| 1 |
424 |
jeremybenn |
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
|
| 2 |
|
|
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
|
| 3 |
|
|
<html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>Chapter 41. Concurrency</title><meta name="generator" content="DocBook XSL Stylesheets V1.75.2" /><meta name="keywords" content=" ISO C++ , library " /><meta name="keywords" content=" ISO C++ , library " /><link rel="home" href="../spine.html" title="The GNU C++ Library Documentation" /><link rel="up" href="extensions.html" title="Part XII. Extensions" /><link rel="prev" href="ext_demangling.html" title="Chapter 40. Demangling" /><link rel="next" href="bk01pt12ch41s02.html" title="Implementation" /></head><body><div class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="3" align="center">Chapter 41. Concurrency</th></tr><tr><td width="20%" align="left"><a accesskey="p" href="ext_demangling.html">Prev</a> </td><th width="60%" align="center">Part XII.
|
| 4 |
|
|
Extensions
|
| 5 |
|
|
|
| 6 |
|
|
</th><td width="20%" align="right"> <a accesskey="n" href="bk01pt12ch41s02.html">Next</a></td></tr></table><hr /></div><div class="chapter" title="Chapter 41. Concurrency"><div class="titlepage"><div><div><h2 class="title"><a id="manual.ext.concurrency"></a>Chapter 41. Concurrency</h2></div></div></div><div class="toc"><p><b>Table of Contents</b></p><dl><dt><span class="sect1"><a href="ext_concurrency.html#manual.ext.concurrency.design">Design</a></span></dt><dd><dl><dt><span class="sect2"><a href="ext_concurrency.html#manual.ext.concurrency.design.threads">Interface to Locks and Mutexes</a></span></dt><dt><span class="sect2"><a href="ext_concurrency.html#manual.ext.concurrency.design.atomics">Interface to Atomic Functions</a></span></dt></dl></dd><dt><span class="sect1"><a href="bk01pt12ch41s02.html">Implementation</a></span></dt><dd><dl><dt><span class="sect2"><a href="bk01pt12ch41s02.html#manual.ext.concurrency.impl.atomic_fallbacks">Using Builtin Atomic Functions</a></span></dt><dt><span class="sect2"><a href="bk01pt12ch41s02.html#manual.ext.concurrency.impl.thread">Thread Abstraction</a></span></dt></dl></dd><dt><span class="sect1"><a href="bk01pt12ch41s03.html">Use</a></span></dt></dl></div><div class="sect1" title="Design"><div class="titlepage"><div><div><h2 class="title" style="clear: both"><a id="manual.ext.concurrency.design"></a>Design</h2></div></div></div><div class="sect2" title="Interface to Locks and Mutexes"><div class="titlepage"><div><div><h3 class="title"><a id="manual.ext.concurrency.design.threads"></a>Interface to Locks and Mutexes</h3></div></div></div><p>The file <ext/concurrence.h> contains all the higher-level
|
| 7 |
|
|
constructs for playing with threads. In contrast to the atomics layer,
|
| 8 |
|
|
the concurrence layer consists largely of types. All types are defined within <code class="code">namespace __gnu_cxx</code>.
|
| 9 |
|
|
</p><p>
|
| 10 |
|
|
These types can be used in a portable manner, regardless of the
|
| 11 |
|
|
specific environment. They are carefully designed to provide optimum
|
| 12 |
|
|
efficiency and speed, abstracting out underlying thread calls and
|
| 13 |
|
|
accesses when compiling for single-threaded situations (even on hosts
|
| 14 |
|
|
that support multiple threads.)
|
| 15 |
|
|
</p><p>The enumerated type <code class="code">_Lock_policy</code> details the set of
|
| 16 |
|
|
available locking
|
| 17 |
|
|
policies: <code class="code">_S_single</code>, <code class="code">_S_mutex</code>,
|
| 18 |
|
|
and <code class="code">_S_atomic</code>.
|
| 19 |
|
|
</p><div class="itemizedlist"><ul class="itemizedlist" type="disc"><li class="listitem"><p><code class="code">_S_single</code></p><p>Indicates single-threaded code that does not need locking.
|
| 20 |
|
|
</p></li><li class="listitem"><p><code class="code">_S_mutex</code></p><p>Indicates multi-threaded code using thread-layer abstractions.
|
| 21 |
|
|
</p></li><li class="listitem"><p><code class="code">_S_atomic</code></p><p>Indicates multi-threaded code using atomic operations.
|
| 22 |
|
|
</p></li></ul></div><p>The compile-time constant <code class="code">__default_lock_policy</code> is set
|
| 23 |
|
|
to one of the three values above, depending on characteristics of the
|
| 24 |
|
|
host environment and the current compilation flags.
|
| 25 |
|
|
</p><p>Two more datatypes make up the rest of the
|
| 26 |
|
|
interface: <code class="code">__mutex</code>, and <code class="code">__scoped_lock</code>.
