[www-releases] r297634 - Check in the 4.0.0 release
Hans Wennborg via llvm-commits
llvm-commits at lists.llvm.org
Mon Mar 13 09:30:21 PDT 2017
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llc.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llc.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llc.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llc.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,315 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llc - LLVM static compiler — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="lli - directly execute programs from LLVM bitcode" href="lli.html" />
+ <link rel="prev" title="opt - LLVM optimizer" href="opt.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="lli.html" title="lli - directly execute programs from LLVM bitcode"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="opt.html" title="opt - LLVM optimizer"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llc-llvm-static-compiler">
+<h1>llc - LLVM static compiler<a class="headerlink" href="#llc-llvm-static-compiler" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llc</strong> [<em>options</em>] [<em>filename</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p>The <strong class="program">llc</strong> command compiles LLVM source inputs into assembly language
+for a specified architecture. The assembly language output can then be passed
+through a native assembler and linker to generate a native executable.</p>
+<p>The choice of architecture for the output assembly code is automatically
+determined from the input file, unless the <a class="reference internal" href="lli.html#cmdoption-march"><em class="xref std std-option">-march</em></a> option is used to
+override the default.</p>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<p>If <tt class="docutils literal"><span class="pre">filename</span></tt> is “<tt class="docutils literal"><span class="pre">-</span></tt>” or omitted, <strong class="program">llc</strong> reads from standard input.
+Otherwise, it will from <tt class="docutils literal"><span class="pre">filename</span></tt>. Inputs can be in either the LLVM assembly
+language format (<tt class="docutils literal"><span class="pre">.ll</span></tt>) or the LLVM bitcode format (<tt class="docutils literal"><span class="pre">.bc</span></tt>).</p>
+<p>If the <a class="reference internal" href="opt.html#cmdoption-o"><em class="xref std std-option">-o</em></a> option is omitted, then <strong class="program">llc</strong> will send its output
+to standard output if the input is from standard input. If the <a class="reference internal" href="opt.html#cmdoption-o"><em class="xref std std-option">-o</em></a>
+option specifies “<tt class="docutils literal"><span class="pre">-</span></tt>”, then the output will also be sent to standard output.</p>
+<p>If no <a class="reference internal" href="opt.html#cmdoption-o"><em class="xref std std-option">-o</em></a> option is specified and an input file other than “<tt class="docutils literal"><span class="pre">-</span></tt>” is
+specified, then <strong class="program">llc</strong> creates the output filename by taking the input
+filename, removing any existing <tt class="docutils literal"><span class="pre">.bc</span></tt> extension, and adding a <tt class="docutils literal"><span class="pre">.s</span></tt> suffix.</p>
+<p>Other <strong class="program">llc</strong> options are described below.</p>
+<div class="section" id="end-user-options">
+<h3>End-user Options<a class="headerlink" href="#end-user-options" title="Permalink to this headline">¶</a></h3>
+<dl class="option">
+<dt id="cmdoption-help">
+<tt class="descname">-help</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-help" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print a summary of command line options.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-O">
+<tt class="descname">-O</tt><tt class="descclassname">=uint</tt><a class="headerlink" href="#cmdoption-O" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate code at different optimization levels. These correspond to the
+<tt class="docutils literal"><span class="pre">-O0</span></tt>, <tt class="docutils literal"><span class="pre">-O1</span></tt>, <tt class="docutils literal"><span class="pre">-O2</span></tt>, and <tt class="docutils literal"><span class="pre">-O3</span></tt> optimization levels used by
+<strong class="program">clang</strong>.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-mtriple">
+<tt class="descname">-mtriple</tt><tt class="descclassname">=<target triple></tt><a class="headerlink" href="#cmdoption-mtriple" title="Permalink to this definition">¶</a></dt>
+<dd><p>Override the target triple specified in the input file with the specified
+string.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-march">
+<tt class="descname">-march</tt><tt class="descclassname">=<arch></tt><a class="headerlink" href="#cmdoption-march" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the architecture for which to generate assembly, overriding the target
+encoded in the input file. See the output of <tt class="docutils literal"><span class="pre">llc</span> <span class="pre">-help</span></tt> for a list of
+valid architectures. By default this is inferred from the target triple or
+autodetected to the current architecture.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-mcpu">
+<tt class="descname">-mcpu</tt><tt class="descclassname">=<cpuname></tt><a class="headerlink" href="#cmdoption-mcpu" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify a specific chip in the current architecture to generate code for.
+By default this is inferred from the target triple and autodetected to
+the current architecture. For a list of available CPUs, use:</p>
+<div class="highlight-none"><div class="highlight"><pre>llvm-as < /dev/null | llc -march=xyz -mcpu=help
+</pre></div>
+</div>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-filetype">
+<tt class="descname">-filetype</tt><tt class="descclassname">=<output file type></tt><a class="headerlink" href="#cmdoption-filetype" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify what kind of output <tt class="docutils literal"><span class="pre">llc</span></tt> should generated. Options are: <tt class="docutils literal"><span class="pre">asm</span></tt>
+for textual assembly ( <tt class="docutils literal"><span class="pre">'.s'</span></tt>), <tt class="docutils literal"><span class="pre">obj</span></tt> for native object files (<tt class="docutils literal"><span class="pre">'.o'</span></tt>)
+and <tt class="docutils literal"><span class="pre">null</span></tt> for not emitting anything (for performance testing).</p>
+<p>Note that not all targets support all options.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-mattr">
+<tt class="descname">-mattr</tt><tt class="descclassname">=a1,+a2,-a3,...</tt><a class="headerlink" href="#cmdoption-mattr" title="Permalink to this definition">¶</a></dt>
+<dd><p>Override or control specific attributes of the target, such as whether SIMD
+operations are enabled or not. The default set of attributes is set by the
+current CPU. For a list of available attributes, use:</p>
+<div class="highlight-none"><div class="highlight"><pre>llvm-as < /dev/null | llc -march=xyz -mattr=help
+</pre></div>
+</div>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption--disable-fp-elim">
+<tt class="descname">--disable-fp-elim</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption--disable-fp-elim" title="Permalink to this definition">¶</a></dt>
+<dd><p>Disable frame pointer elimination optimization.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption--disable-excess-fp-precision">
+<tt class="descname">--disable-excess-fp-precision</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption--disable-excess-fp-precision" title="Permalink to this definition">¶</a></dt>
+<dd><p>Disable optimizations that may produce excess precision for floating point.
+Note that this option can dramatically slow down code on some systems
+(e.g. X86).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption--enable-no-infs-fp-math">
+<tt class="descname">--enable-no-infs-fp-math</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption--enable-no-infs-fp-math" title="Permalink to this definition">¶</a></dt>
+<dd><p>Enable optimizations that assume no Inf values.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption--enable-no-nans-fp-math">
+<tt class="descname">--enable-no-nans-fp-math</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption--enable-no-nans-fp-math" title="Permalink to this definition">¶</a></dt>
+<dd><p>Enable optimizations that assume no NAN values.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption--enable-unsafe-fp-math">
+<tt class="descname">--enable-unsafe-fp-math</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption--enable-unsafe-fp-math" title="Permalink to this definition">¶</a></dt>
+<dd><p>Enable optimizations that make unsafe assumptions about IEEE math (e.g. that
+addition is associative) or may not work for all input ranges. These
+optimizations allow the code generator to make use of some instructions which
+would otherwise not be usable (such as <tt class="docutils literal"><span class="pre">fsin</span></tt> on X86).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption--stats">
+<tt class="descname">--stats</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption--stats" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print statistics recorded by code-generation passes.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption--time-passes">
+<tt class="descname">--time-passes</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption--time-passes" title="Permalink to this definition">¶</a></dt>
+<dd><p>Record the amount of time needed for each pass and print a report to standard
+error.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption--load">
+<tt class="descname">--load</tt><tt class="descclassname">=<dso_path></tt><a class="headerlink" href="#cmdoption--load" title="Permalink to this definition">¶</a></dt>
+<dd><p>Dynamically load <tt class="docutils literal"><span class="pre">dso_path</span></tt> (a path to a dynamically shared object) that
+implements an LLVM target. This will permit the target name to be used with
+the <a class="reference internal" href="lli.html#cmdoption-march"><em class="xref std std-option">-march</em></a> option so that code can be generated for that target.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-meabi">
+<tt class="descname">-meabi</tt><tt class="descclassname">=[default|gnu|4|5]</tt><a class="headerlink" href="#cmdoption-meabi" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify which EABI version should conform to. Valid EABI versions are <em>gnu</em>,
+<em>4</em> and <em>5</em>. Default value (<em>default</em>) depends on the triple.</p>
+</dd></dl>
+
+</div>
+<div class="section" id="tuning-configuration-options">
+<h3>Tuning/Configuration Options<a class="headerlink" href="#tuning-configuration-options" title="Permalink to this headline">¶</a></h3>
+<dl class="option">
+<dt id="cmdoption--print-machineinstrs">
+<tt class="descname">--print-machineinstrs</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption--print-machineinstrs" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print generated machine code between compilation phases (useful for debugging).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption--regalloc">
+<tt class="descname">--regalloc</tt><tt class="descclassname">=<allocator></tt><a class="headerlink" href="#cmdoption--regalloc" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the register allocator to use.
+Valid register allocators are:</p>
+<p><em>basic</em></p>
+<blockquote>
+<div>Basic register allocator.</div></blockquote>
+<p><em>fast</em></p>
+<blockquote>
+<div>Fast register allocator. It is the default for unoptimized code.</div></blockquote>
+<p><em>greedy</em></p>
+<blockquote>
+<div>Greedy register allocator. It is the default for optimized code.</div></blockquote>
+<p><em>pbqp</em></p>
+<blockquote>
+<div>Register allocator based on ‘Partitioned Boolean Quadratic Programming’.</div></blockquote>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption--spiller">
+<tt class="descname">--spiller</tt><tt class="descclassname">=<spiller></tt><a class="headerlink" href="#cmdoption--spiller" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the spiller to use for register allocators that support it. Currently
+this option is used only by the linear scan register allocator. The default
+<tt class="docutils literal"><span class="pre">spiller</span></tt> is <em>local</em>. Valid spillers are:</p>
+<p><em>simple</em></p>
+<blockquote>
+<div>Simple spiller</div></blockquote>
+<p><em>local</em></p>
+<blockquote>
+<div>Local spiller</div></blockquote>
+</dd></dl>
+
+</div>
+<div class="section" id="intel-ia-32-specific-options">
+<h3>Intel IA-32-specific Options<a class="headerlink" href="#intel-ia-32-specific-options" title="Permalink to this headline">¶</a></h3>
+<dl class="option">
+<dt id="cmdoption--x86-asm-syntax">
+<tt class="descname">--x86-asm-syntax</tt><tt class="descclassname">=[att|intel]</tt><a class="headerlink" href="#cmdoption--x86-asm-syntax" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify whether to emit assembly code in AT&T syntax (the default) or Intel
+syntax.</p>
+</dd></dl>
+
+</div>
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p>If <strong class="program">llc</strong> succeeds, it will exit with 0. Otherwise, if an error
+occurs, it will exit with a non-zero value.</p>
+</div>
+<div class="section" id="see-also">
+<h2>SEE ALSO<a class="headerlink" href="#see-also" title="Permalink to this headline">¶</a></h2>
+<p>lli</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="lli.html" title="lli - directly execute programs from LLVM bitcode"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="opt.html" title="opt - LLVM optimizer"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/lli.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/lli.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/lli.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/lli.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,353 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>lli - directly execute programs from LLVM bitcode — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-link - LLVM bitcode linker" href="llvm-link.html" />
+ <link rel="prev" title="llc - LLVM static compiler" href="llc.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-link.html" title="llvm-link - LLVM bitcode linker"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llc.html" title="llc - LLVM static compiler"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="lli-directly-execute-programs-from-llvm-bitcode">
+<h1>lli - directly execute programs from LLVM bitcode<a class="headerlink" href="#lli-directly-execute-programs-from-llvm-bitcode" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">lli</strong> [<em>options</em>] [<em>filename</em>] [<em>program args</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">lli</strong> directly executes programs in LLVM bitcode format. It takes a program
+in LLVM bitcode format and executes it using a just-in-time compiler or an
+interpreter.</p>
+<p><strong class="program">lli</strong> is <em>not</em> an emulator. It will not execute IR of different architectures
+and it can only interpret (or JIT-compile) for the host architecture.</p>
+<p>The JIT compiler takes the same arguments as other tools, like <strong class="program">llc</strong>,
+but they don’t necessarily work for the interpreter.</p>
+<p>If <cite>filename</cite> is not specified, then <strong class="program">lli</strong> reads the LLVM bitcode for the
+program from standard input.</p>
+<p>The optional <em>args</em> specified on the command line are passed to the program as
+arguments.</p>
+</div>
+<div class="section" id="general-options">
+<h2>GENERAL OPTIONS<a class="headerlink" href="#general-options" title="Permalink to this headline">¶</a></h2>
+<dl class="option">
+<dt id="cmdoption-fake-argv0">
+<tt class="descname">-fake-argv0</tt><tt class="descclassname">=executable</tt><a class="headerlink" href="#cmdoption-fake-argv0" title="Permalink to this definition">¶</a></dt>
+<dd><p>Override the <tt class="docutils literal"><span class="pre">argv[0]</span></tt> value passed into the executing program.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-force-interpreter">
+<tt class="descname">-force-interpreter</tt><tt class="descclassname">={false,true}</tt><a class="headerlink" href="#cmdoption-force-interpreter" title="Permalink to this definition">¶</a></dt>
+<dd><p>If set to true, use the interpreter even if a just-in-time compiler is available
+for this architecture. Defaults to false.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-help">
+<tt class="descname">-help</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-help" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print a summary of command line options.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-load">
+<tt class="descname">-load</tt><tt class="descclassname">=pluginfilename</tt><a class="headerlink" href="#cmdoption-load" title="Permalink to this definition">¶</a></dt>
+<dd><p>Causes <strong class="program">lli</strong> to load the plugin (shared object) named <em>pluginfilename</em> and use
+it for optimization.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-stats">
+<tt class="descname">-stats</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-stats" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print statistics from the code-generation passes. This is only meaningful for
+the just-in-time compiler, at present.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-time-passes">
+<tt class="descname">-time-passes</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-time-passes" title="Permalink to this definition">¶</a></dt>
+<dd><p>Record the amount of time needed for each code-generation pass and print it to
+standard error.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-version">
+<tt class="descname">-version</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-version" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print out the version of <strong class="program">lli</strong> and exit without doing anything else.</p>
+</dd></dl>
+
+</div>
+<div class="section" id="target-options">
+<h2>TARGET OPTIONS<a class="headerlink" href="#target-options" title="Permalink to this headline">¶</a></h2>
+<dl class="option">
+<dt id="cmdoption-mtriple">
+<tt class="descname">-mtriple</tt><tt class="descclassname">=target triple</tt><a class="headerlink" href="#cmdoption-mtriple" title="Permalink to this definition">¶</a></dt>
+<dd><p>Override the target triple specified in the input bitcode file with the
+specified string. This may result in a crash if you pick an
+architecture which is not compatible with the current system.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-march">
+<tt class="descname">-march</tt><tt class="descclassname">=arch</tt><a class="headerlink" href="#cmdoption-march" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the architecture for which to generate assembly, overriding the target
+encoded in the bitcode file. See the output of <strong>llc -help</strong> for a list of
+valid architectures. By default this is inferred from the target triple or
+autodetected to the current architecture.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-mcpu">
+<tt class="descname">-mcpu</tt><tt class="descclassname">=cpuname</tt><a class="headerlink" href="#cmdoption-mcpu" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify a specific chip in the current architecture to generate code for.
