2
0
mirror of https://github.com/boostorg/thread.git synced 2026-01-23 18:12:12 +00:00
Files
thread/doc/condition.html
2001-09-18 21:24:51 +00:00

334 lines
10 KiB
HTML

<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta name="keywords" content="threads, BTL, thread library, C++">
<link rel="stylesheet" type="text/css" href="styles.css">
<title>Boost.Threads, condition</title>
</head>
<body bgcolor="#ffffff" link="#0000ff" vlink="#800080" text="#000000">
<table border="0" cellpadding="7" cellspacing="0" width="100%">
<tr>
<td valign="top" width="300">
<h3><IMG height=86 alt="C++ Boost" src="../../../c++boost.gif" width=277></h3>
</td>
<td valign="top">
<h1 align="center">Boost.Threads</h1>
<h2 align="center">condition</h2>
</td>
</tr>
</table>
<hr>
<p><A href="#Introduction">Introduction</a><br>
<A href="#Header">Header</a><br>
<A href="#Synopsis">Synopsis</a><br>
<A href="#Members">Members</a><br>
<A href="#Example">Example</a></p>
<h2><a name="Introduction">Introduction</a></h2>
<p>An object of class <code>condition</code> is a synchronization primitive used to
cause a thread to wait until a particular shared-data condition (or time) is met.
A <code>condition</code> object is always used in conjunction with a mutex
object modeling a <a href="mutex_concept.html">Mutex Concept</a>. The mutex must be locked prior to waiting on the
<code>condition</code>, which is ensured by passing a lock object modeling a <a href="lock_concept.html">Lock
Concept</a> to the <code>condition</code> object's <code>wait</code> functions. While the thread is waiting on the <code>condition</code>
object,
the mutex associated with the lock is unlocked. When the thread returns
from a call to one of the <code>condition</code> object's <code>wait</code> functions,
the mutex is again locked. The tricky lock/unlock/lock sequence is performed
automatically by the <code>condition</code> object's <code>wait</code>
functions.</p>
<p>The <code>condition</code> type is often used to implement the <i>Monitor Object</i>
and other important patterns. See <A href="bibliography.html#Schmidt-00">[Schmidt-00]</a>
and <A href="bibliography.html#Hoare-74">[Hoare 74]</a>. Monitors are one of the most
important patterns for creating reliable multithreaded programs.</p>
<p>See <A href="definitions.html">Formal Definitions</a> for definitions of thread
states <A href="definitions.html#state">blocked</a> and
<A href="definitions.html#state">ready</a>. Note that "waiting" is a synonym
for blocked.</p>
<h2><a name="Header">Header</a></h2>
<pre>
#include <A href="../../../boost/thread/condition.hpp">&lt;boost/thread/condition.hpp&gt;</a>
</pre>
<h2><a name="Synopsis">Synopsis</a></h2>
<pre>
namespace boost {
class condition : private <A href="../../utility/utility.htm#Class noncopyable">boost::noncopyable</a> // Exposition only.
// Class condition meets the <a href="overview.html#NonCopyable">NonCopyable</a> requirement.
