readdir $pathname, $callback->($entries)

Unlike the POSIX call of the same name, C<aio_readdir> reads an entire
directory (i.e. opendir + readdir + closedir). The entries will not be
sorted, and will B<NOT> include the C<.> and C<..> entries.

The callback a single argument which is either C<undef> or an array-ref
with the filenames.

=item aio_scandir $path, $maxreq, $callback->($dirs, $nondirs)

Scans a directory (similar to C<aio_readdir>) but additionally tries to
efficiently separate the entries of directory C<$path> into two sets of
names, directories you can recurse into (directories), and ones you cannot
recurse into (everything else, including symlinks to directories).

C<aio_scandir> is a composite request that creates of many sub requests_
C<$maxreq> specifies the maximum number of outstanding aio requests that
this function generates. If it is C<< <= 0 >>, then a suitable default
will be chosen (currently 6).

On error, the callback is called without arguments, otherwise it receives
two array-refs with path-relative entry names.

Example:

   aio_scandir $dir, 0, sub {
      my ($dirs, $nondirs) = @_;
      print "real directories: @$dirs\n";
      print "everything else: @$nondirs\n";
   };

Implementation notes.

The C<aio_readdir> cannot be avoided, but C<stat()>'ing every entry can.

After reading the directory, the modification time, size etc. of the
directory before and after the readdir is checked, and if they match (and
isn't the current time), the link count will be used to decide how many
entries are directories (if >= 2). Otherwise, no knowledge of the number
of subdirectories will be assumed.

Then entries will be sorted into likely directories (everything without
a non-initial dot currently) and likely non-directories (everything
else). Then every entry plus an appended C</.> will be C<stat>'ed,
likely directories first. If that succeeds, it assumes that the entry
is a directory or a symlink to directory (which will be checked
seperately). This is often faster than stat'ing the entry itself because
filesystems might detect the type of the entry without reading the inode
data (e.g. ext2fs filetype feature).

If the known number of directories (link count - 2) has been reached, the
rest of the entries is assumed to be non-directories.

This only works with certainty on POSIX (= UNIX) filesystems, which
fortunately are the vast majority of filesystems around.

It will also likely work on non-POSIX filesystems with reduced efficiency
as those tend to return 0 or 1 as link counts, which disables the
directory counting heuristic.

=cut

sub aio_scandir($$$) {
   my ($path, $maxreq, $cb) = @_;

   my $grp = aio_group $cb;

   $maxreq = 6 if $maxreq <= 0;

   # stat once
   add $grp aio_stat $path, sub {
      return $grp->result () if $_[0];
      my $now = time;
      my $hash1 = join ":", (stat _)[0,1,3,7,9];

      # read the directory entries
      add $grp aio_readdir $path, sub {
         my $entries = shift
            or return $grp->result ();

         # stat the dir another time
         add $grp aio_stat $path, sub {
            my $hash2 = join ":", (stat _)[0,1,3,7,9];

            my $ndirs;

            # take the slow route if anything looks fishy
            if ($hash1 ne $hash2 or (stat _)[9] == $now) {
               $ndirs = -1;
            } else {
               # if nlink == 2, we are finished
               # on non-posix-fs's, we rely on nlink < 2
               $ndirs = (stat _)[3] - 2
                  or return $grp->result ([], $entries);
            }

            # sort into likely dirs and likely nondirs
            # dirs == files without ".", short entries first
            $entries = [map $_->[0],
                           sort { $b->[1] cmp $a->[1] }
                              map [$_, sprintf "%s%04d", (/.\./ ? "1" : "0"), length],
                                 @$entries];

            my (@dirs, @nondirs);

            my $statgrp = add $grp aio_group sub {
               $grp->result (\@dirs, \@nondirs);
            };

            limit $statgrp $maxreq;
            feed $statgrp sub {
               return unless @$entries;
               my $entry = pop @$entries;

               add $statgrp aio_stat "$path/$entry/.", sub {
                  if ($_[0] < 0) {
                     push @nondirs, $entry;
                  } else {
                     # need to check for real directory
                     add $statgrp aio_lstat "$path/$entry", sub {
                        if (-d _) {
                           push @dirs, $entry;