+By default this is inferred from the target triple and autodetected to
+the current architecture. For a list of available CPUs, use:
+<strong>llvm-as < /dev/null | llc -march=xyz -mcpu=help</strong></p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-mattr">
+<tt class="descname">-mattr</tt><tt class="descclassname">=a1,+a2,-a3,...</tt><a class="headerlink" href="#cmdoption-mattr" title="Permalink to this definition">¶</a></dt>
+<dd><p>Override or control specific attributes of the target, such as whether SIMD
+operations are enabled or not. The default set of attributes is set by the
+current CPU. For a list of available attributes, use:
+<strong>llvm-as < /dev/null | llc -march=xyz -mattr=help</strong></p>
+</dd></dl>
+
+</div>
+<div class="section" id="floating-point-options">
+<h2>FLOATING POINT OPTIONS<a class="headerlink" href="#floating-point-options" title="Permalink to this headline">¶</a></h2>
+<dl class="option">
+<dt id="cmdoption-disable-excess-fp-precision">
+<tt class="descname">-disable-excess-fp-precision</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-disable-excess-fp-precision" title="Permalink to this definition">¶</a></dt>
+<dd><p>Disable optimizations that may increase floating point precision.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-enable-no-infs-fp-math">
+<tt class="descname">-enable-no-infs-fp-math</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-enable-no-infs-fp-math" title="Permalink to this definition">¶</a></dt>
+<dd><p>Enable optimizations that assume no Inf values.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-enable-no-nans-fp-math">
+<tt class="descname">-enable-no-nans-fp-math</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-enable-no-nans-fp-math" title="Permalink to this definition">¶</a></dt>
+<dd><p>Enable optimizations that assume no NAN values.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-enable-unsafe-fp-math">
+<tt class="descname">-enable-unsafe-fp-math</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-enable-unsafe-fp-math" title="Permalink to this definition">¶</a></dt>
+<dd><p>Causes <strong class="program">lli</strong> to enable optimizations that may decrease floating point
+precision.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-soft-float">
+<tt class="descname">-soft-float</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-soft-float" title="Permalink to this definition">¶</a></dt>
+<dd><p>Causes <strong class="program">lli</strong> to generate software floating point library calls instead of
+equivalent hardware instructions.</p>
+</dd></dl>
+
+</div>
+<div class="section" id="code-generation-options">
+<h2>CODE GENERATION OPTIONS<a class="headerlink" href="#code-generation-options" title="Permalink to this headline">¶</a></h2>
+<dl class="option">
+<dt id="cmdoption-code-model">
+<tt class="descname">-code-model</tt><tt class="descclassname">=model</tt><a class="headerlink" href="#cmdoption-code-model" title="Permalink to this definition">¶</a></dt>
+<dd><p>Choose the code model from:</p>
+<div class="highlight-perl"><div class="highlight"><pre><span class="n">default:</span> <span class="n">Target</span> <span class="n">default</span> <span class="n">code</span> <span class="n">model</span>
+<span class="n">small:</span> <span class="n">Small</span> <span class="n">code</span> <span class="n">model</span>
+<span class="n">kernel:</span> <span class="n">Kernel</span> <span class="n">code</span> <span class="n">model</span>
+<span class="n">medium:</span> <span class="n">Medium</span> <span class="n">code</span> <span class="n">model</span>
+<span class="n">large:</span> <span class="n">Large</span> <span class="n">code</span> <span class="n">model</span>
+</pre></div>
+</div>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-disable-post-RA-scheduler">
+<tt class="descname">-disable-post-RA-scheduler</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-disable-post-RA-scheduler" title="Permalink to this definition">¶</a></dt>
+<dd><p>Disable scheduling after register allocation.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-disable-spill-fusing">
+<tt class="descname">-disable-spill-fusing</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-disable-spill-fusing" title="Permalink to this definition">¶</a></dt>
+<dd><p>Disable fusing of spill code into instructions.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-jit-enable-eh">
+<tt class="descname">-jit-enable-eh</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-jit-enable-eh" title="Permalink to this definition">¶</a></dt>
+<dd><p>Exception handling should be enabled in the just-in-time compiler.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-join-liveintervals">
+<tt class="descname">-join-liveintervals</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-join-liveintervals" title="Permalink to this definition">¶</a></dt>
+<dd><p>Coalesce copies (default=true).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-nozero-initialized-in-bss">
+<tt class="descname">-nozero-initialized-in-bss</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-nozero-initialized-in-bss" title="Permalink to this definition">¶</a></dt>
+<dd><p>Don’t place zero-initialized symbols into the BSS section.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-pre-RA-sched">
+<tt class="descname">-pre-RA-sched</tt><tt class="descclassname">=scheduler</tt><a class="headerlink" href="#cmdoption-pre-RA-sched" title="Permalink to this definition">¶</a></dt>
+<dd><p>Instruction schedulers available (before register allocation):</p>
+<div class="highlight-perl"><div class="highlight"><pre><span class="o">=</span><span class="n">default:</span> <span class="n">Best</span> <span class="n">scheduler</span> <span class="k">for</span> <span class="n">the</span> <span class="n">target</span>
+<span class="o">=</span><span class="n">none:</span> <span class="n">No</span> <span class="n">scheduling:</span> <span class="n">breadth</span> <span class="n">first</span> <span class="n">sequencing</span>
+<span class="o">=</span><span class="n">simple:</span> <span class="n">Simple</span> <span class="n">two</span> <span class="n">pass</span> <span class="n">scheduling:</span> <span class="n">minimize</span> <span class="n">critical</span> <span class="n">path</span> <span class="ow">and</span> <span class="n">maximize</span> <span class="n">processor</span> <span class="n">utilization</span>
+<span class="o">=</span><span class="n">simple</span><span class="o">-</span><span class="n">noitin:</span> <span class="n">Simple</span> <span class="n">two</span> <span class="n">pass</span> <span class="n">scheduling:</span> <span class="n">Same</span> <span class="n">as</span> <span class="n">simple</span> <span class="n">except</span> <span class="n">using</span> <span class="n">generic</span> <span class="n">latency</span>
+<span class="o">=</span><span class="n">list</span><span class="o">-</span><span class="n">burr:</span> <span class="n">Bottom</span><span class="o">-</span><span class="n">up</span> <span class="n">register</span> <span class="n">reduction</span> <span class="n">list</span> <span class="n">scheduling</span>
+<span class="o">=</span><span class="n">list</span><span class="o">-</span><span class="n">tdrr:</span> <span class="n">Top</span><span class="o">-</span><span class="n">down</span> <span class="n">register</span> <span class="n">reduction</span> <span class="n">list</span> <span class="n">scheduling</span>
+<span class="o">=</span><span class="n">list</span><span class="o">-</span><span class="n">td:</span> <span class="n">Top</span><span class="o">-</span><span class="n">down</span> <span class="n">list</span> <span class="n">scheduler</span> <span class="o">-</span><span class="k">print</span><span class="o">-</span><span class="n">machineinstrs</span> <span class="o">-</span> <span class="n">Print</span> <span class="n">generated</span> <span class="n">machine</span> <span class="n">code</span>
+</pre></div>
+</div>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-regalloc">
+<tt class="descname">-regalloc</tt><tt class="descclassname">=allocator</tt><a class="headerlink" href="#cmdoption-regalloc" title="Permalink to this definition">¶</a></dt>
+<dd><p>Register allocator to use (default=linearscan)</p>
+<div class="highlight-perl"><div class="highlight"><pre><span class="o">=</span><span class="n">bigblock:</span> <span class="n">Big</span><span class="o">-</span><span class="n">block</span> <span class="n">register</span> <span class="n">allocator</span>
+<span class="o">=</span><span class="n">linearscan:</span> <span class="n">linear</span> <span class="n">scan</span> <span class="n">register</span> <span class="n">allocator</span> <span class="o">=</span><span class="nb">local</span> <span class="o">-</span> <span class="nb">local</span> <span class="n">register</span> <span class="n">allocator</span>
+<span class="o">=</span><span class="n">simple:</span> <span class="n">simple</span> <span class="n">register</span> <span class="n">allocator</span>
+</pre></div>
+</div>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-relocation-model">
+<tt class="descname">-relocation-model</tt><tt class="descclassname">=model</tt><a class="headerlink" href="#cmdoption-relocation-model" title="Permalink to this definition">¶</a></dt>
+<dd><p>Choose relocation model from:</p>
+<div class="highlight-perl"><div class="highlight"><pre><span class="o">=</span><span class="n">default:</span> <span class="n">Target</span> <span class="n">default</span> <span class="n">relocation</span> <span class="n">model</span>
+<span class="o">=</span><span class="n">static:</span> <span class="n">Non</span><span class="o">-</span><span class="n">relocatable</span> <span class="n">code</span> <span class="o">=</span><span class="n">pic</span> <span class="o">-</span> <span class="n">Fully</span> <span class="n">relocatable</span><span class="p">,</span> <span class="n">position</span> <span class="n">independent</span> <span class="n">code</span>
+<span class="o">=</span><span class="n">dynamic</span><span class="o">-</span><span class="nb">no</span><span class="o">-</span><span class="n">pic:</span> <span class="n">Relocatable</span> <span class="n">external</span> <span class="n">references</span><span class="p">,</span> <span class="n">non</span><span class="o">-</span><span class="n">relocatable</span> <span class="n">code</span>
+</pre></div>
+</div>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-spiller">
+<tt class="descname">-spiller</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-spiller" title="Permalink to this definition">¶</a></dt>
+<dd><p>Spiller to use (default=local)</p>
+<div class="highlight-perl"><div class="highlight"><pre><span class="o">=</span><span class="n">simple:</span> <span class="n">simple</span> <span class="n">spiller</span>
+<span class="o">=</span><span class="nb">local</span><span class="p">:</span> <span class="nb">local</span> <span class="n">spiller</span>
+</pre></div>
+</div>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-x86-asm-syntax">
+<tt class="descname">-x86-asm-syntax</tt><tt class="descclassname">=syntax</tt><a class="headerlink" href="#cmdoption-x86-asm-syntax" title="Permalink to this definition">¶</a></dt>
+<dd><p>Choose style of code to emit from X86 backend:</p>
+<div class="highlight-perl"><div class="highlight"><pre><span class="o">=</span><span class="n">att:</span> <span class="n">Emit</span> <span class="n">AT</span><span class="o">&</span><span class="n">T</span><span class="o">-</span><span class="n">style</span> <span class="n">assembly</span>
+<span class="o">=</span><span class="n">intel:</span> <span class="n">Emit</span> <span class="n">Intel</span><span class="o">-</span><span class="n">style</span> <span class="n">assembly</span>
+</pre></div>
+</div>
+</dd></dl>
+
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p>If <strong class="program">lli</strong> fails to load the program, it will exit with an exit code of 1.
+Otherwise, it will return the exit code of the program it executes.</p>
+</div>
+<div class="section" id="see-also">
+<h2>SEE ALSO<a class="headerlink" href="#see-also" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llc</strong></p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-link.html" title="llvm-link - LLVM bitcode linker"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llc.html" title="llc - LLVM static compiler"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-ar.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-ar.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-ar.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-ar.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,346 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-ar - LLVM archiver — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-lib - LLVM lib.exe compatible library tool" href="llvm-lib.html" />
+ <link rel="prev" title="llvm-link - LLVM bitcode linker" href="llvm-link.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-lib.html" title="llvm-lib - LLVM lib.exe compatible library tool"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-link.html" title="llvm-link - LLVM bitcode linker"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-ar-llvm-archiver">
+<h1>llvm-ar - LLVM archiver<a class="headerlink" href="#llvm-ar-llvm-archiver" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong>llvm-ar</strong> [-]{dmpqrtx}[Rabfikou] [relpos] [count] <archive> [files...]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p>The <strong>llvm-ar</strong> command is similar to the common Unix utility, <tt class="docutils literal"><span class="pre">ar</span></tt>. It
+archives several files together into a single file. The intent for this is
+to produce archive libraries by LLVM bitcode that can be linked into an
+LLVM program. However, the archive can contain any kind of file. By default,
+<strong>llvm-ar</strong> generates a symbol table that makes linking faster because
+only the symbol table needs to be consulted, not each individual file member
+of the archive.</p>
+<p>The <strong>llvm-ar</strong> command can be used to <em>read</em> SVR4, GNU and BSD style archive
+files. However, right now it can only write in the GNU format. If an
+SVR4 or BSD style archive is used with the <tt class="docutils literal"><span class="pre">r</span></tt> (replace) or <tt class="docutils literal"><span class="pre">q</span></tt> (quick
+update) operations, the archive will be reconstructed in GNU format.</p>
+<p>Here’s where <strong>llvm-ar</strong> departs from previous <tt class="docutils literal"><span class="pre">ar</span></tt> implementations:</p>
+<p><em>Symbol Table</em></p>
+<blockquote>
+<div>Since <strong>llvm-ar</strong> supports bitcode files. The symbol table it creates
+is in GNU format and includes both native and bitcode files.</div></blockquote>
+<p><em>Long Paths</em></p>
+<blockquote>
+<div>Currently <strong>llvm-ar</strong> can read GNU and BSD long file names, but only writes
+archives with the GNU format.</div></blockquote>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<p>The options to <strong>llvm-ar</strong> are compatible with other <tt class="docutils literal"><span class="pre">ar</span></tt> implementations.
+However, there are a few modifiers (<em>R</em>) that are not found in other <tt class="docutils literal"><span class="pre">ar</span></tt>
+implementations. The options to <strong>llvm-ar</strong> specify a single basic operation to
+perform on the archive, a variety of modifiers for that operation, the name of
+the archive file, and an optional list of file names. These options are used to
+determine how <strong>llvm-ar</strong> should process the archive file.</p>
+<p>The Operations and Modifiers are explained in the sections below. The minimal
+set of options is at least one operator and the name of the archive. Typically
+archive files end with a <tt class="docutils literal"><span class="pre">.a</span></tt> suffix, but this is not required. Following
+the <em>archive-name</em> comes a list of <em>files</em> that indicate the specific members
+of the archive to operate on. If the <em>files</em> option is not specified, it
+generally means either “none” or “all” members, depending on the operation.</p>
+<div class="section" id="operations">
+<h3>Operations<a class="headerlink" href="#operations" title="Permalink to this headline">¶</a></h3>
+<p>d</p>
+<blockquote>
+<div>Delete files from the archive. No modifiers are applicable to this operation.
+The <em>files</em> options specify which members should be removed from the
+archive. It is not an error if a specified file does not appear in the archive.
+If no <em>files</em> are specified, the archive is not modified.</div></blockquote>
+<p>m[abi]</p>
+<blockquote>
+<div>Move files from one location in the archive to another. The <em>a</em>, <em>b</em>, and
+<em>i</em> modifiers apply to this operation. The <em>files</em> will all be moved
+to the location given by the modifiers. If no modifiers are used, the files
+will be moved to the end of the archive. If no <em>files</em> are specified, the
+archive is not modified.</div></blockquote>
+<p>p</p>
+<blockquote>
+<div>Print files to the standard output. This operation simply prints the
+<em>files</em> indicated to the standard output. If no <em>files</em> are
+specified, the entire archive is printed. Printing bitcode files is
+ill-advised as they might confuse your terminal settings. The <em>p</em>
+operation never modifies the archive.</div></blockquote>
+<p>q</p>
+<blockquote>
+<div>Quickly append files to the end of the archive. This operation quickly adds the
+<em>files</em> to the archive without checking for duplicates that should be
+removed first. If no <em>files</em> are specified, the archive is not modified.