{
public:
condition();
~condition();
void notify_one();
void notify_all();
template &lt;typename <a href="scoped_lock.html">ScopedLock</a>&gt;
void wait(<a href="scoped_lock.html">ScopedLock</a>&amp; lock);
template &lt;typename <a href="scoped_lock.html">ScopedLock</a>, typename <A href="http://www.sgi.com/tech/stl/Predicate.html">Predicate</A>&gt;
void wait(<a href="scoped_lock.html">ScopedLock</a>&amp; lock, <A href="http://www.sgi.com/tech/stl/Predicate.html">Predicate</A> pred);
template &lt;typename <a href="scoped_lock.html">ScopedLock</a>&gt;
bool timed_wait(<a href="scoped_lock.html">ScopedLock</a>&amp; lock, const xtime&amp; xt);
template &lt;typename <a href="scoped_lock.html">ScopedLock</a>, typename <A href="http://www.sgi.com/tech/stl/Predicate.html">Predicate</A>&gt;
bool timed_wait(<a href="scoped_lock.html">ScopedLock</a>&amp; lock, const xtime&amp; xt, <A href="http://www.sgi.com/tech/stl/Predicate.html">Predicate</A> pred);
};
} // namespace boost
</pre>
<h2><a name="Members">Members</a></h2>
<hr>
<h3>Constructor</h3>
<pre>
condition();
</pre>
<p><b>Effects:</b> Constructs a <code>condition</code>.</p>
<hr>
<h3>Destructor</h3>
<pre>
~condition();
</pre>
<p><b>Effects:</b> Destroys <code>*this</code>.</p>
<hr>
<h3>notify_one</h3>
<pre>
void notify_one();
</pre>
<p><b>Effects:</b> If there is a thread waiting on <code>*this</code>, change
that thread's state to ready. Otherwise there is no effect.</p>
<p><b>Notes:</b> If more that one thread is waiting on the condition, it is
unspecified which is made ready.</p>
<hr>
<h3>notify_all</h3>
<pre>
void notify_all();
</pre>
<p><b>Effects:</b> Change the state of all threads waiting on <code>*this</code>
to ready. If there are no waiting threads, <code>notify_all()</code> has no effect.</p>
<hr>
<h3>wait</h3>
<pre>
template &lt;typename ScopedLock&gt;
void wait(ScopedLock&amp; lock);
</pre>
<p><b>Requires:</b> ScopedLock meets the
<A href="lock_concept.html#ScopedLock">ScopedLock</a> requirements.</p>
<p><b>Effects:</b> Releases the lock on the <A href="mutex_concept.html">mutex model</a>
associated with <code>lock</code>, blocks the current thread of execution until readied
by a call to <code>this-&gt;notify_one()</code> or <code>this-&gt;notify_all()</code>,
and then reacquires the lock. All effects occur in an atomic fashion.</p>
<p><b>Throws:</b> <code><A href="lock_error.html">lock_error</a></code>
if <code>!lock.locked()</code></p>
<p><b>Danger:</b> This version should always be used within a loop checking that the
state logically associated with the <code>condition</code> has become true. Without
the loop, race conditions can ensue due to possible "spurious wake ups". The second
version encapsulates this loop idiom internally and is generally the preferred method.</p>
<pre>
template &lt;typename ScopedLock, typename Pr&gt;
void wait(ScopedLock&amp; lock, Pr pred);
</pre>
<p><b>Requires:</b> ScopedLock meets the
<A href="lock_concept.html#ScopedLock">ScopedLock</a> requirements, return from
<code>pred()</code> convertible to bool.</p>
<p><b>Effects:</b> As if:</p>
<code>
&nbsp;&nbsp;&nbsp;while (!pred()) wait(lock)
</code>
<p><b>Throws:</b> <code><A href="lock_error.html">lock_error</a></code> if
<code>!lock.locked()</code></p>
<hr>
<h3>timed_wait</h3>
<pre>
template &lt;typename ScopedTimedLock&gt;
bool timed_wait(ScopedTimedLock&amp; lock, const <a href="xtime.html">xtime</a>&amp; xt);
</pre>
<p><b>Requires:</b> ScopedTimeLock meets the
<A href="lock_concept.html#ScopedTimedLock">ScopedTimedLock</a> requirements.</p>
<p><b>Effects:</b> Releases the lock on the <A href="mutex_concept.html">mutex model</a>
associated with the <code>lock</code>, blocks the current thread of execution until