                           unless (--$ndirs) {
                              push @nondirs, @$entries;
                              feed $statgrp;
                           }
                        } else {
                           push @nondirs, $entry;
                        }
                     }
                  }
               };
            };
         };
      };
   };

   $grp
}

=item aio_fsync $fh, $callback->($status)

Asynchronously call fsync on the given filehandle and call the callback
with the fsync result code.

=item aio_fdatasync $fh, $callback->($status)

Asynchronously call fdatasync on the given filehandle and call the
callback with the fdatasync result code.

If this call isn't available because your OS lacks it or it couldn't be
detected, it will be emulated by calling C<fsync> instead.

=item aio_group $callback->(...)

This is a very special aio request: Instead of doing something, it is a
container for other aio requests, which is useful if you want to bundle
many requests into a single, composite, request with a definite callback
and the ability to cancel the whole request with its subrequests.

Returns an object of class L<IO::AIO::GRP>. See its documentation below
for more info.

Example:

   my $grp = aio_group sub {
      print "all stats done\n";
   };

   add $grp
      (aio_stat ...),
      (aio_stat ...),
      ...;

=item aio_nop $callback->()

This is a special request - it does nothing in itself and is only used for
side effects, such as when you want to add a dummy request to a group so
that finishing the requests in the group depends on executing the given
code.

While this request does nothing, it still goes through the execution
phase and still requires a worker thread. Thus, the callback will not
be executed immediately but only after other requests in the queue have
entered their execution phase. This can be used to measure request
latency.

=item IO::AIO::aio_busy $fractional_seconds, $callback->()  *NOT EXPORTED*

Mainly used for debugging and benchmarking, this aio request puts one of
the request workers to sleep for the given time.

While it is theoretically handy to have simple I/O scheduling requests
like sleep and file handle readable/writable, the overhead this creates is
immense (it blocks a thread for a long time) so do not use this function
except to put your application under artificial I/O pressure.

=back

=head2 IO::AIO::REQ CLASS

All non-aggregate C<aio_*> functions return an object of this class when
called in non-void context.

=over 4

=item cancel $req

Cancels the request, if possible. Has the effect of skipping execution
when entering the B<execute> state and skipping calling the callback when
entering the the B<result> state, but will leave the request otherwise
untouched. That means that requests that currently execute will not be
stopped and resources held by the request will not be freed prematurely.

=item cb $req $callback->(...)

Replace (or simply set) the callback registered to the request.

=back

=head2 IO::AIO::GRP CLASS

This class is a subclass of L<IO::AIO::REQ>, so all its methods apply to
objects of this class, too.

A IO::AIO::GRP object is a special request that can contain multiple other
aio requests.

You create one by calling the C<aio_group> constructing function with a
callback that will be called when all contained requests have entered the
C<done> state:

   my $grp = aio_group sub {
      print "all requests are done\n";
   };

You add requests by calling the C<add> method with one or more
C<IO::AIO::REQ> objects:

   $grp->add (aio_unlink "...");

   add $grp aio_stat "...", sub {
      $_[0] or return $grp->result ("error");

      # add another request dynamically, if first succeeded
      add $grp aio_open "...", sub {
         $grp->result ("ok");
      };
   };

This makes it very easy to create composite requests (see the source of
C<aio_move> for an application) that work and feel like simple requests.

=over 4

=item * The IO::AIO::GRP objects will be cleaned up during calls to
C<IO::AIO::poll_cb>, just like any other request.

=item * They can be canceled like any other request. Canceling will cancel not
only the request itself, but also all requests it contains.