+Because of the way that <strong>llvm-ar</strong> constructs the archive file, its dubious
+whether the <em>q</em> operation is any faster than the <em>r</em> operation.</div></blockquote>
+<p>r[abu]</p>
+<blockquote>
+<div>Replace or insert file members. The <em>a</em>, <em>b</em>, and <em>u</em>
+modifiers apply to this operation. This operation will replace existing
+<em>files</em> or insert them at the end of the archive if they do not exist. If no
+<em>files</em> are specified, the archive is not modified.</div></blockquote>
+<p>t[v]</p>
+<blockquote>
+<div>Print the table of contents. Without any modifiers, this operation just prints
+the names of the members to the standard output. With the <em>v</em> modifier,
+<strong>llvm-ar</strong> also prints out the file type (B=bitcode, S=symbol
+table, blank=regular file), the permission mode, the owner and group, the
+size, and the date. If any <em>files</em> are specified, the listing is only for
+those files. If no <em>files</em> are specified, the table of contents for the
+whole archive is printed.</div></blockquote>
+<p>x[oP]</p>
+<blockquote>
+<div>Extract archive members back to files. The <em>o</em> modifier applies to this
+operation. This operation retrieves the indicated <em>files</em> from the archive
+and writes them back to the operating system’s file system. If no
+<em>files</em> are specified, the entire archive is extract.</div></blockquote>
+</div>
+<div class="section" id="modifiers-operation-specific">
+<h3>Modifiers (operation specific)<a class="headerlink" href="#modifiers-operation-specific" title="Permalink to this headline">¶</a></h3>
+<p>The modifiers below are specific to certain operations. See the Operations
+section (above) to determine which modifiers are applicable to which operations.</p>
+<p>[a]</p>
+<blockquote>
+<div>When inserting or moving member files, this option specifies the destination of
+the new files as being after the <em>relpos</em> member. If <em>relpos</em> is not found,
+the files are placed at the end of the archive.</div></blockquote>
+<p>[b]</p>
+<blockquote>
+<div>When inserting or moving member files, this option specifies the destination of
+the new files as being before the <em>relpos</em> member. If <em>relpos</em> is not
+found, the files are placed at the end of the archive. This modifier is
+identical to the <em>i</em> modifier.</div></blockquote>
+<p>[i]</p>
+<blockquote>
+<div>A synonym for the <em>b</em> option.</div></blockquote>
+<p>[o]</p>
+<blockquote>
+<div>When extracting files, this option will cause <strong>llvm-ar</strong> to preserve the
+original modification times of the files it writes.</div></blockquote>
+<p>[u]</p>
+<blockquote>
+<div>When replacing existing files in the archive, only replace those files that have
+a time stamp than the time stamp of the member in the archive.</div></blockquote>
+</div>
+<div class="section" id="modifiers-generic">
+<h3>Modifiers (generic)<a class="headerlink" href="#modifiers-generic" title="Permalink to this headline">¶</a></h3>
+<p>The modifiers below may be applied to any operation.</p>
+<p>[c]</p>
+<blockquote>
+<div>For all operations, <strong>llvm-ar</strong> will always create the archive if it doesn’t
+exist. Normally, <strong>llvm-ar</strong> will print a warning message indicating that the
+archive is being created. Using this modifier turns off that warning.</div></blockquote>
+<p>[s]</p>
+<blockquote>
+<div>This modifier requests that an archive index (or symbol table) be added to the
+archive. This is the default mode of operation. The symbol table will contain
+all the externally visible functions and global variables defined by all the
+bitcode files in the archive.</div></blockquote>
+<p>[S]</p>
+<blockquote>
+<div>This modifier is the opposite of the <em>s</em> modifier. It instructs <strong>llvm-ar</strong> to
+not build the symbol table. If both <em>s</em> and <em>S</em> are used, the last modifier to
+occur in the options will prevail.</div></blockquote>
+<p>[v]</p>
+<blockquote>
+<div>This modifier instructs <strong>llvm-ar</strong> to be verbose about what it is doing. Each
+editing operation taken against the archive will produce a line of output saying
+what is being done.</div></blockquote>
+</div>
+</div>
+<div class="section" id="standards">
+<h2>STANDARDS<a class="headerlink" href="#standards" title="Permalink to this headline">¶</a></h2>
+<p>The <strong>llvm-ar</strong> utility is intended to provide a superset of the IEEE Std 1003.2
+(POSIX.2) functionality for <tt class="docutils literal"><span class="pre">ar</span></tt>. <strong>llvm-ar</strong> can read both SVR4 and BSD4.4 (or
+Mac OS X) archives. If the <tt class="docutils literal"><span class="pre">f</span></tt> modifier is given to the <tt class="docutils literal"><span class="pre">x</span></tt> or <tt class="docutils literal"><span class="pre">r</span></tt> operations
+then <strong>llvm-ar</strong> will write SVR4 compatible archives. Without this modifier,
+<strong>llvm-ar</strong> will write BSD4.4 compatible archives that have long names
+immediately after the header and indicated using the “#1/ddd” notation for the
+name in the header.</p>
+</div>
+<div class="section" id="file-format">
+<h2>FILE FORMAT<a class="headerlink" href="#file-format" title="Permalink to this headline">¶</a></h2>
+<p>The file format for LLVM Archive files is similar to that of BSD 4.4 or Mac OSX
+archive files. In fact, except for the symbol table, the <tt class="docutils literal"><span class="pre">ar</span></tt> commands on those
+operating systems should be able to read LLVM archive files. The details of the
+file format follow.</p>
+<p>Each archive begins with the archive magic number which is the eight printable
+characters ”!<arch>n” where n represents the newline character (0x0A).
+Following the magic number, the file is composed of even length members that
+begin with an archive header and end with a n padding character if necessary
+(to make the length even). Each file member is composed of a header (defined
+below), an optional newline-terminated “long file name” and the contents of
+the file.</p>
+<p>The fields of the header are described in the items below. All fields of the
+header contain only ASCII characters, are left justified and are right padded
+with space characters.</p>
+<p>name - char[16]</p>
+<blockquote>
+<div>This field of the header provides the name of the archive member. If the name is
+longer than 15 characters or contains a slash (/) character, then this field
+contains <tt class="docutils literal"><span class="pre">#1/nnn</span></tt> where <tt class="docutils literal"><span class="pre">nnn</span></tt> provides the length of the name and the <tt class="docutils literal"><span class="pre">#1/</span></tt>
+is literal. In this case, the actual name of the file is provided in the <tt class="docutils literal"><span class="pre">nnn</span></tt>
+bytes immediately following the header. If the name is 15 characters or less, it
+is contained directly in this field and terminated with a slash (/) character.</div></blockquote>
+<p>date - char[12]</p>
+<blockquote>
+<div>This field provides the date of modification of the file in the form of a
+decimal encoded number that provides the number of seconds since the epoch
+(since 00:00:00 Jan 1, 1970) per Posix specifications.</div></blockquote>
+<p>uid - char[6]</p>
+<blockquote>
+<div>This field provides the user id of the file encoded as a decimal ASCII string.
+This field might not make much sense on non-Unix systems. On Unix, it is the
+same value as the st_uid field of the stat structure returned by the stat(2)
+operating system call.</div></blockquote>
+<p>gid - char[6]</p>
+<blockquote>
+<div>This field provides the group id of the file encoded as a decimal ASCII string.
+This field might not make much sense on non-Unix systems. On Unix, it is the
+same value as the st_gid field of the stat structure returned by the stat(2)
+operating system call.</div></blockquote>
+<p>mode - char[8]</p>
+<blockquote>
+<div>This field provides the access mode of the file encoded as an octal ASCII
+string. This field might not make much sense on non-Unix systems. On Unix, it
+is the same value as the st_mode field of the stat structure returned by the
+stat(2) operating system call.</div></blockquote>
+<p>size - char[10]</p>
+<blockquote>
+<div>This field provides the size of the file, in bytes, encoded as a decimal ASCII
+string.</div></blockquote>
+<p>fmag - char[2]</p>
+<blockquote>
+<div>This field is the archive file member magic number. Its content is always the
+two characters back tick (0x60) and newline (0x0A). This provides some measure
+utility in identifying archive files that have been corrupted.</div></blockquote>
+<p>offset - vbr encoded 32-bit integer</p>
+<blockquote>
+<div>The offset item provides the offset into the archive file where the bitcode
+member is stored that is associated with the symbol. The offset value is 0
+based at the start of the first “normal” file member. To derive the actual
+file offset of the member, you must add the number of bytes occupied by the file
+signature (8 bytes) and the symbol tables. The value of this item is encoded
+using variable bit rate encoding to reduce the size of the symbol table.
+Variable bit rate encoding uses the high bit (0x80) of each byte to indicate
+if there are more bytes to follow. The remaining 7 bits in each byte carry bits
+from the value. The final byte does not have the high bit set.</div></blockquote>
+<p>length - vbr encoded 32-bit integer</p>
+<blockquote>
+<div>The length item provides the length of the symbol that follows. Like this
+<em>offset</em> item, the length is variable bit rate encoded.</div></blockquote>
+<p>symbol - character array</p>
+<blockquote>
+<div>The symbol item provides the text of the symbol that is associated with the
+<em>offset</em>. The symbol is not terminated by any character. Its length is provided
+by the <em>length</em> field. Note that is allowed (but unwise) to use non-printing
+characters (even 0x00) in the symbol. This allows for multiple encodings of
+symbol names.</div></blockquote>
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p>If <strong>llvm-ar</strong> succeeds, it will exit with 0. A usage error, results
+in an exit code of 1. A hard (file system typically) error results in an
+exit code of 2. Miscellaneous or unknown errors result in an
+exit code of 3.</p>
+</div>
+<div class="section" id="see-also">
+<h2>SEE ALSO<a class="headerlink" href="#see-also" title="Permalink to this headline">¶</a></h2>
+<p>ar(1)</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-lib.html" title="llvm-lib - LLVM lib.exe compatible library tool"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-link.html" title="llvm-link - LLVM bitcode linker"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-as.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-as.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-as.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-as.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,147 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-as - LLVM assembler — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-dis - LLVM disassembler" href="llvm-dis.html" />
+ <link rel="prev" title="LLVM Command Guide" href="index.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-dis.html" title="llvm-dis - LLVM disassembler"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="index.html" title="LLVM Command Guide"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-as-llvm-assembler">
+<h1>llvm-as - LLVM assembler<a class="headerlink" href="#llvm-as-llvm-assembler" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong>llvm-as</strong> [<em>options</em>] [<em>filename</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p><strong>llvm-as</strong> is the LLVM assembler. It reads a file containing human-readable
+LLVM assembly language, translates it to LLVM bitcode, and writes the result
+into a file or to standard output.</p>
+<p>If <em>filename</em> is omitted or is <tt class="docutils literal"><span class="pre">-</span></tt>, then <strong>llvm-as</strong> reads its input from
+standard input.</p>
+<p>If an output file is not specified with the <strong>-o</strong> option, then
+<strong>llvm-as</strong> sends its output to a file or standard output by following
+these rules:</p>
+<ul class="simple">
+<li>If the input is standard input, then the output is standard output.</li>
+<li>If the input is a file that ends with <tt class="docutils literal"><span class="pre">.ll</span></tt>, then the output file is of the
+same name, except that the suffix is changed to <tt class="docutils literal"><span class="pre">.bc</span></tt>.</li>
+<li>If the input is a file that does not end with the <tt class="docutils literal"><span class="pre">.ll</span></tt> suffix, then the
+output file has the same name as the input file, except that the <tt class="docutils literal"><span class="pre">.bc</span></tt>
+suffix is appended.</li>
+</ul>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<dl class="docutils">
+<dt><strong>-f</strong></dt>
+<dd>Enable binary output on terminals. Normally, <strong>llvm-as</strong> will refuse to
+write raw bitcode output if the output stream is a terminal. With this option,
+<strong>llvm-as</strong> will write raw bitcode regardless of the output device.</dd>
+<dt><strong>-help</strong></dt>
+<dd>Print a summary of command line options.</dd>
+<dt><strong>-o</strong> <em>filename</em></dt>
+<dd>Specify the output file name. If <em>filename</em> is <tt class="docutils literal"><span class="pre">-</span></tt>, then <strong>llvm-as</strong>
+sends its output to standard output.</dd>
+</dl>
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p>If <strong>llvm-as</strong> succeeds, it will exit with 0. Otherwise, if an error occurs, it
+will exit with a non-zero value.</p>
+</div>
+<div class="section" id="see-also">
+<h2>SEE ALSO<a class="headerlink" href="#see-also" title="Permalink to this headline">¶</a></h2>
+<p>llvm-dis|llvm-dis, gccas|gccas</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-dis.html" title="llvm-dis - LLVM disassembler"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="index.html" title="LLVM Command Guide"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-bcanalyzer.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-bcanalyzer.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-bcanalyzer.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-bcanalyzer.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,353 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-bcanalyzer - LLVM bitcode analyzer — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="FileCheck - Flexible pattern matching file verifier" href="FileCheck.html" />
+ <link rel="prev" title="llvm-extract - extract a function from an LLVM module" href="llvm-extract.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="FileCheck.html" title="FileCheck - Flexible pattern matching file verifier"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-extract.html" title="llvm-extract - extract a function from an LLVM module"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-bcanalyzer-llvm-bitcode-analyzer">
+<h1>llvm-bcanalyzer - LLVM bitcode analyzer<a class="headerlink" href="#llvm-bcanalyzer-llvm-bitcode-analyzer" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-bcanalyzer</strong> [<em>options</em>] [<em>filename</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p>The <strong class="program">llvm-bcanalyzer</strong> command is a small utility for analyzing bitcode
+files. The tool reads a bitcode file (such as generated with the
+<strong class="program">llvm-as</strong> tool) and produces a statistical report on the contents of
+the bitcode file. The tool can also dump a low level but human readable
+version of the bitcode file. This tool is probably not of much interest or
+utility except for those working directly with the bitcode file format. Most
+LLVM users can just ignore this tool.</p>
+<p>If <em>filename</em> is omitted or is <tt class="docutils literal"><span class="pre">-</span></tt>, then <strong class="program">llvm-bcanalyzer</strong> reads its
+input from standard input. This is useful for combining the tool into a
+pipeline. Output is written to the standard output.</p>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<dl class="option">
+<dt id="cmdoption-llvm-bcanalyzer-nodetails">
+<tt class="descname">-nodetails</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-bcanalyzer-nodetails" title="Permalink to this definition">¶</a></dt>
+<dd><p>Causes <strong class="program">llvm-bcanalyzer</strong> to abbreviate its output by writing out only
+a module level summary. The details for individual functions are not
+displayed.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-bcanalyzer-dump">
+<tt class="descname">-dump</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-bcanalyzer-dump" title="Permalink to this definition">¶</a></dt>
+<dd><p>Causes <strong class="program">llvm-bcanalyzer</strong> to dump the bitcode in a human readable
+format. This format is significantly different from LLVM assembly and
+provides details about the encoding of the bitcode file.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-bcanalyzer-verify">
+<tt class="descname">-verify</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-bcanalyzer-verify" title="Permalink to this definition">¶</a></dt>
+<dd><p>Causes <strong class="program">llvm-bcanalyzer</strong> to verify the module produced by reading the
+bitcode. This ensures that the statistics generated are based on a consistent
+module.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-bcanalyzer-help">
+<tt class="descname">-help</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-bcanalyzer-help" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print a summary of command line options.</p>
+</dd></dl>
+
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p>If <strong class="program">llvm-bcanalyzer</strong> succeeds, it will exit with 0. Otherwise, if an
+error occurs, it will exit with a non-zero value, usually 1.</p>
+</div>
+<div class="section" id="summary-output-definitions">
+<h2>SUMMARY OUTPUT DEFINITIONS<a class="headerlink" href="#summary-output-definitions" title="Permalink to this headline">¶</a></h2>
+<p>The following items are always printed by llvm-bcanalyzer. They comprize the
+summary output.</p>
+<p><strong>Bitcode Analysis Of Module</strong></p>
+<blockquote>
+<div>This just provides the name of the module for which bitcode analysis is being
+generated.</div></blockquote>
+<p><strong>Bitcode Version Number</strong></p>
+<blockquote>
+<div>The bitcode version (not LLVM version) of the file read by the analyzer.</div></blockquote>
+<p><strong>File Size</strong></p>
+<blockquote>
+<div>The size, in bytes, of the entire bitcode file.</div></blockquote>
+<p><strong>Module Bytes</strong></p>
+<blockquote>
+<div>The size, in bytes, of the module block. Percentage is relative to File Size.</div></blockquote>
+<p><strong>Function Bytes</strong></p>
+<blockquote>
+<div>The size, in bytes, of all the function blocks. Percentage is relative to File
+Size.</div></blockquote>
+<p><strong>Global Types Bytes</strong></p>
+<blockquote>
+<div>The size, in bytes, of the Global Types Pool. Percentage is relative to File
+Size. This is the size of the definitions of all types in the bitcode file.</div></blockquote>
+<p><strong>Constant Pool Bytes</strong></p>
+<blockquote>
+<div>The size, in bytes, of the Constant Pool Blocks Percentage is relative to File
+Size.</div></blockquote>
+<p><strong>Module Globals Bytes</strong></p>
+<blockquote>
+<div>Ths size, in bytes, of the Global Variable Definitions and their initializers.