readied by a call to <code>this-&gt;notify_one()</code> or
<code>this-&gt;notify_all()</code>, or until <code>xt</code>, and then reacquires the
lock. All effects occur in an atomic fashion.</p>
<p><b>Throws:</b> <code><A href="lock_error.html">lock_error</a></code> if
<code>!lock.locked()</code></p>
<p><b>Danger:</b> This version should always be used within a loop checking that the
state logically associated with the <code>condition</code> has become true. Without
the loop, race conditions can ensue due to "spurious wake ups". The second version
encapsulates this loop idiom internally and is generally the preferred method.</p>
<p><b>Returns:</b> <code>false</code> if <code>xt</code> is reached, otherwise
<code>true</code>.</p>
<pre>
template &lt;typename ScopedTimedLock, typename Pr&gt;
bool timed_wait(ScopedTimedLock&amp; lock, const <a href="xtime.html">xtime</a>&amp; xt, Pr pred);
</pre>
<p><b>Requires: </b>ScopedTimeLock meets the
<A href="lock_concept.html#ScopedTimedLock">ScopedTimedLock</a> requirements,
return from <code>pred()</code> convertible to bool.</p>
<p><b>Effects: </b>As if:</p>
<code>
&nbsp;&nbsp;&nbsp;while (!pred())<br>
&nbsp;&nbsp;&nbsp;{<br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;if (!timed_wait(lock, xt))<br>
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;return false;<br>
&nbsp;&nbsp;&nbsp;}
</code>
<p><b>Throws:</b> <code><A href="lock_error.html">lock_error</a></code> if
<code>!lock.locked()</code></p>
<p><b>Returns:</b> <code>false</code> if <code>xt</code> is reached, otherwise
<code>true</code>.</p>
<hr>
<h2><a name="Example">Example Usage</a></h2>
<pre>
#include &lt;iostream&gt;
#include &lt;vector&gt;
#include <A href="../../../boost/utility.hpp">&lt;boost/utility.hpp&gt;</a>
#include <A href="../../../boost/thread/condition.hpp">&lt;boost/thread/condition.hpp&gt;</a>
#include <A href="../../../boost/thread/thread.hpp">&lt;boost/thread/thread.hpp&gt;</a>
class bounded_buffer : private boost::noncopyable
{
public:
typedef boost::mutex::scoped_lock lock;
bounded_buffer(int n) : begin(0), end(0), buffered(0), circular_buf(n) { }
void send (int m) {
lock lk(monitor);
while (buffered == circular_buf.size())
buffer_not_full.wait(lk);
circular_buf[end] = m;
end = (end+1) % circular_buf.size();
++buffered;
buffer_not_empty.notify_one();
}
int receive() {
lock lk(monitor);
while (buffered == 0)
buffer_not_empty.wait(lk);
int i = circular_buf[begin];
begin = (begin+1) % circular_buf.size();
--buffered;
buffer_not_full.notify_one();
return i;
}
private:
int begin, end, buffered;
std::vector&lt;int&gt; circular_buf;
boost::condition buffer_not_full, buffer_not_empty;
boost::mutex monitor;
};
bounded_buffer buf(2);
void sender() {
int n = 0;
while (n &lt; 100) {
buf.send(n);
std::cout &lt;&lt; &quot;sent: &quot; &lt;&lt; n &lt;&lt; std::endl;
++n;
}
buf.send(-1);
}
void receiver() {
int n;
do {
n = buf.receive();
std::cout &lt;&lt; &quot;received: &quot; &lt;&lt; n &lt;&lt; std::endl;
} while (n != -1); // -1 indicates end of buffer
}
int main(int, char*[])
{
boost::thread thrd1(&amp;sender);
boost::thread thrd2(&amp;receiver);
thrd1.join();
thrd2.join();
return 0;
}
</pre>
<p>Typical output (dependent on scheduling policies) is:</p>
<pre>
sent: 0
sent: 1
received: 0
received: 1
sent: 2
sent: 3
received: 2
received: 3
sent: 4
received: 4
</pre>
<hr>
<p>Revised <!--webbot bot="Timestamp" S-Type="EDITED" S-Format="%d %B, %Y" startspan -->13 September, 2001<!--webbot bot="Timestamp" endspan i-checksum="39334" -->
</p>
<p><i>© Copyright <A href="mailto:williamkempf@hotmail.com">William E. Kempf</a>
2001 all rights reserved.</i></p>
</body>
</html>