=item * They can also can also be added to other IO::AIO::GRP objects.

=item * You must not add requests to a group from within the group callback (or
any later time).

=back

Their lifetime, simplified, looks like this: when they are empty, they
will finish very quickly. If they contain only requests that are in the
C<done> state, they will also finish. Otherwise they will continue to
exist.

That means after creating a group you have some time to add requests. And
in the callbacks of those requests, you can add further requests to the
group. And only when all those requests have finished will the the group
itself finish.

=over 4

=item add $grp ...

=item $grp->add (...)

Add one or more requests to the group. Any type of L<IO::AIO::REQ> can
be added, including other groups, as long as you do not create circular
dependencies.

Returns all its arguments.

=item $grp->cancel_subs

Cancel all subrequests and clears any feeder, but not the group request
itself. Useful when you queued a lot of events but got a result early.

=item $grp->result (...)

Set the result value(s) that will be passed to the group callback when all
subrequests have finished. By default, no argument will be passed.

=item feed $grp $callback->($grp)

Sets a feeder/generator on this group: every group can have an attached
generator that generates requests if idle. The idea behind this is that,
although you could just queue as many requests as you want in a group,
this might starve other requests for a potentially long time.  For
example, C<aio_scandir> might generate hundreds of thousands C<aio_stat>
requests, delaying any later requests for a long time.

To avoid this, and allow incremental generation of requests, you can
instead a group and set a feeder on it that generates those requests. The
feed callback will be called whenever there are few enough (see C<limit>,
below) requests active in the group itself and is expected to queue more
requests.

The feed callback can queue as many requests as it likes (i.e. C<add> does
not impose any limits).

If the feed does not queue more requests when called, it will be
automatically removed from the group.

If the feed limit is C<0>, it will be set to C<2> automatically.

Example:

   # stat all files in @files, but only ever use four aio requests concurrently:

   my $grp = aio_group sub { print "finished\n" };
   limit $grp 4;
   feed $grp sub {
      my $file = pop @files
         or return;

      add $grp aio_stat $file, sub { ... };
   };

=item limit $grp $num

Sets the feeder limit for the group: The feeder will be called whenever
the group contains less than this many requests.

Setting the limit to C<0> will pause the feeding process.

=back

=head2 SUPPORT FUNCTIONS

=over 4

=item $fileno = IO::AIO::poll_fileno

Return the I<request result pipe file descriptor>. This filehandle must be
polled for reading by some mechanism outside this module (e.g. Event or
select, see below or the SYNOPSIS). If the pipe becomes readable you have
to call C<poll_cb> to check the results.

See C<poll_cb> for an example.

=item IO::AIO::poll_cb

Process all outstanding events on the result pipe. You have to call this
regularly. Returns the number of events processed. Returns immediately
when no events are outstanding.

Example: Install an Event watcher that automatically calls
IO::AIO::poll_cb with high priority:

   Event->io (fd => IO::AIO::poll_fileno,
              poll => 'r', async => 1,
              cb => \&IO::AIO::poll_cb);

=item IO::AIO::poll_wait

Wait till the result filehandle becomes ready for reading (simply does a
C<select> on the filehandle. This is useful if you want to synchronously wait
for some requests to finish).

See C<nreqs> for an example.

=item IO::AIO::nreqs

Returns the number of requests currently outstanding (i.e. for which their
callback has not been invoked yet).

Example: wait till there are no outstanding requests anymore:

   IO::AIO::poll_wait, IO::AIO::poll_cb
      while IO::AIO::nreqs;

=item IO::AIO::flush

Wait till all outstanding AIO requests have been handled.

Strictly equivalent to:

   IO::AIO::poll_wait, IO::AIO::poll_cb
      while IO::AIO::nreqs;

=item IO::AIO::poll

Waits until some requests have been handled.