+Percentage is relative to File Size.</div></blockquote>
+<p><strong>Instruction List Bytes</strong></p>
+<blockquote>
+<div>The size, in bytes, of all the instruction lists in all the functions.
+Percentage is relative to File Size. Note that this value is also included in
+the Function Bytes.</div></blockquote>
+<p><strong>Compaction Table Bytes</strong></p>
+<blockquote>
+<div>The size, in bytes, of all the compaction tables in all the functions.
+Percentage is relative to File Size. Note that this value is also included in
+the Function Bytes.</div></blockquote>
+<p><strong>Symbol Table Bytes</strong></p>
+<blockquote>
+<div>The size, in bytes, of all the symbol tables in all the functions. Percentage is
+relative to File Size. Note that this value is also included in the Function
+Bytes.</div></blockquote>
+<p><strong>Dependent Libraries Bytes</strong></p>
+<blockquote>
+<div>The size, in bytes, of the list of dependent libraries in the module. Percentage
+is relative to File Size. Note that this value is also included in the Module
+Global Bytes.</div></blockquote>
+<p><strong>Number Of Bitcode Blocks</strong></p>
+<blockquote>
+<div>The total number of blocks of any kind in the bitcode file.</div></blockquote>
+<p><strong>Number Of Functions</strong></p>
+<blockquote>
+<div>The total number of function definitions in the bitcode file.</div></blockquote>
+<p><strong>Number Of Types</strong></p>
+<blockquote>
+<div>The total number of types defined in the Global Types Pool.</div></blockquote>
+<p><strong>Number Of Constants</strong></p>
+<blockquote>
+<div>The total number of constants (of any type) defined in the Constant Pool.</div></blockquote>
+<p><strong>Number Of Basic Blocks</strong></p>
+<blockquote>
+<div>The total number of basic blocks defined in all functions in the bitcode file.</div></blockquote>
+<p><strong>Number Of Instructions</strong></p>
+<blockquote>
+<div>The total number of instructions defined in all functions in the bitcode file.</div></blockquote>
+<p><strong>Number Of Long Instructions</strong></p>
+<blockquote>
+<div>The total number of long instructions defined in all functions in the bitcode
+file. Long instructions are those taking greater than 4 bytes. Typically long
+instructions are GetElementPtr with several indices, PHI nodes, and calls to
+functions with large numbers of arguments.</div></blockquote>
+<p><strong>Number Of Operands</strong></p>
+<blockquote>
+<div>The total number of operands used in all instructions in the bitcode file.</div></blockquote>
+<p><strong>Number Of Compaction Tables</strong></p>
+<blockquote>
+<div>The total number of compaction tables in all functions in the bitcode file.</div></blockquote>
+<p><strong>Number Of Symbol Tables</strong></p>
+<blockquote>
+<div>The total number of symbol tables in all functions in the bitcode file.</div></blockquote>
+<p><strong>Number Of Dependent Libs</strong></p>
+<blockquote>
+<div>The total number of dependent libraries found in the bitcode file.</div></blockquote>
+<p><strong>Total Instruction Size</strong></p>
+<blockquote>
+<div>The total size of the instructions in all functions in the bitcode file.</div></blockquote>
+<p><strong>Average Instruction Size</strong></p>
+<blockquote>
+<div>The average number of bytes per instruction across all functions in the bitcode
+file. This value is computed by dividing Total Instruction Size by Number Of
+Instructions.</div></blockquote>
+<p><strong>Maximum Type Slot Number</strong></p>
+<blockquote>
+<div>The maximum value used for a type’s slot number. Larger slot number values take
+more bytes to encode.</div></blockquote>
+<p><strong>Maximum Value Slot Number</strong></p>
+<blockquote>
+<div>The maximum value used for a value’s slot number. Larger slot number values take
+more bytes to encode.</div></blockquote>
+<p><strong>Bytes Per Value</strong></p>
+<blockquote>
+<div>The average size of a Value definition (of any type). This is computed by
+dividing File Size by the total number of values of any type.</div></blockquote>
+<p><strong>Bytes Per Global</strong></p>
+<blockquote>
+<div>The average size of a global definition (constants and global variables).</div></blockquote>
+<p><strong>Bytes Per Function</strong></p>
+<blockquote>
+<div>The average number of bytes per function definition. This is computed by
+dividing Function Bytes by Number Of Functions.</div></blockquote>
+<p><strong># of VBR 32-bit Integers</strong></p>
+<blockquote>
+<div>The total number of 32-bit integers encoded using the Variable Bit Rate
+encoding scheme.</div></blockquote>
+<p><strong># of VBR 64-bit Integers</strong></p>
+<blockquote>
+<div>The total number of 64-bit integers encoded using the Variable Bit Rate encoding
+scheme.</div></blockquote>
+<p><strong># of VBR Compressed Bytes</strong></p>
+<blockquote>
+<div>The total number of bytes consumed by the 32-bit and 64-bit integers that use
+the Variable Bit Rate encoding scheme.</div></blockquote>
+<p><strong># of VBR Expanded Bytes</strong></p>
+<blockquote>
+<div>The total number of bytes that would have been consumed by the 32-bit and 64-bit
+integers had they not been compressed with the Variable Bit Rage encoding
+scheme.</div></blockquote>
+<p><strong>Bytes Saved With VBR</strong></p>
+<blockquote>
+<div>The total number of bytes saved by using the Variable Bit Rate encoding scheme.
+The percentage is relative to # of VBR Expanded Bytes.</div></blockquote>
+</div>
+<div class="section" id="detailed-output-definitions">
+<h2>DETAILED OUTPUT DEFINITIONS<a class="headerlink" href="#detailed-output-definitions" title="Permalink to this headline">¶</a></h2>
+<p>The following definitions occur only if the -nodetails option was not given.
+The detailed output provides additional information on a per-function basis.</p>
+<p><strong>Type</strong></p>
+<blockquote>
+<div>The type signature of the function.</div></blockquote>
+<p><strong>Byte Size</strong></p>
+<blockquote>
+<div>The total number of bytes in the function’s block.</div></blockquote>
+<p><strong>Basic Blocks</strong></p>
+<blockquote>
+<div>The number of basic blocks defined by the function.</div></blockquote>
+<p><strong>Instructions</strong></p>
+<blockquote>
+<div>The number of instructions defined by the function.</div></blockquote>
+<p><strong>Long Instructions</strong></p>
+<blockquote>
+<div>The number of instructions using the long instruction format in the function.</div></blockquote>
+<p><strong>Operands</strong></p>
+<blockquote>
+<div>The number of operands used by all instructions in the function.</div></blockquote>
+<p><strong>Instruction Size</strong></p>
+<blockquote>
+<div>The number of bytes consumed by instructions in the function.</div></blockquote>
+<p><strong>Average Instruction Size</strong></p>
+<blockquote>
+<div>The average number of bytes consumed by the instructions in the function.
+This value is computed by dividing Instruction Size by Instructions.</div></blockquote>
+<p><strong>Bytes Per Instruction</strong></p>
+<blockquote>
+<div>The average number of bytes used by the function per instruction. This value
+is computed by dividing Byte Size by Instructions. Note that this is not the
+same as Average Instruction Size. It computes a number relative to the total
+function size not just the size of the instruction list.</div></blockquote>
+<p><strong>Number of VBR 32-bit Integers</strong></p>
+<blockquote>
+<div>The total number of 32-bit integers found in this function (for any use).</div></blockquote>
+<p><strong>Number of VBR 64-bit Integers</strong></p>
+<blockquote>
+<div>The total number of 64-bit integers found in this function (for any use).</div></blockquote>
+<p><strong>Number of VBR Compressed Bytes</strong></p>
+<blockquote>
+<div>The total number of bytes in this function consumed by the 32-bit and 64-bit
+integers that use the Variable Bit Rate encoding scheme.</div></blockquote>
+<p><strong>Number of VBR Expanded Bytes</strong></p>
+<blockquote>
+<div>The total number of bytes in this function that would have been consumed by
+the 32-bit and 64-bit integers had they not been compressed with the Variable
+Bit Rate encoding scheme.</div></blockquote>
+<p><strong>Bytes Saved With VBR</strong></p>
+<blockquote>
+<div>The total number of bytes saved in this function by using the Variable Bit
+Rate encoding scheme. The percentage is relative to # of VBR Expanded Bytes.</div></blockquote>
+</div>
+<div class="section" id="see-also">
+<h2>SEE ALSO<a class="headerlink" href="#see-also" title="Permalink to this headline">¶</a></h2>
+<p><a class="reference internal" href="llvm-dis.html"><em>llvm-dis - LLVM disassembler</em></a>, <a class="reference internal" href="../BitCodeFormat.html"><em>LLVM Bitcode File Format</em></a></p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="FileCheck.html" title="FileCheck - Flexible pattern matching file verifier"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-extract.html" title="llvm-extract - extract a function from an LLVM module"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-build.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-build.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-build.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-build.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,164 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-build - LLVM Project Build Utility — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-readobj - LLVM Object Reader" href="llvm-readobj.html" />
+ <link rel="prev" title="lit - LLVM Integrated Tester" href="lit.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-readobj.html" title="llvm-readobj - LLVM Object Reader"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="lit.html" title="lit - LLVM Integrated Tester"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-build-llvm-project-build-utility">
+<h1>llvm-build - LLVM Project Build Utility<a class="headerlink" href="#llvm-build-llvm-project-build-utility" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong>llvm-build</strong> [<em>options</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p><strong>llvm-build</strong> is a tool for working with LLVM projects that use the LLVMBuild
+system for describing their components.</p>
+<p>At heart, <strong>llvm-build</strong> is responsible for loading, verifying, and manipulating
+the project’s component data. The tool is primarily designed for use in
+implementing build systems and tools which need access to the project structure
+information.</p>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<p><strong>-h</strong>, <strong>–help</strong></p>
+<blockquote>
+<div>Print the builtin program help.</div></blockquote>
+<p><strong>–source-root</strong>=<em>PATH</em></p>
+<blockquote>
+<div>If given, load the project at the given source root path. If this option is not
+given, the location of the project sources will be inferred from the location of
+the <strong>llvm-build</strong> script itself.</div></blockquote>
+<p><strong>–print-tree</strong></p>
+<blockquote>
+<div>Print the component tree for the project.</div></blockquote>
+<p><strong>–write-library-table</strong></p>
+<blockquote>
+<div>Write out the C++ fragment which defines the components, library names, and
+required libraries. This C++ fragment is built into llvm-config|llvm-config
+in order to provide clients with the list of required libraries for arbitrary
+component combinations.</div></blockquote>
+<p><strong>–write-llvmbuild</strong></p>
+<blockquote>
+<div>Write out new <em>LLVMBuild.txt</em> files based on the loaded components. This is
+useful for auto-upgrading the schema of the files. <strong>llvm-build</strong> will try to a
+limited extent to preserve the comments which were written in the original
+source file, although at this time it only preserves block comments that precede
+the section names in the <em>LLVMBuild</em> files.</div></blockquote>
+<p><strong>–write-cmake-fragment</strong></p>
+<blockquote>
+<div>Write out the LLVMBuild in the form of a CMake fragment, so it can easily be
+consumed by the CMake based build system. The exact contents and format of this
+file are closely tied to how LLVMBuild is integrated with CMake, see LLVM’s
+top-level CMakeLists.txt.</div></blockquote>
+<p><strong>–write-make-fragment</strong></p>
+<blockquote>
+<div>Write out the LLVMBuild in the form of a Makefile fragment, so it can easily be
+consumed by a Make based build system. The exact contents and format of this
+file are closely tied to how LLVMBuild is integrated with the Makefiles, see
+LLVM’s Makefile.rules.</div></blockquote>
+<p><strong>–llvmbuild-source-root</strong>=<em>PATH</em></p>
+<blockquote>
+<div>If given, expect the <em>LLVMBuild</em> files for the project to be rooted at the
+given path, instead of inside the source tree itself. This option is primarily
+designed for use in conjunction with <strong>–write-llvmbuild</strong> to test changes to
+<em>LLVMBuild</em> schema.</div></blockquote>
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p><strong>llvm-build</strong> exits with 0 if operation was successful. Otherwise, it will exist
+with a non-zero value.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-readobj.html" title="llvm-readobj - LLVM Object Reader"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="lit.html" title="lit - LLVM Integrated Tester"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-config.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-config.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-config.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-config.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,196 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-config - Print LLVM compilation options — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-diff - LLVM structural âdiffâ" href="llvm-diff.html" />
+ <link rel="prev" title="llvm-nm - list LLVM bitcode and object fileâs symbol table" href="llvm-nm.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-diff.html" title="llvm-diff - LLVM structural âdiffâ"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-nm.html" title="llvm-nm - list LLVM bitcode and object fileâs symbol table"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-config-print-llvm-compilation-options">
+<h1>llvm-config - Print LLVM compilation options<a class="headerlink" href="#llvm-config-print-llvm-compilation-options" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong>llvm-config</strong> <em>option</em> [<em>components</em>...]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p><strong>llvm-config</strong> makes it easier to build applications that use LLVM. It can
+print the compiler flags, linker flags and object libraries needed to link
+against LLVM.</p>
+</div>
+<div class="section" id="examples">
+<h2>EXAMPLES<a class="headerlink" href="#examples" title="Permalink to this headline">¶</a></h2>
+<p>To link against the JIT:</p>
+<div class="highlight-sh"><div class="highlight"><pre>g++ <span class="sb">`</span>llvm-config --cxxflags<span class="sb">`</span> -o HowToUseJIT.o -c HowToUseJIT.cpp
+g++ <span class="sb">`</span>llvm-config --ldflags<span class="sb">`</span> -o HowToUseJIT HowToUseJIT.o <span class="se">\</span>
+ <span class="sb">`</span>llvm-config --libs engine bcreader scalaropts<span class="sb">`</span>
+</pre></div>
+</div>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<p><strong>–version</strong></p>
+<blockquote>
+<div>Print the version number of LLVM.</div></blockquote>
+<p><strong>-help</strong></p>
+<blockquote>
+<div>Print a summary of <strong>llvm-config</strong> arguments.</div></blockquote>
+<p><strong>–prefix</strong></p>
+<blockquote>
+<div>Print the installation prefix for LLVM.</div></blockquote>
+<p><strong>–src-root</strong></p>
+<blockquote>
+<div>Print the source root from which LLVM was built.</div></blockquote>
+<p><strong>–obj-root</strong></p>
+<blockquote>
+<div>Print the object root used to build LLVM.</div></blockquote>
+<p><strong>–bindir</strong></p>
+<blockquote>
+<div>Print the installation directory for LLVM binaries.</div></blockquote>
+<p><strong>–includedir</strong></p>
+<blockquote>
+<div>Print the installation directory for LLVM headers.</div></blockquote>
+<p><strong>–libdir</strong></p>
+<blockquote>
+<div>Print the installation directory for LLVM libraries.</div></blockquote>
+<p><strong>–cxxflags</strong></p>
+<blockquote>
+<div>Print the C++ compiler flags needed to use LLVM headers.</div></blockquote>
+<p><strong>–ldflags</strong></p>
+<blockquote>
+<div>Print the flags needed to link against LLVM libraries.</div></blockquote>
+<p><strong>–libs</strong></p>
+<blockquote>
+<div>Print all the libraries needed to link against the specified LLVM
+<em>components</em>, including any dependencies.</div></blockquote>
+<p><strong>–libnames</strong></p>
+<blockquote>
+<div>Similar to <strong>–libs</strong>, but prints the bare filenames of the libraries
+without <strong>-l</strong> or pathnames. Useful for linking against a not-yet-installed
+copy of LLVM.</div></blockquote>
+<p><strong>–libfiles</strong></p>
+<blockquote>
+<div>Similar to <strong>–libs</strong>, but print the full path to each library file. This is
+useful when creating makefile dependencies, to ensure that a tool is relinked if
+any library it uses changes.</div></blockquote>
+<p><strong>–components</strong></p>
+<blockquote>
+<div>Print all valid component names.</div></blockquote>
+<p><strong>–targets-built</strong></p>
+<blockquote>
+<div>Print the component names for all targets supported by this copy of LLVM.</div></blockquote>
+<p><strong>–build-mode</strong></p>
+<blockquote>
+<div>Print the build mode used when LLVM was built (e.g. Debug or Release)</div></blockquote>
+</div>
+<div class="section" id="components">
+<h2>COMPONENTS<a class="headerlink" href="#components" title="Permalink to this headline">¶</a></h2>
+<p>To print a list of all available components, run <strong>llvm-config
+–components</strong>. In most cases, components correspond directly to LLVM
+libraries. Useful “virtual” components include:</p>
+<p><strong>all</strong></p>
+<blockquote>
+<div>Includes all LLVM libraries. The default if no components are specified.</div></blockquote>
+<p><strong>backend</strong></p>
+<blockquote>
+<div>Includes either a native backend or the C backend.</div></blockquote>
+<p><strong>engine</strong></p>
+<blockquote>
+<div>Includes either a native JIT or the bitcode interpreter.</div></blockquote>
