=head1 NAME Coro::Semaphore - counting semaphores =head1 SYNOPSIS use Coro; $sig = new Coro::Semaphore [initial value]; $sig->down; # wait for signal # ... some other "thread" $sig->up; =head1 DESCRIPTION This module implements counting semaphores. You can initialize a mutex with any level of parallel users, that is, you can initialize a sempahore that can be Ced more than once until it blocks. There is no owner associated with semaphores, so one thread can C it while another can C it (or vice versa), C can be called before C and so on: the semaphore is really just an integer counter that optionally blocks when it is 0. Counting semaphores are typically used to coordinate access to resources, with the semaphore count initialized to the number of free resources. Threads then increment the count when resources are added and decrement the count when resources are removed. You don't have to load C manually, it will be loaded automatically when you C and call the C constructor. =over 4 =cut package Coro::Semaphore; use common::sense; use Coro (); our $VERSION = 6.514; =item new [initial count] Creates a new sempahore object with the given initial lock count. The default lock count is 1, which means it is unlocked by default. Zero (or negative values) are also allowed, in which case the semaphore is locked by default. =item $sem->count Returns the current semaphore count. The semaphore can be down'ed without blocking when the count is strictly higher than C<0>. =item $sem->adjust ($diff) Atomically adds the amount given to the current semaphore count. If the count becomes positive, wakes up any waiters. Does not block if the count becomes negative, however. =item $sem->down Decrement the counter, therefore "locking" the semaphore. This method waits until the semaphore is available if the counter is zero or less. =item $sem->wait Similar to C, but does not actually decrement the counter. Instead, when this function returns, a following call to C or C is guaranteed to succeed without blocking, until the next thread switch (C etc.). Note that using C is much less efficient than using C, so try to prefer C whenever possible. =item $sem->wait ($callback) If you pass a callback argument to C, it will not wait, but immediately return. The callback will be called as soon as the semaphore becomes available (which might be instantly), and gets passed the semaphore as first argument. The callback might C the semaphore exactly once, might wake up other threads, but is I allowed to block (switch to other threads). =cut #=item $status = $sem->timed_down ($timeout) # #Like C, but returns false if semaphore couldn't be acquired within #$timeout seconds, otherwise true. #sub timed_down { # require Coro::Timer; # my $timeout = Coro::Timer::timeout ($_[1]); # # while ($_[0][0] <= 0) { # push @{$_[0][1]}, $Coro::current; # &Coro::schedule; # if ($timeout) { # # ugly as hell. slow, too, btw! # for (0..$#{$_[0][1]}) { # if ($_[0][1][$_] == $Coro::current) { # splice @{$_[0][1]}, $_, 1; # return; # } # } # die; # } # } # # --$_[0][0]; # return 1; #} =item $sem->up Unlock the semaphore again. =item $sem->try Try to C the semaphore. Returns true when this was possible, otherwise return false and leave the semaphore unchanged. =item $sem->waiters In scalar context, returns the number of threads waiting for this semaphore. Might accidentally cause WW3 if called in other contexts, so don't use these. =item $guard = $sem->guard This method calls C and then creates a guard object. When the guard object is destroyed it automatically calls C. =cut sub guard { &down; bless [$_[0]], Coro::Semaphore::guard:: } #=item $guard = $sem->timed_guard ($timeout) # #Like C, but returns undef if semaphore couldn't be acquired within #$timeout seconds, otherwise the guard object. #sub timed_guard { # &timed_down # ? bless \\$_[0], Coro::Semaphore::guard:: # : (); #} sub Coro::Semaphore::guard::DESTROY { &up($_[0][0]); } =back =head1 AUTHOR/SUPPORT/CONTACT Marc A. Lehmann http://software.schmorp.de/pkg/Coro.html =cut 1