Strictly equivalent to:

   IO::AIO::poll_wait, IO::AIO::poll_cb
      if IO::AIO::nreqs;

=item IO::AIO::min_parallel $nthreads

Set the minimum number of AIO threads to C<$nthreads>. The current
default is C<8>, which means eight asynchronous operations can execute
concurrently at any one time (the number of outstanding requests,
however, is unlimited).

IO::AIO starts threads only on demand, when an AIO request is queued and
no free thread exists.

It is recommended to keep the number of threads relatively low, as some
Linux kernel versions will scale negatively with the number of threads
(higher parallelity => MUCH higher latency). With current Linux 2.6
versions, 4-32 threads should be fine.

Under most circumstances you don't need to call this function, as the
module selects a default that is suitable for low to moderate load.

=item IO::AIO::max_parallel $nthreads

Sets the maximum number of AIO threads to C<$nthreads>. If more than the
specified number of threads are currently running, this function kills
them. This function blocks until the limit is reached.

While C<$nthreads> are zero, aio requests get queued but not executed
until the number of threads has been increased again.

This module automatically runs C<max_parallel 0> at program end, to ensure
that all threads are killed and that there are no outstanding requests.

Under normal circumstances you don't need to call this function.

=item $oldnreqs = IO::AIO::max_outstanding $nreqs

[REMOVED]

Pre-2.x versions used max_outstanding for a crude request queue length limit.

In 2.x+ you are advised to use a group and a feeder to limit
concurrency. The max_outstanding feature ran very unstable (endless
recursions causing segfaults, bad interaction with groups etc.) and was
removed.

I am deeply sorry, but I am still on the hunt for a good limiting interface.

Original description was as follows:

Sets the maximum number of outstanding requests to C<$nreqs>. If you try
to queue up more than this number of requests, the caller will block until
some requests have been handled.

=back

=cut

# support function to convert a fd into a perl filehandle
sub _fd2fh {
   return undef if $_[0] < 0;

   # try to generate nice filehandles
   my $sym = "IO::AIO::fd#$_[0]";
   local *$sym;

   open *$sym, "+<&=$_[0]"      # usually works under any unix
      or open *$sym, "<&=$_[0]" # cygwin needs this
      or open *$sym, ">&=$_[0]" # or this
      or return undef;

   *$sym
}

min_parallel 8;

END {
   max_parallel 0;
}

1;

=head2 FORK BEHAVIOUR

This module should do "the right thing" when the process using it forks:

Before the fork, IO::AIO enters a quiescent state where no requests
can be added in other threads and no results will be processed. After
the fork the parent simply leaves the quiescent state and continues
request/result processing, while the child frees the request/result queue
(so that the requests started before the fork will only be handled in the
parent). Threads will be started on demand until the limit set in the
parent process has been reached again.

In short: the parent will, after a short pause, continue as if fork had
not been called, while the child will act as if IO::AIO has not been used
yet.

=head2 MEMORY USAGE

Per-request usage:

Each aio request uses - depending on your architecture - around 100-200
bytes of memory. In addition, stat requests need a stat buffer (possibly
a few hundred bytes), readdir requires a result buffer and so on. Perl
scalars and other data passed into aio requests will also be locked and
will consume memory till the request has entered the done state.

This is now awfully much, so queuing lots of requests is not usually a
problem.

Per-thread usage:

In the execution phase, some aio requests require more memory for
temporary buffers, and each thread requires a stack and other data
structures (usually around 16k-128k, depending on the OS).

=head1 KNOWN BUGS

Known bugs will be fixed in the next release.

=head1 SEE ALSO

L<Coro::AIO>.

=head1 AUTHOR

 Marc Lehmann <schmorp@schmorp.de>
 http://home.schmorp.de/

=cut

Revision:	1.77
Committed:	Wed Oct 25 17:57:30 2006 UTC (17 years, 6 months ago) by root
Branch:	MAIN
Changes since 1.76:	+14 -17 lines
Log Message:	* empty log message *