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p>If <strong>llvm-config</strong> succeeds, it will exit with 0. Otherwise, if an error
+occurs, it will exit with a non-zero value.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-diff.html" title="llvm-diff - LLVM structural âdiffâ"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-nm.html" title="llvm-nm - list LLVM bitcode and object fileâs symbol table"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-cov.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-cov.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-cov.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-cov.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,478 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-cov - emit coverage information — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-profdata - Profile data tool" href="llvm-profdata.html" />
+ <link rel="prev" title="llvm-diff - LLVM structural âdiffâ" href="llvm-diff.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-profdata.html" title="llvm-profdata - Profile data tool"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-diff.html" title="llvm-diff - LLVM structural âdiffâ"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-cov-emit-coverage-information">
+<h1>llvm-cov - emit coverage information<a class="headerlink" href="#llvm-cov-emit-coverage-information" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-cov</strong> <em>command</em> [<em>args...</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p>The <strong class="program">llvm-cov</strong> tool shows code coverage information for
+programs that are instrumented to emit profile data. It can be used to
+work with <tt class="docutils literal"><span class="pre">gcov</span></tt>-style coverage or with <tt class="docutils literal"><span class="pre">clang</span></tt>‘s instrumentation
+based profiling.</p>
+<p>If the program is invoked with a base name of <tt class="docutils literal"><span class="pre">gcov</span></tt>, it will behave as if
+the <strong class="program">llvm-cov gcov</strong> command were called. Otherwise, a command should
+be provided.</p>
+</div>
+<div class="section" id="commands">
+<h2>COMMANDS<a class="headerlink" href="#commands" title="Permalink to this headline">¶</a></h2>
+<ul class="simple">
+<li><a class="reference internal" href="#llvm-cov-gcov"><em>gcov</em></a></li>
+<li><a class="reference internal" href="#llvm-cov-show"><em>show</em></a></li>
+<li><a class="reference internal" href="#llvm-cov-report"><em>report</em></a></li>
+<li><a class="reference internal" href="#llvm-cov-export"><em>export</em></a></li>
+</ul>
+</div>
+<div class="section" id="gcov-command">
+<span id="llvm-cov-gcov"></span><h2>GCOV COMMAND<a class="headerlink" href="#gcov-command" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="id1">
+<h3>SYNOPSIS<a class="headerlink" href="#id1" title="Permalink to this headline">¶</a></h3>
+<p><strong class="program">llvm-cov gcov</strong> [<em>options</em>] <em>SOURCEFILE</em></p>
+</div>
+<div class="section" id="id2">
+<h3>DESCRIPTION<a class="headerlink" href="#id2" title="Permalink to this headline">¶</a></h3>
+<p>The <strong class="program">llvm-cov gcov</strong> tool reads code coverage data files and displays
+the coverage information for a specified source file. It is compatible with the
+<tt class="docutils literal"><span class="pre">gcov</span></tt> tool from version 4.2 of <tt class="docutils literal"><span class="pre">GCC</span></tt> and may also be compatible with some
+later versions of <tt class="docutils literal"><span class="pre">gcov</span></tt>.</p>
+<p>To use <strong class="program">llvm-cov gcov</strong>, you must first build an instrumented version
+of your application that collects coverage data as it runs. Compile with the
+<tt class="docutils literal"><span class="pre">-fprofile-arcs</span></tt> and <tt class="docutils literal"><span class="pre">-ftest-coverage</span></tt> options to add the
+instrumentation. (Alternatively, you can use the <tt class="docutils literal"><span class="pre">--coverage</span></tt> option, which
+includes both of those other options.) You should compile with debugging
+information (<tt class="docutils literal"><span class="pre">-g</span></tt>) and without optimization (<tt class="docutils literal"><span class="pre">-O0</span></tt>); otherwise, the
+coverage data cannot be accurately mapped back to the source code.</p>
+<p>At the time you compile the instrumented code, a <tt class="docutils literal"><span class="pre">.gcno</span></tt> data file will be
+generated for each object file. These <tt class="docutils literal"><span class="pre">.gcno</span></tt> files contain half of the
+coverage data. The other half of the data comes from <tt class="docutils literal"><span class="pre">.gcda</span></tt> files that are
+generated when you run the instrumented program, with a separate <tt class="docutils literal"><span class="pre">.gcda</span></tt>
+file for each object file. Each time you run the program, the execution counts
+are summed into any existing <tt class="docutils literal"><span class="pre">.gcda</span></tt> files, so be sure to remove any old
+files if you do not want their contents to be included.</p>
+<p>By default, the <tt class="docutils literal"><span class="pre">.gcda</span></tt> files are written into the same directory as the
+object files, but you can override that by setting the <tt class="docutils literal"><span class="pre">GCOV_PREFIX</span></tt> and
+<tt class="docutils literal"><span class="pre">GCOV_PREFIX_STRIP</span></tt> environment variables. The <tt class="docutils literal"><span class="pre">GCOV_PREFIX_STRIP</span></tt>
+variable specifies a number of directory components to be removed from the
+start of the absolute path to the object file directory. After stripping those
+directories, the prefix from the <tt class="docutils literal"><span class="pre">GCOV_PREFIX</span></tt> variable is added. These
+environment variables allow you to run the instrumented program on a machine
+where the original object file directories are not accessible, but you will
+then need to copy the <tt class="docutils literal"><span class="pre">.gcda</span></tt> files back to the object file directories
+where <strong class="program">llvm-cov gcov</strong> expects to find them.</p>
+<p>Once you have generated the coverage data files, run <strong class="program">llvm-cov gcov</strong>
+for each main source file where you want to examine the coverage results. This
+should be run from the same directory where you previously ran the
+compiler. The results for the specified source file are written to a file named
+by appending a <tt class="docutils literal"><span class="pre">.gcov</span></tt> suffix. A separate output file is also created for
+each file included by the main source file, also with a <tt class="docutils literal"><span class="pre">.gcov</span></tt> suffix added.</p>
+<p>The basic content of an <tt class="docutils literal"><span class="pre">.gcov</span></tt> output file is a copy of the source file with
+an execution count and line number prepended to every line. The execution
+count is shown as <tt class="docutils literal"><span class="pre">-</span></tt> if a line does not contain any executable code. If
+a line contains code but that code was never executed, the count is displayed
+as <tt class="docutils literal"><span class="pre">#####</span></tt>.</p>
+</div>
+<div class="section" id="options">
+<h3>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h3>
+<dl class="option">
+<dt id="cmdoption-llvm-cov-gcov-a">
+<span id="cmdoption-llvm-cov-gcov--all-blocks"></span><tt class="descname">-a</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">--all-blocks</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-gcov-a" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display all basic blocks. If there are multiple blocks for a single line of
+source code, this option causes llvm-cov to show the count for each block
+instead of just one count for the entire line.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-gcov-b">
+<span id="cmdoption-llvm-cov-gcov--branch-probabilities"></span><tt class="descname">-b</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">--branch-probabilities</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-gcov-b" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display conditional branch probabilities and a summary of branch information.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-gcov-c">
+<span id="cmdoption-llvm-cov-gcov--branch-counts"></span><tt class="descname">-c</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">--branch-counts</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-gcov-c" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display branch counts instead of probabilities (requires -b).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-gcov-f">
+<span id="cmdoption-llvm-cov-gcov--function-summaries"></span><tt class="descname">-f</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">--function-summaries</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-gcov-f" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show a summary of coverage for each function instead of just one summary for
+an entire source file.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-gcov--help">
+<tt class="descname">--help</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-gcov--help" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display available options (–help-hidden for more).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-gcov-l">
+<span id="cmdoption-llvm-cov-gcov--long-file-names"></span><tt class="descname">-l</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">--long-file-names</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-gcov-l" title="Permalink to this definition">¶</a></dt>
+<dd><p>For coverage output of files included from the main source file, add the
+main file name followed by <tt class="docutils literal"><span class="pre">##</span></tt> as a prefix to the output file names. This
+can be combined with the –preserve-paths option to use complete paths for
+both the main file and the included file.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-gcov-n">
+<span id="cmdoption-llvm-cov-gcov--no-output"></span><tt class="descname">-n</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">--no-output</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-gcov-n" title="Permalink to this definition">¶</a></dt>
+<dd><p>Do not output any <tt class="docutils literal"><span class="pre">.gcov</span></tt> files. Summary information is still
+displayed.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-gcov-o">
+<span id="cmdoption-llvm-cov-gcov--object-directory"></span><span id="cmdoption-llvm-cov-gcov--object-file"></span><tt class="descname">-o</tt><tt class="descclassname">=<DIR|FILE></tt><tt class="descclassname">, </tt><tt class="descname">--object-directory</tt><tt class="descclassname">=<DIR></tt><tt class="descclassname">, </tt><tt class="descname">--object-file</tt><tt class="descclassname">=<FILE></tt><a class="headerlink" href="#cmdoption-llvm-cov-gcov-o" title="Permalink to this definition">¶</a></dt>
+<dd><p>Find objects in DIR or based on FILE’s path. If you specify a particular
+object file, the coverage data files are expected to have the same base name
+with <tt class="docutils literal"><span class="pre">.gcno</span></tt> and <tt class="docutils literal"><span class="pre">.gcda</span></tt> extensions. If you specify a directory, the
+files are expected in that directory with the same base name as the source
+file.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-gcov-p">
+<span id="cmdoption-llvm-cov-gcov--preserve-paths"></span><tt class="descname">-p</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">--preserve-paths</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-gcov-p" title="Permalink to this definition">¶</a></dt>
+<dd><p>Preserve path components when naming the coverage output files. In addition
+to the source file name, include the directories from the path to that
+file. The directories are separate by <tt class="docutils literal"><span class="pre">#</span></tt> characters, with <tt class="docutils literal"><span class="pre">.</span></tt> directories
+removed and <tt class="docutils literal"><span class="pre">..</span></tt> directories replaced by <tt class="docutils literal"><span class="pre">^</span></tt> characters. When used with
+the –long-file-names option, this applies to both the main file name and the
+included file name.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-gcov-u">
+<span id="cmdoption-llvm-cov-gcov--unconditional-branches"></span><tt class="descname">-u</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">--unconditional-branches</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-gcov-u" title="Permalink to this definition">¶</a></dt>
+<dd><p>Include unconditional branches in the output for the –branch-probabilities
+option.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-gcov-version">
+<tt class="descname">-version</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-gcov-version" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display the version of llvm-cov.</p>
+</dd></dl>
+
+</div>
+<div class="section" id="exit-status">
+<h3>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h3>
+<p><strong class="program">llvm-cov gcov</strong> returns 1 if it cannot read input files. Otherwise,
+it exits with zero.</p>
+</div>
+</div>
+<div class="section" id="show-command">
+<span id="llvm-cov-show"></span><h2>SHOW COMMAND<a class="headerlink" href="#show-command" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="id3">
+<h3>SYNOPSIS<a class="headerlink" href="#id3" title="Permalink to this headline">¶</a></h3>
+<p><strong class="program">llvm-cov show</strong> [<em>options</em>] -instr-profile <em>PROFILE</em> <em>BIN</em> [<em>-object BIN,...</em>] [[<em>-object BIN</em>]] [<em>SOURCES</em>]</p>
+</div>
+<div class="section" id="id4">
+<h3>DESCRIPTION<a class="headerlink" href="#id4" title="Permalink to this headline">¶</a></h3>
+<p>The <strong class="program">llvm-cov show</strong> command shows line by line coverage of the
+binaries <em>BIN</em>,... using the profile data <em>PROFILE</em>. It can optionally be
+filtered to only show the coverage for the files listed in <em>SOURCES</em>.</p>
+<p>To use <strong class="program">llvm-cov show</strong>, you need a program that is compiled with
+instrumentation to emit profile and coverage data. To build such a program with
+<tt class="docutils literal"><span class="pre">clang</span></tt> use the <tt class="docutils literal"><span class="pre">-fprofile-instr-generate</span></tt> and <tt class="docutils literal"><span class="pre">-fcoverage-mapping</span></tt>
+flags. If linking with the <tt class="docutils literal"><span class="pre">clang</span></tt> driver, pass <tt class="docutils literal"><span class="pre">-fprofile-instr-generate</span></tt>
+to the link stage to make sure the necessary runtime libraries are linked in.</p>
+<p>The coverage information is stored in the built executable or library itself,
+and this is what you should pass to <strong class="program">llvm-cov show</strong> as a <em>BIN</em>
+argument. The profile data is generated by running this instrumented program
+normally. When the program exits it will write out a raw profile file,
+typically called <tt class="docutils literal"><span class="pre">default.profraw</span></tt>, which can be converted to a format that
+is suitable for the <em>PROFILE</em> argument using the <strong class="program">llvm-profdata merge</strong>
+tool.</p>
+</div>
+<div class="section" id="id5">
+<h3>OPTIONS<a class="headerlink" href="#id5" title="Permalink to this headline">¶</a></h3>
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-show-line-counts">
+<tt class="descname">-show-line-counts</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-show-line-counts" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show the execution counts for each line. This is enabled by default, unless
+another <tt class="docutils literal"><span class="pre">-show</span></tt> option is used.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-show-expansions">
+<tt class="descname">-show-expansions</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-show-expansions" title="Permalink to this definition">¶</a></dt>
+<dd><p>Expand inclusions, such as preprocessor macros or textual inclusions, inline
+in the display of the source file.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-show-instantiations">
+<tt class="descname">-show-instantiations</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-show-instantiations" title="Permalink to this definition">¶</a></dt>
+<dd><p>For source regions that are instantiated multiple times, such as templates in
+<tt class="docutils literal"><span class="pre">C++</span></tt>, show each instantiation separately as well as the combined summary.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-show-regions">
+<tt class="descname">-show-regions</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-show-regions" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show the execution counts for each region by displaying a caret that points to
+the character where the region starts.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-show-line-counts-or-regions">
+<tt class="descname">-show-line-counts-or-regions</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-show-line-counts-or-regions" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show the execution counts for each line if there is only one region on the
+line, but show the individual regions if there are multiple on the line.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-use-color">
+<tt class="descname">-use-color</tt><tt class="descclassname">[=VALUE]</tt><a class="headerlink" href="#cmdoption-llvm-cov-show-use-color" title="Permalink to this definition">¶</a></dt>
+<dd><p>Enable or disable color output. By default this is autodetected.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-arch">
+<tt class="descname">-arch</tt><tt class="descclassname">=<name></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-arch" title="Permalink to this definition">¶</a></dt>
+<dd><p>If the covered binary is a universal binary, select the architecture to use.