|
| 27 |
|
|
</p><p>
|
| 28 |
|
|
</p><p>The scoped lock idiom is well-discussed within the C++
|
| 29 |
|
|
community. This version takes a <code class="code">__mutex</code> reference, and
|
| 30 |
|
|
locks it during construction of <code class="code">__scoped_locke</code> and
|
| 31 |
|
|
unlocks it during destruction. This is an efficient way of locking
|
| 32 |
|
|
critical sections, while retaining exception-safety.
|
| 33 |
|
|
</p></div><div class="sect2" title="Interface to Atomic Functions"><div class="titlepage"><div><div><h3 class="title"><a id="manual.ext.concurrency.design.atomics"></a>Interface to Atomic Functions</h3></div></div></div><p>
|
| 34 |
|
|
Two functions and one type form the base of atomic support.
|
| 35 |
|
|
</p><p>The type <code class="code">_Atomic_word</code> is a signed integral type
|
| 36 |
|
|
supporting atomic operations.
|
| 37 |
|
|
</p><p>
|
| 38 |
|
|
The two functions functions are:
|
| 39 |
|
|
</p><pre class="programlisting">
|
| 40 |
|
|
_Atomic_word
|
| 41 |
|
|
__exchange_and_add_dispatch(volatile _Atomic_word*, int);
|
| 42 |
|
|
|
| 43 |
|
|
void
|
| 44 |
|
|
__atomic_add_dispatch(volatile _Atomic_word*, int);
|
| 45 |
|
|
</pre><p>Both of these functions are declared in the header file
|
| 46 |
|
|
<ext/atomicity.h>, and are in <code class="code">namespace __gnu_cxx</code>.
|
| 47 |
|
|
</p><div class="itemizedlist"><ul class="itemizedlist" type="disc"><li class="listitem"><p>
|
| 48 |
|
|
<code class="code">
|
| 49 |
|
|
__exchange_and_add_dispatch
|
| 50 |
|
|
</code>
|
| 51 |
|
|
</p><p>Adds the second argument's value to the first argument. Returns the old value.
|
| 52 |
|
|
</p></li><li class="listitem"><p>
|
| 53 |
|
|
<code class="code">
|
| 54 |
|
|
__atomic_add_dispatch
|
| 55 |
|
|
</code>
|
| 56 |
|
|
</p><p>Adds the second argument's value to the first argument. Has no return value.
|
| 57 |
|
|
</p></li></ul></div><p>
|
| 58 |
|
|
These functions forward to one of several specialized helper
|
| 59 |
|
|
functions, depending on the circumstances. For instance,
|
| 60 |
|
|
</p><p>
|
| 61 |
|
|
<code class="code">
|
| 62 |
|
|
__exchange_and_add_dispatch
|
| 63 |
|
|
</code>
|
| 64 |
|
|
</p><p>
|
| 65 |
|
|
Calls through to either of:
|
| 66 |
|
|
</p><div class="itemizedlist"><ul class="itemizedlist" type="disc"><li class="listitem"><p><code class="code">__exchange_and_add</code>
|
| 67 |
|
|
</p><p>Multi-thread version. Inlined if compiler-generated builtin atomics
|
| 68 |
|
|
can be used, otherwise resolved at link time to a non-builtin code
|
| 69 |
|
|
sequence.
|
| 70 |
|
|
</p></li><li class="listitem"><p><code class="code">__exchange_and_add_single</code>
|
| 71 |
|
|
</p><p>Single threaded version. Inlined.</p></li></ul></div><p>However, only <code class="code">__exchange_and_add_dispatch</code>
|
| 72 |
|
|
and <code class="code">__atomic_add_dispatch</code> should be used. These functions
|
| 73 |
|
|
can be used in a portable manner, regardless of the specific
|
| 74 |
|
|
environment. They are carefully designed to provide optimum efficiency
|
| 75 |
|
|
and speed, abstracting out atomic accesses when they are not required
|
| 76 |
|
|
(even on hosts that support compiler intrinsics for atomic
|
| 77 |
|
|
operations.)
|
| 78 |
|
|
</p><p>
|
| 79 |
|
|
In addition, there are two macros
|
| 80 |
|
|
</p><p>
|
| 81 |
|
|
<code class="code">
|
| 82 |
|
|
_GLIBCXX_READ_MEM_BARRIER
|
| 83 |
|
|
</code>
|
| 84 |
|
|
</p><p>
|
| 85 |
|
|
<code class="code">
|
| 86 |
|
|
_GLIBCXX_WRITE_MEM_BARRIER
|
| 87 |
|
|
</code>
|
| 88 |
|
|
</p><p>
|
| 89 |
|
|
Which expand to the appropriate write and read barrier required by the
|
| 90 |
|
|
host hardware and operating system.
|
| 91 |
|
|
</p></div></div></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="ext_demangling.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="extensions.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="bk01pt12ch41s02.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">Chapter 40. Demangling </td><td width="20%" align="center"><a accesskey="h" href="../spine.html">Home</a></td><td width="40%" align="right" valign="top"> Implementation</td></tr></table></div></body></html>
|