+It is an error to specify an architecture that is not included in the
+universal binary or to use an architecture that does not match a
+non-universal binary.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-name">
+<tt class="descname">-name</tt><tt class="descclassname">=<NAME></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-name" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show code coverage only for functions with the given name.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-name-regex">
+<tt class="descname">-name-regex</tt><tt class="descclassname">=<PATTERN></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-name-regex" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show code coverage only for functions that match the given regular expression.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-format">
+<tt class="descname">-format</tt><tt class="descclassname">=<FORMAT></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-format" title="Permalink to this definition">¶</a></dt>
+<dd><p>Use the specified output format. The supported formats are: “text”, “html”.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-tab-size">
+<tt class="descname">-tab-size</tt><tt class="descclassname">=<TABSIZE></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-tab-size" title="Permalink to this definition">¶</a></dt>
+<dd><p>Replace tabs with <TABSIZE> spaces when preparing reports. Currently, this is
+only supported for the html format.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-output-dir">
+<tt class="descname">-output-dir</tt><tt class="descclassname">=PATH</tt><a class="headerlink" href="#cmdoption-llvm-cov-show-output-dir" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify a directory to write coverage reports into. If the directory does not
+exist, it is created. When used in function view mode (i.e when -name or
+-name-regex are used to select specific functions), the report is written to
+PATH/functions.EXTENSION. When used in file view mode, a report for each file
+is written to PATH/REL_PATH_TO_FILE.EXTENSION.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-Xdemangler">
+<tt class="descname">-Xdemangler</tt><tt class="descclassname">=<TOOL>|<TOOL-OPTION></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-Xdemangler" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify a symbol demangler. This can be used to make reports more
+human-readable. This option can be specified multiple times to supply
+arguments to the demangler (e.g <cite>-Xdemangler c++filt -Xdemangler -n</cite> for C++).
+The demangler is expected to read a newline-separated list of symbols from
+stdin and write a newline-separated list of the same length to stdout.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-line-coverage-gt">
+<tt class="descname">-line-coverage-gt</tt><tt class="descclassname">=<N></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-line-coverage-gt" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show code coverage only for functions with line coverage greater than the
+given threshold.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-line-coverage-lt">
+<tt class="descname">-line-coverage-lt</tt><tt class="descclassname">=<N></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-line-coverage-lt" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show code coverage only for functions with line coverage less than the given
+threshold.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-region-coverage-gt">
+<tt class="descname">-region-coverage-gt</tt><tt class="descclassname">=<N></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-region-coverage-gt" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show code coverage only for functions with region coverage greater than the
+given threshold.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-show-region-coverage-lt">
+<tt class="descname">-region-coverage-lt</tt><tt class="descclassname">=<N></tt><a class="headerlink" href="#cmdoption-llvm-cov-show-region-coverage-lt" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show code coverage only for functions with region coverage less than the given
+threshold.</p>
+</dd></dl>
+
+</div>
+</div>
+<div class="section" id="report-command">
+<span id="llvm-cov-report"></span><h2>REPORT COMMAND<a class="headerlink" href="#report-command" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="id6">
+<h3>SYNOPSIS<a class="headerlink" href="#id6" title="Permalink to this headline">¶</a></h3>
+<p><strong class="program">llvm-cov report</strong> [<em>options</em>] -instr-profile <em>PROFILE</em> <em>BIN</em> [<em>-object BIN,...</em>] [[<em>-object BIN</em>]] [<em>SOURCES</em>]</p>
+</div>
+<div class="section" id="id7">
+<h3>DESCRIPTION<a class="headerlink" href="#id7" title="Permalink to this headline">¶</a></h3>
+<p>The <strong class="program">llvm-cov report</strong> command displays a summary of the coverage of
+the binaries <em>BIN</em>,... using the profile data <em>PROFILE</em>. It can optionally be
+filtered to only show the coverage for the files listed in <em>SOURCES</em>.</p>
+<p>If no source files are provided, a summary line is printed for each file in the
+coverage data. If any files are provided, summaries are shown for each function
+in the listed files instead.</p>
+<p>For information on compiling programs for coverage and generating profile data,
+see <a class="reference internal" href="#llvm-cov-show"><em>SHOW COMMAND</em></a>.</p>
+</div>
+<div class="section" id="id8">
+<h3>OPTIONS<a class="headerlink" href="#id8" title="Permalink to this headline">¶</a></h3>
+<dl class="option">
+<dt id="cmdoption-llvm-cov-report-use-color">
+<tt class="descname">-use-color</tt><tt class="descclassname">[=VALUE]</tt><a class="headerlink" href="#cmdoption-llvm-cov-report-use-color" title="Permalink to this definition">¶</a></dt>
+<dd><p>Enable or disable color output. By default this is autodetected.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-cov-report-arch">
+<tt class="descname">-arch</tt><tt class="descclassname">=<name></tt><a class="headerlink" href="#cmdoption-llvm-cov-report-arch" title="Permalink to this definition">¶</a></dt>
+<dd><p>If the covered binary is a universal binary, select the architecture to use.
+It is an error to specify an architecture that is not included in the
+universal binary or to use an architecture that does not match a
+non-universal binary.</p>
+</dd></dl>
+
+</div>
+</div>
+<div class="section" id="export-command">
+<span id="llvm-cov-export"></span><h2>EXPORT COMMAND<a class="headerlink" href="#export-command" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="id9">
+<h3>SYNOPSIS<a class="headerlink" href="#id9" title="Permalink to this headline">¶</a></h3>
+<p><strong class="program">llvm-cov export</strong> [<em>options</em>] -instr-profile <em>PROFILE</em> <em>BIN</em> [<em>-object BIN,...</em>] [[<em>-object BIN</em>]]</p>
+</div>
+<div class="section" id="id10">
+<h3>DESCRIPTION<a class="headerlink" href="#id10" title="Permalink to this headline">¶</a></h3>
+<p>The <strong class="program">llvm-cov export</strong> command exports regions, functions, expansions,
+and summaries of the coverage of the binaries <em>BIN</em>,... using the profile data
+<em>PROFILE</em> as JSON.</p>
+<p>For information on compiling programs for coverage and generating profile data,
+see <a class="reference internal" href="#llvm-cov-show"><em>SHOW COMMAND</em></a>.</p>
+</div>
+<div class="section" id="id11">
+<h3>OPTIONS<a class="headerlink" href="#id11" title="Permalink to this headline">¶</a></h3>
+<dl class="option">
+<dt id="cmdoption-llvm-cov-export-arch">
+<tt class="descname">-arch</tt><tt class="descclassname">=<name></tt><a class="headerlink" href="#cmdoption-llvm-cov-export-arch" title="Permalink to this definition">¶</a></dt>
+<dd><p>If the covered binary is a universal binary, select the architecture to use.
+It is an error to specify an architecture that is not included in the
+universal binary or to use an architecture that does not match a
+non-universal binary.</p>
+</dd></dl>
+
+</div>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-profdata.html" title="llvm-profdata - Profile data tool"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-diff.html" title="llvm-diff - LLVM structural âdiffâ"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-diff.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-diff.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-diff.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-diff.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,138 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-diff - LLVM structural âdiffâ — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-cov - emit coverage information" href="llvm-cov.html" />
+ <link rel="prev" title="llvm-config - Print LLVM compilation options" href="llvm-config.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-cov.html" title="llvm-cov - emit coverage information"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-config.html" title="llvm-config - Print LLVM compilation options"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-diff-llvm-structural-diff">
+<h1>llvm-diff - LLVM structural ‘diff’<a class="headerlink" href="#llvm-diff-llvm-structural-diff" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong>llvm-diff</strong> [<em>options</em>] <em>module 1</em> <em>module 2</em> [<em>global name ...</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p><strong>llvm-diff</strong> compares the structure of two LLVM modules, primarily
+focusing on differences in function definitions. Insignificant
+differences, such as changes in the ordering of globals or in the
+names of local values, are ignored.</p>
+<p>An input module will be interpreted as an assembly file if its name
+ends in ‘.ll’; otherwise it will be read in as a bitcode file.</p>
+<p>If a list of global names is given, just the values with those names
+are compared; otherwise, all global values are compared, and
+diagnostics are produced for globals which only appear in one module
+or the other.</p>
+<p><strong>llvm-diff</strong> compares two functions by comparing their basic blocks,
+beginning with the entry blocks. If the terminators seem to match,
+then the corresponding successors are compared; otherwise they are
+ignored. This algorithm is very sensitive to changes in control flow,
+which tend to stop any downstream changes from being detected.</p>
+<p><strong>llvm-diff</strong> is intended as a debugging tool for writers of LLVM
+passes and frontends. It does not have a stable output format.</p>
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p>If <strong>llvm-diff</strong> finds no differences between the modules, it will exit
+with 0 and produce no output. Otherwise it will exit with a non-zero
+value.</p>
+</div>
+<div class="section" id="bugs">
+<h2>BUGS<a class="headerlink" href="#bugs" title="Permalink to this headline">¶</a></h2>
+<p>Many important differences, like changes in linkage or function
+attributes, are not diagnosed.</p>
+<p>Changes in memory behavior (for example, coalescing loads) can cause
+massive detected differences in blocks.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-cov.html" title="llvm-cov - emit coverage information"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-config.html" title="llvm-config - Print LLVM compilation options"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-dis.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-dis.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-dis.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-dis.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,142 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-dis - LLVM disassembler — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="opt - LLVM optimizer" href="opt.html" />
+ <link rel="prev" title="llvm-as - LLVM assembler" href="llvm-as.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="opt.html" title="opt - LLVM optimizer"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-as.html" title="llvm-as - LLVM assembler"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-dis-llvm-disassembler">
+<h1>llvm-dis - LLVM disassembler<a class="headerlink" href="#llvm-dis-llvm-disassembler" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong>llvm-dis</strong> [<em>options</em>] [<em>filename</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p>The <strong>llvm-dis</strong> command is the LLVM disassembler. It takes an LLVM
+bitcode file and converts it into human-readable LLVM assembly language.</p>
+<p>If filename is omitted or specified as <tt class="docutils literal"><span class="pre">-</span></tt>, <strong>llvm-dis</strong> reads its
+input from standard input.</p>
+<p>If the input is being read from standard input, then <strong>llvm-dis</strong>
+will send its output to standard output by default. Otherwise, the
+output will be written to a file named after the input file, with
+a <tt class="docutils literal"><span class="pre">.ll</span></tt> suffix added (any existing <tt class="docutils literal"><span class="pre">.bc</span></tt> suffix will first be
+removed). You can override the choice of output file using the
+<strong>-o</strong> option.</p>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<p><strong>-f</strong></p>
+<blockquote>
+<div>Enable binary output on terminals. Normally, <strong>llvm-dis</strong> will refuse to
+write raw bitcode output if the output stream is a terminal. With this option,
+<strong>llvm-dis</strong> will write raw bitcode regardless of the output device.</div></blockquote>
+<p><strong>-help</strong></p>
+<blockquote>
+<div>Print a summary of command line options.</div></blockquote>
+<p><strong>-o</strong> <em>filename</em></p>
+<blockquote>
+<div>Specify the output file name. If <em>filename</em> is -, then the output is sent
+to standard output.</div></blockquote>
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p>If <strong>llvm-dis</strong> succeeds, it will exit with 0. Otherwise, if an error
+occurs, it will exit with a non-zero value.</p>
+</div>
+<div class="section" id="see-also">
+<h2>SEE ALSO<a class="headerlink" href="#see-also" title="Permalink to this headline">¶</a></h2>
+<p>llvm-as|llvm-as</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="opt.html" title="opt - LLVM optimizer"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-as.html" title="llvm-as - LLVM assembler"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-dwarfdump.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-dwarfdump.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-dwarfdump.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-dwarfdump.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,128 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-dwarfdump - print contents of DWARF sections — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="bugpoint - automatic test case reduction tool" href="bugpoint.html" />
+ <link rel="prev" title="llvm-symbolizer - convert addresses into source code locations" href="llvm-symbolizer.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="bugpoint.html" title="bugpoint - automatic test case reduction tool"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-symbolizer.html" title="llvm-symbolizer - convert addresses into source code locations"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-dwarfdump-print-contents-of-dwarf-sections">
+<h1>llvm-dwarfdump - print contents of DWARF sections<a class="headerlink" href="#llvm-dwarfdump-print-contents-of-dwarf-sections" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-dwarfdump</strong> [<em>options</em>] [<em>filenames...</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-dwarfdump</strong> parses DWARF sections in the object files
+and prints their contents in human-readable form.</p>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<dl class="option">
+<dt id="cmdoption-debug-dump">
+<tt class="descname">-debug-dump</tt><tt class="descclassname">=section</tt><a class="headerlink" href="#cmdoption-debug-dump" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the DWARF section to dump.
+For example, use <tt class="docutils literal"><span class="pre">abbrev</span></tt> to dump the contents of <tt class="docutils literal"><span class="pre">.debug_abbrev</span></tt> section,
+<tt class="docutils literal"><span class="pre">loc.dwo</span></tt> to dump the contents of <tt class="docutils literal"><span class="pre">.debug_loc.dwo</span></tt> etc.
+See <tt class="docutils literal"><span class="pre">llvm-dwarfdump</span> <span class="pre">--help</span></tt> for the complete list of supported sections.
+Use <tt class="docutils literal"><span class="pre">all</span></tt> to dump all DWARF sections. It is the default.</p>
+</dd></dl>
+
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-dwarfdump</strong> returns 0 if the input files were parsed and dumped
+successfully. Otherwise, it returns 1.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="bugpoint.html" title="bugpoint - automatic test case reduction tool"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-symbolizer.html" title="llvm-symbolizer - convert addresses into source code locations"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-extract.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-extract.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-extract.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-extract.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,163 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-extract - extract a function from an LLVM module — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-bcanalyzer - LLVM bitcode analyzer" href="llvm-bcanalyzer.html" />
+ <link rel="prev" title="bugpoint - automatic test case reduction tool" href="bugpoint.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-bcanalyzer.html" title="llvm-bcanalyzer - LLVM bitcode analyzer"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="bugpoint.html" title="bugpoint - automatic test case reduction tool"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-extract-extract-a-function-from-an-llvm-module">
+<h1>llvm-extract - extract a function from an LLVM module<a class="headerlink" href="#llvm-extract-extract-a-function-from-an-llvm-module" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-extract</strong> [<em>options</em>] <strong>–func</strong> <em>function-name</em> [<em>filename</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p>The <strong class="program">llvm-extract</strong> command takes the name of a function and extracts
+it from the specified LLVM bitcode file. It is primarily used as a debugging
+tool to reduce test cases from larger programs that are triggering a bug.</p>
+<p>In addition to extracting the bitcode of the specified function,
+<strong class="program">llvm-extract</strong> will also remove unreachable global variables,
+prototypes, and unused types.</p>
+<p>The <strong class="program">llvm-extract</strong> command reads its input from standard input if
+filename is omitted or if filename is <tt class="docutils literal"><span class="pre">-</span></tt>. The output is always written to
+standard output, unless the <strong>-o</strong> option is specified (see below).</p>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<p><strong>-f</strong></p>
+<blockquote>
+<div>Enable binary output on terminals. Normally, <strong class="program">llvm-extract</strong> will
+refuse to write raw bitcode output if the output stream is a terminal. With
+this option, <strong class="program">llvm-extract</strong> will write raw bitcode regardless of the
+output device.</div></blockquote>
+<p><strong>–func</strong> <em>function-name</em></p>
+<blockquote>
+<div>Extract the function named <em>function-name</em> from the LLVM bitcode. May be
+specified multiple times to extract multiple functions at once.</div></blockquote>
+<p><strong>–rfunc</strong> <em>function-regular-expr</em></p>
+<blockquote>
+<div>Extract the function(s) matching <em>function-regular-expr</em> from the LLVM bitcode.
+All functions matching the regular expression will be extracted. May be
+specified multiple times.</div></blockquote>
+<p><strong>–glob</strong> <em>global-name</em></p>
+<blockquote>
+<div>Extract the global variable named <em>global-name</em> from the LLVM bitcode. May be
+specified multiple times to extract multiple global variables at once.</div></blockquote>
+<p><strong>–rglob</strong> <em>glob-regular-expr</em></p>
+<blockquote>
+<div>Extract the global variable(s) matching <em>global-regular-expr</em> from the LLVM
+bitcode. All global variables matching the regular expression will be
+extracted. May be specified multiple times.</div></blockquote>
+<p><strong>-help</strong></p>
+<blockquote>
+<div>Print a summary of command line options.</div></blockquote>
+<p><strong>-o</strong> <em>filename</em></p>
+<blockquote>
+<div>Specify the output filename. If filename is “-” (the default), then
+<strong class="program">llvm-extract</strong> sends its output to standard output.</div></blockquote>
+<p><strong>-S</strong></p>
+<blockquote>
+<div>Write output in LLVM intermediate language (instead of bitcode).</div></blockquote>
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p>If <strong class="program">llvm-extract</strong> succeeds, it will exit with 0. Otherwise, if an error
+occurs, it will exit with a non-zero value.</p>
+</div>
+<div class="section" id="see-also">
+<h2>SEE ALSO<a class="headerlink" href="#see-also" title="Permalink to this headline">¶</a></h2>
+<p>bugpoint</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-bcanalyzer.html" title="llvm-bcanalyzer - LLVM bitcode analyzer"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="bugpoint.html" title="bugpoint - automatic test case reduction tool"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-lib.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-lib.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-lib.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-lib.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,123 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-lib - LLVM lib.exe compatible library tool — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-nm - list LLVM bitcode and object fileâs symbol table" href="llvm-nm.html" />
+ <link rel="prev" title="llvm-ar - LLVM archiver" href="llvm-ar.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-nm.html" title="llvm-nm - list LLVM bitcode and object fileâs symbol table"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-ar.html" title="llvm-ar - LLVM archiver"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-lib-llvm-lib-exe-compatible-library-tool">
+<h1>llvm-lib - LLVM lib.exe compatible library tool<a class="headerlink" href="#llvm-lib-llvm-lib-exe-compatible-library-tool" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong>llvm-lib</strong> [/libpath:<path>] [/out:<output>] [/llvmlibthin]
+[/ignore] [/machine] [/nologo] [files...]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p>The <strong>llvm-lib</strong> command is intended to be a <tt class="docutils literal"><span class="pre">lib.exe</span></tt> compatible
+tool. See <a class="reference external" href="https://msdn.microsoft.com/en-us/library/7ykb2k5f">https://msdn.microsoft.com/en-us/library/7ykb2k5f</a> for the
+general description.</p>
+<p><strong>llvm-lib</strong> has the following extensions:</p>
+<ul class="simple">
+<li>Bitcode files in symbol tables.
+<strong>llvm-lib</strong> includes symbols from both bitcode files and regular
+object files in the symbol table.</li>
+<li>Creating thin archives.
+The /llvmlibthin option causes <strong>llvm-lib</strong> to create thin archive
+that contain only the symbol table and the header for the various
+members. These files are much smaller, but are not compatible with
+link.exe (lld can handle them).</li>
+</ul>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-nm.html" title="llvm-nm - list LLVM bitcode and object fileâs symbol table"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-ar.html" title="llvm-ar - LLVM archiver"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-link.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-link.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-link.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-link.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,162 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-link - LLVM bitcode linker — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-ar - LLVM archiver" href="llvm-ar.html" />
+ <link rel="prev" title="lli - directly execute programs from LLVM bitcode" href="lli.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-ar.html" title="llvm-ar - LLVM archiver"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="lli.html" title="lli - directly execute programs from LLVM bitcode"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-link-llvm-bitcode-linker">
+<h1>llvm-link - LLVM bitcode linker<a class="headerlink" href="#llvm-link-llvm-bitcode-linker" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-link</strong> [<em>options</em>] <em>filename ...</em></p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-link</strong> takes several LLVM bitcode files and links them together
+into a single LLVM bitcode file. It writes the output file to standard output,
+unless the <a class="reference internal" href="opt.html#cmdoption-o"><em class="xref std std-option">-o</em></a> option is used to specify a filename.</p>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<dl class="option">
+<dt id="cmdoption-f">
+<tt class="descname">-f</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-f" title="Permalink to this definition">¶</a></dt>
+<dd><p>Enable binary output on terminals. Normally, <strong class="program">llvm-link</strong> will refuse
+to write raw bitcode output if the output stream is a terminal. With this
+option, <strong class="program">llvm-link</strong> will write raw bitcode regardless of the output
+device.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-o">
+<tt class="descname">-o</tt><tt class="descclassname"> filename</tt><a class="headerlink" href="#cmdoption-o" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the output file name. If <tt class="docutils literal"><span class="pre">filename</span></tt> is “<tt class="docutils literal"><span class="pre">-</span></tt>”, then
+<strong class="program">llvm-link</strong> will write its output to standard output.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-S">
+<tt class="descname">-S</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-S" title="Permalink to this definition">¶</a></dt>
+<dd><p>Write output in LLVM intermediate language (instead of bitcode).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-d">
+<tt class="descname">-d</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-d" title="Permalink to this definition">¶</a></dt>
+<dd><p>If specified, <strong class="program">llvm-link</strong> prints a human-readable version of the
+output bitcode file to standard error.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-help">
+<tt class="descname">-help</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-help" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print a summary of command line options.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-v">
+<tt class="descname">-v</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-v" title="Permalink to this definition">¶</a></dt>
+<dd><p>Verbose mode. Print information about what <strong class="program">llvm-link</strong> is doing.
+This typically includes a message for each bitcode file linked in and for each
+library found.</p>
+</dd></dl>
+
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p>If <strong class="program">llvm-link</strong> succeeds, it will exit with 0. Otherwise, if an error
+occurs, it will exit with a non-zero value.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-ar.html" title="llvm-ar - LLVM archiver"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="lli.html" title="lli - directly execute programs from LLVM bitcode"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-nm.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-nm.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-nm.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-nm.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,265 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-nm - list LLVM bitcode and object fileâs symbol table — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-config - Print LLVM compilation options" href="llvm-config.html" />
+ <link rel="prev" title="llvm-lib - LLVM lib.exe compatible library tool" href="llvm-lib.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-config.html" title="llvm-config - Print LLVM compilation options"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-lib.html" title="llvm-lib - LLVM lib.exe compatible library tool"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-nm-list-llvm-bitcode-and-object-file-s-symbol-table">
+<h1>llvm-nm - list LLVM bitcode and object file’s symbol table<a class="headerlink" href="#llvm-nm-list-llvm-bitcode-and-object-file-s-symbol-table" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-nm</strong> [<em>options</em>] [<em>filenames...</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p>The <strong class="program">llvm-nm</strong> utility lists the names of symbols from the LLVM bitcode
+files, object files, or <strong class="program">ar</strong> archives containing them, named on the
+command line. Each symbol is listed along with some simple information about
+its provenance. If no file name is specified, or <em>-</em> is used as a file name,
+<strong class="program">llvm-nm</strong> will process a file on its standard input stream.</p>
+<p><strong class="program">llvm-nm</strong>‘s default output format is the traditional BSD <strong class="program">nm</strong>
+output format. Each such output record consists of an (optional) 8-digit
+hexadecimal address, followed by a type code character, followed by a name, for
+each symbol. One record is printed per line; fields are separated by spaces.
+When the address is omitted, it is replaced by 8 spaces.</p>
+<p>Type code characters currently supported, and their meanings, are as follows:</p>
+<p>U</p>
+<blockquote>
+<div>Named object is referenced but undefined in this bitcode file</div></blockquote>
+<p>C</p>
+<blockquote>
+<div>Common (multiple definitions link together into one def)</div></blockquote>
+<p>W</p>
+<blockquote>
+<div>Weak reference (multiple definitions link together into zero or one definitions)</div></blockquote>
+<p>t</p>
+<blockquote>
+<div>Local function (text) object</div></blockquote>
+<p>T</p>
+<blockquote>
+<div>Global function (text) object</div></blockquote>
+<p>d</p>
+<blockquote>
+<div>Local data object</div></blockquote>
+<p>D</p>
+<blockquote>
+<div>Global data object</div></blockquote>
+<p>?</p>
+<blockquote>
+<div>Something unrecognizable</div></blockquote>
+<p>Because LLVM bitcode files typically contain objects that are not considered to
+have addresses until they are linked into an executable image or dynamically
+compiled “just-in-time”, <strong class="program">llvm-nm</strong> does not print an address for any
+symbol in an LLVM bitcode file, even symbols which are defined in the bitcode
+file.</p>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<dl class="option">
+<dt id="cmdoption-llvm-nm-B">
+<tt class="descname">-B</tt><tt class="descclassname"> (default)</tt><a class="headerlink" href="#cmdoption-llvm-nm-B" title="Permalink to this definition">¶</a></dt>
+<dd><p>Use BSD output format. Alias for <cite>–format=bsd</cite>.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm-P">
+<tt class="descname">-P</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm-P" title="Permalink to this definition">¶</a></dt>
+<dd><p>Use POSIX.2 output format. Alias for <cite>–format=posix</cite>.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm--debug-syms">
+<span id="cmdoption-llvm-nm-a"></span><tt class="descname">--debug-syms</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-a</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm--debug-syms" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show all symbols, even debugger only.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm--defined-only">
+<tt class="descname">--defined-only</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm--defined-only" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print only symbols defined in this file (as opposed to
+symbols which may be referenced by objects in this file, but not
+defined in this file.)</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm--dynamic">
+<span id="cmdoption-llvm-nm-D"></span><tt class="descname">--dynamic</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-D</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm--dynamic" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display dynamic symbols instead of normal symbols.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm--extern-only">
+<span id="cmdoption-llvm-nm-g"></span><tt class="descname">--extern-only</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-g</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm--extern-only" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print only symbols whose definitions are external; that is, accessible
+from other files.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm--format">
+<span id="cmdoption-llvm-nm-f"></span><tt class="descname">--format</tt><tt class="descclassname">=format</tt><tt class="descclassname">, </tt><tt class="descname">-f</tt><tt class="descclassname"> format</tt><a class="headerlink" href="#cmdoption-llvm-nm--format" title="Permalink to this definition">¶</a></dt>
+<dd><p>Select an output format; <em>format</em> may be <em>sysv</em>, <em>posix</em>, or <em>bsd</em>. The default
+is <em>bsd</em>.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm-help">
+<tt class="descname">-help</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm-help" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print a summary of command-line options and their meanings.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm--no-sort">
+<span id="cmdoption-llvm-nm-p"></span><tt class="descname">--no-sort</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-p</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm--no-sort" title="Permalink to this definition">¶</a></dt>
+<dd><p>Shows symbols in order encountered.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm--numeric-sort">
+<span id="cmdoption-llvm-nm-n"></span><span id="cmdoption-llvm-nm-v"></span><tt class="descname">--numeric-sort</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-n</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-v</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm--numeric-sort" title="Permalink to this definition">¶</a></dt>
+<dd><p>Sort symbols by address.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm--print-file-name">
+<span id="cmdoption-llvm-nm-A"></span><span id="cmdoption-llvm-nm-o"></span><tt class="descname">--print-file-name</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-A</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-o</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm--print-file-name" title="Permalink to this definition">¶</a></dt>
+<dd><p>Precede each symbol with the file it came from.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm--print-size">
+<span id="cmdoption-llvm-nm-S"></span><tt class="descname">--print-size</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-S</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm--print-size" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show symbol size instead of address.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm--size-sort">
+<tt class="descname">--size-sort</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm--size-sort" title="Permalink to this definition">¶</a></dt>
+<dd><p>Sort symbols by size.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm--undefined-only">
+<span id="cmdoption-llvm-nm-u"></span><tt class="descname">--undefined-only</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-u</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm--undefined-only" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print only symbols referenced but not defined in this file.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-nm--radix">
+<span id="cmdoption-llvm-nm-t"></span><tt class="descname">--radix</tt><tt class="descclassname">=RADIX</tt><tt class="descclassname">, </tt><tt class="descname">-t</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-nm--radix" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the radix of the symbol address(es). Values accepted d(decimal),
+x(hexadecomal) and o(octal).</p>
+</dd></dl>
+
+</div>
+<div class="section" id="bugs">
+<h2>BUGS<a class="headerlink" href="#bugs" title="Permalink to this headline">¶</a></h2>
+<blockquote>
+<div><ul class="simple">
+<li><strong class="program">llvm-nm</strong> cannot demangle C++ mangled names, like GNU <strong class="program">nm</strong>
+can.</li>
+<li><strong class="program">llvm-nm</strong> does not support the full set of arguments that GNU
+<strong class="program">nm</strong> does.</li>
+</ul>
+</div></blockquote>
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-nm</strong> exits with an exit code of zero.</p>
+</div>
+<div class="section" id="see-also">
+<h2>SEE ALSO<a class="headerlink" href="#see-also" title="Permalink to this headline">¶</a></h2>
+<p>llvm-dis, ar(1), nm(1)</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-config.html" title="llvm-config - Print LLVM compilation options"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-lib.html" title="llvm-lib - LLVM lib.exe compatible library tool"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-profdata.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-profdata.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-profdata.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-profdata.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,318 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-profdata - Profile data tool — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-stress - generate random .ll files" href="llvm-stress.html" />
+ <link rel="prev" title="llvm-cov - emit coverage information" href="llvm-cov.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-stress.html" title="llvm-stress - generate random .ll files"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-cov.html" title="llvm-cov - emit coverage information"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-profdata-profile-data-tool">
+<h1>llvm-profdata - Profile data tool<a class="headerlink" href="#llvm-profdata-profile-data-tool" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-profdata</strong> <em>command</em> [<em>args...</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p>The <strong class="program">llvm-profdata</strong> tool is a small utility for working with profile
+data files.</p>
+</div>
+<div class="section" id="commands">
+<h2>COMMANDS<a class="headerlink" href="#commands" title="Permalink to this headline">¶</a></h2>
+<ul class="simple">
+<li><a class="reference internal" href="#profdata-merge"><em>merge</em></a></li>
+<li><a class="reference internal" href="#profdata-show"><em>show</em></a></li>
+</ul>
+</div>
+<div class="section" id="merge">
+<span id="profdata-merge"></span><h2>MERGE<a class="headerlink" href="#merge" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="id1">
+<h3>SYNOPSIS<a class="headerlink" href="#id1" title="Permalink to this headline">¶</a></h3>
+<p><strong class="program">llvm-profdata merge</strong> [<em>options</em>] [<em>filename...</em>]</p>
+</div>
+<div class="section" id="id2">
+<h3>DESCRIPTION<a class="headerlink" href="#id2" title="Permalink to this headline">¶</a></h3>
+<p><strong class="program">llvm-profdata merge</strong> takes several profile data files
+generated by PGO instrumentation and merges them together into a single
+indexed profile data file.</p>
+<p>By default profile data is merged without modification. This means that the
+relative importance of each input file is proportional to the number of samples
+or counts it contains. In general, the input from a longer training run will be
+interpreted as relatively more important than a shorter run. Depending on the
+nature of the training runs it may be useful to adjust the weight given to each
+input file by using the <tt class="docutils literal"><span class="pre">-weighted-input</span></tt> option.</p>
+<p>Profiles passed in via <tt class="docutils literal"><span class="pre">-weighted-input</span></tt>, <tt class="docutils literal"><span class="pre">-input-files</span></tt>, or via positional
+arguments are processed once for each time they are seen.</p>
+</div>
+<div class="section" id="options">
+<h3>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h3>
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-merge-help">
+<tt class="descname">-help</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-profdata-merge-help" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print a summary of command line options.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-merge-output">
+<span id="cmdoption-llvm-profdata-merge-o"></span><tt class="descname">-output</tt><tt class="descclassname">=output</tt><tt class="descclassname">, </tt><tt class="descname">-o</tt><tt class="descclassname">=output</tt><a class="headerlink" href="#cmdoption-llvm-profdata-merge-output" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the output file name. <em>Output</em> cannot be <tt class="docutils literal"><span class="pre">-</span></tt> as the resulting
+indexed profile data can’t be written to standard output.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-merge-weighted-input">
+<tt class="descname">-weighted-input</tt><tt class="descclassname">=weight,filename</tt><a class="headerlink" href="#cmdoption-llvm-profdata-merge-weighted-input" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify an input file name along with a weight. The profile counts of the
+supplied <tt class="docutils literal"><span class="pre">filename</span></tt> will be scaled (multiplied) by the supplied
+<tt class="docutils literal"><span class="pre">weight</span></tt>, where where <tt class="docutils literal"><span class="pre">weight</span></tt> is a decimal integer >= 1.
+Input files specified without using this option are assigned a default
+weight of 1. Examples are shown below.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-merge-input-files">
+<span id="cmdoption-llvm-profdata-merge-f"></span><tt class="descname">-input-files</tt><tt class="descclassname">=path</tt><tt class="descclassname">, </tt><tt class="descname">-f</tt><tt class="descclassname">=path</tt><a class="headerlink" href="#cmdoption-llvm-profdata-merge-input-files" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify a file which contains a list of files to merge. The entries in this
+file are newline-separated. Lines starting with ‘#’ are skipped. Entries may
+be of the form <filename> or <weight>,<filename>.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-merge-instr">
+<tt class="descname">-instr</tt><tt class="descclassname"> (default)</tt><a class="headerlink" href="#cmdoption-llvm-profdata-merge-instr" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify that the input profile is an instrumentation-based profile.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-merge-sample">
+<tt class="descname">-sample</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-profdata-merge-sample" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify that the input profile is a sample-based profile.</p>
+<p>The format of the generated file can be generated in one of three ways:</p>
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-merge-binary">
+<tt class="descname">-binary</tt><tt class="descclassname"> (default)</tt><a class="headerlink" href="#cmdoption-llvm-profdata-merge-binary" title="Permalink to this definition">¶</a></dt>
+<dd></dd></dl>
+
+<p>Emit the profile using a binary encoding. For instrumentation-based profile
+the output format is the indexed binary format.</p>
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-merge-text">
+<tt class="descname">-text</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-profdata-merge-text" title="Permalink to this definition">¶</a></dt>
+<dd></dd></dl>
+
+<p>Emit the profile in text mode. This option can also be used with both
+sample-based and instrumentation-based profile. When this option is used
+the profile will be dumped in the text format that is parsable by the profile
+reader.</p>
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-merge-gcc">
+<tt class="descname">-gcc</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-profdata-merge-gcc" title="Permalink to this definition">¶</a></dt>
+<dd></dd></dl>
+
+<p>Emit the profile using GCC’s gcov format (Not yet supported).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-merge-sparse">
+<tt class="descname">-sparse</tt><tt class="descclassname">[=true|false]</tt><a class="headerlink" href="#cmdoption-llvm-profdata-merge-sparse" title="Permalink to this definition">¶</a></dt>
+<dd><p>Do not emit function records with 0 execution count. Can only be used in
+conjunction with -instr. Defaults to false, since it can inhibit compiler
+optimization during PGO.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-merge-num-threads">
+<span id="cmdoption-llvm-profdata-merge-j"></span><tt class="descname">-num-threads</tt><tt class="descclassname">=N</tt><tt class="descclassname">, </tt><tt class="descname">-j</tt><tt class="descclassname">=N</tt><a class="headerlink" href="#cmdoption-llvm-profdata-merge-num-threads" title="Permalink to this definition">¶</a></dt>
+<dd><p>Use N threads to perform profile merging. When N=0, llvm-profdata auto-detects
+an appropriate number of threads to use. This is the default.</p>
+</dd></dl>
+
+</div>
+<div class="section" id="examples">
+<h3>EXAMPLES<a class="headerlink" href="#examples" title="Permalink to this headline">¶</a></h3>
+<div class="section" id="basic-usage">
+<h4>Basic Usage<a class="headerlink" href="#basic-usage" title="Permalink to this headline">¶</a></h4>
+<p>Merge three profiles:</p>
+<div class="highlight-python"><div class="highlight"><pre>llvm-profdata merge foo.profdata bar.profdata baz.profdata -output merged.profdata
+</pre></div>
+</div>
+</div>
+<div class="section" id="weighted-input">
+<h4>Weighted Input<a class="headerlink" href="#weighted-input" title="Permalink to this headline">¶</a></h4>
+<p>The input file <cite>foo.profdata</cite> is especially important, multiply its counts by 10:</p>
+<div class="highlight-python"><div class="highlight"><pre>llvm-profdata merge -weighted-input=10,foo.profdata bar.profdata baz.profdata -output merged.profdata
+</pre></div>
+</div>
+<p>Exactly equivalent to the previous invocation (explicit form; useful for programmatic invocation):</p>
+<div class="highlight-python"><div class="highlight"><pre>llvm-profdata merge -weighted-input=10,foo.profdata -weighted-input=1,bar.profdata -weighted-input=1,baz.profdata -output merged.profdata
+</pre></div>
+</div>
+</div>
+</div>
+</div>
+<div class="section" id="show">
+<span id="profdata-show"></span><h2>SHOW<a class="headerlink" href="#show" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="id3">
+<h3>SYNOPSIS<a class="headerlink" href="#id3" title="Permalink to this headline">¶</a></h3>
+<p><strong class="program">llvm-profdata show</strong> [<em>options</em>] [<em>filename</em>]</p>
+</div>
+<div class="section" id="id4">
+<h3>DESCRIPTION<a class="headerlink" href="#id4" title="Permalink to this headline">¶</a></h3>
+<p><strong class="program">llvm-profdata show</strong> takes a profile data file and displays the
+information about the profile counters for this file and
+for any of the specified function(s).</p>
+<p>If <em>filename</em> is omitted or is <tt class="docutils literal"><span class="pre">-</span></tt>, then <strong>llvm-profdata show</strong> reads its
+input from standard input.</p>
+</div>
+<div class="section" id="id5">
+<h3>OPTIONS<a class="headerlink" href="#id5" title="Permalink to this headline">¶</a></h3>
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-show-all-functions">
+<tt class="descname">-all-functions</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-profdata-show-all-functions" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print details for every function.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-show-counts">
+<tt class="descname">-counts</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-profdata-show-counts" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print the counter values for the displayed functions.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-show-function">
+<tt class="descname">-function</tt><tt class="descclassname">=string</tt><a class="headerlink" href="#cmdoption-llvm-profdata-show-function" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print details for a function if the function’s name contains the given string.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-show-help">
+<tt class="descname">-help</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-profdata-show-help" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print a summary of command line options.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-show-output">
+<span id="cmdoption-llvm-profdata-show-o"></span><tt class="descname">-output</tt><tt class="descclassname">=output</tt><tt class="descclassname">, </tt><tt class="descname">-o</tt><tt class="descclassname">=output</tt><a class="headerlink" href="#cmdoption-llvm-profdata-show-output" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the output file name. If <em>output</em> is <tt class="docutils literal"><span class="pre">-</span></tt> or it isn’t specified,
+then the output is sent to standard output.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-show-instr">
+<tt class="descname">-instr</tt><tt class="descclassname"> (default)</tt><a class="headerlink" href="#cmdoption-llvm-profdata-show-instr" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify that the input profile is an instrumentation-based profile.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-show-text">
+<tt class="descname">-text</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-profdata-show-text" title="Permalink to this definition">¶</a></dt>
+<dd><p>Instruct the profile dumper to show profile counts in the text format of the
+instrumentation-based profile data representation. By default, the profile
+information is dumped in a more human readable form (also in text) with
+annotations.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-llvm-profdata-show-sample">
+<tt class="descname">-sample</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-llvm-profdata-show-sample" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify that the input profile is a sample-based profile.</p>
+</dd></dl>
+
+</div>
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-profdata</strong> returns 1 if the command is omitted or is invalid,
+if it cannot read input files, or if there is a mismatch between their data.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-stress.html" title="llvm-stress - generate random .ll files"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-cov.html" title="llvm-cov - emit coverage information"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-readobj.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-readobj.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-readobj.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-readobj.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,217 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-readobj - LLVM Object Reader — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="Getting Started with the LLVM System" href="../GettingStarted.html" />
+ <link rel="prev" title="llvm-build - LLVM Project Build Utility" href="llvm-build.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="../GettingStarted.html" title="Getting Started with the LLVM System"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-build.html" title="llvm-build - LLVM Project Build Utility"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-readobj-llvm-object-reader">
+<h1>llvm-readobj - LLVM Object Reader<a class="headerlink" href="#llvm-readobj-llvm-object-reader" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-readobj</strong> [<em>options</em>] [<em>input...</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p>The <strong class="program">llvm-readobj</strong> tool displays low-level format-specific information
+about one or more object files. The tool and its output is primarily designed
+for use in FileCheck-based tests.</p>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<p>If <tt class="docutils literal"><span class="pre">input</span></tt> is “<tt class="docutils literal"><span class="pre">-</span></tt>” or omitted, <strong class="program">llvm-readobj</strong> reads from standard
+input. Otherwise, it will read from the specified <tt class="docutils literal"><span class="pre">filenames</span></tt>.</p>
+<dl class="option">
+<dt id="cmdoption-help">
+<tt class="descname">-help</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-help" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print a summary of command line options.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-version">
+<tt class="descname">-version</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-version" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display the version of this program</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-file-headers">
+<span id="cmdoption-h"></span><tt class="descname">-file-headers</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-h</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-file-headers" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display file headers.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-sections">
+<span id="cmdoption-s"></span><tt class="descname">-sections</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-s</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-sections" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display all sections.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-section-data">
+<span id="cmdoption-sd"></span><tt class="descname">-section-data</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-sd</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-section-data" title="Permalink to this definition">¶</a></dt>
+<dd><p>When used with <tt class="docutils literal"><span class="pre">-sections</span></tt>, display section data for each section shown.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-section-relocations">
+<span id="cmdoption-sr"></span><tt class="descname">-section-relocations</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-sr</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-section-relocations" title="Permalink to this definition">¶</a></dt>
+<dd><p>When used with <tt class="docutils literal"><span class="pre">-sections</span></tt>, display relocations for each section shown.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-section-symbols">
+<span id="cmdoption-st"></span><tt class="descname">-section-symbols</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-st</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-section-symbols" title="Permalink to this definition">¶</a></dt>
+<dd><p>When used with <tt class="docutils literal"><span class="pre">-sections</span></tt>, display symbols for each section shown.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-relocations">
+<span id="cmdoption-r"></span><tt class="descname">-relocations</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-r</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-relocations" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display the relocation entries in the file.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-symbols">
+<span id="cmdoption-t"></span><tt class="descname">-symbols</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-t</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-symbols" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display the symbol table.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-dyn-symbols">
+<tt class="descname">-dyn-symbols</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-dyn-symbols" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display the dynamic symbol table (only for ELF object files).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-unwind">
+<span id="cmdoption-u"></span><tt class="descname">-unwind</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-u</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-unwind" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display unwind information.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-expand-relocs">
+<tt class="descname">-expand-relocs</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-expand-relocs" title="Permalink to this definition">¶</a></dt>
+<dd><p>When used with <tt class="docutils literal"><span class="pre">-relocations</span></tt>, display each relocation in an expanded
+multi-line format.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-dynamic-table">
+<tt class="descname">-dynamic-table</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-dynamic-table" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display the ELF .dynamic section table (only for ELF object files).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-needed-libs">
+<tt class="descname">-needed-libs</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-needed-libs" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display the needed libraries (only for ELF object files).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-program-headers">
+<tt class="descname">-program-headers</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-program-headers" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display the ELF program headers (only for ELF object files).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-elf-section-groups">
+<span id="cmdoption-g"></span><tt class="descname">-elf-section-groups</tt><tt class="descclassname"></tt><tt class="descclassname">, </tt><tt class="descname">-g</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-elf-section-groups" title="Permalink to this definition">¶</a></dt>
+<dd><p>Display section groups (only for ELF object files).</p>
+</dd></dl>
+
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-readobj</strong> returns 0.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="../GettingStarted.html" title="Getting Started with the LLVM System"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-build.html" title="llvm-build - LLVM Project Build Utility"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-stress.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-stress.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-stress.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-stress.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,135 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-stress - generate random .ll files — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-symbolizer - convert addresses into source code locations" href="llvm-symbolizer.html" />
+ <link rel="prev" title="llvm-profdata - Profile data tool" href="llvm-profdata.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-symbolizer.html" title="llvm-symbolizer - convert addresses into source code locations"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-profdata.html" title="llvm-profdata - Profile data tool"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-stress-generate-random-ll-files">
+<h1>llvm-stress - generate random .ll files<a class="headerlink" href="#llvm-stress-generate-random-ll-files" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-stress</strong> [-size=filesize] [-seed=initialseed] [-o=outfile]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p>The <strong class="program">llvm-stress</strong> tool is used to generate random <tt class="docutils literal"><span class="pre">.ll</span></tt> files that
+can be used to test different components of LLVM.</p>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<dl class="option">
+<dt id="cmdoption-o">
+<tt class="descname">-o</tt><tt class="descclassname"> filename</tt><a class="headerlink" href="#cmdoption-o" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the output filename.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-size">
+<tt class="descname">-size</tt><tt class="descclassname"> size</tt><a class="headerlink" href="#cmdoption-size" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the size of the generated <tt class="docutils literal"><span class="pre">.ll</span></tt> file.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-seed">
+<tt class="descname">-seed</tt><tt class="descclassname"> seed</tt><a class="headerlink" href="#cmdoption-seed" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the seed to be used for the randomly generated instructions.</p>
+</dd></dl>
+
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-stress</strong> returns 0.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-symbolizer.html" title="llvm-symbolizer - convert addresses into source code locations"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-profdata.html" title="llvm-profdata - Profile data tool"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/llvm-symbolizer.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/llvm-symbolizer.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/llvm-symbolizer.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/llvm-symbolizer.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,234 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>llvm-symbolizer - convert addresses into source code locations — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llvm-dwarfdump - print contents of DWARF sections" href="llvm-dwarfdump.html" />
+ <link rel="prev" title="llvm-stress - generate random .ll files" href="llvm-stress.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llvm-dwarfdump.html" title="llvm-dwarfdump - print contents of DWARF sections"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-stress.html" title="llvm-stress - generate random .ll files"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-symbolizer-convert-addresses-into-source-code-locations">
+<h1>llvm-symbolizer - convert addresses into source code locations<a class="headerlink" href="#llvm-symbolizer-convert-addresses-into-source-code-locations" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-symbolizer</strong> [options]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-symbolizer</strong> reads object file names and addresses from standard
+input and prints corresponding source code locations to standard output.
+If object file is specified in command line, <strong class="program">llvm-symbolizer</strong>
+processes only addresses from standard input, the rest is output verbatim.
+This program uses debug info sections and symbol table in the object files.</p>
+</div>
+<div class="section" id="example">
+<h2>EXAMPLE<a class="headerlink" href="#example" title="Permalink to this headline">¶</a></h2>
+<div class="highlight-console"><div class="highlight"><pre><span class="gp">$</span> cat addr.txt
+<span class="go">a.out 0x4004f4</span>
+<span class="go">/tmp/b.out 0x400528</span>
+<span class="go">/tmp/c.so 0x710</span>
+<span class="go">/tmp/mach_universal_binary:i386 0x1f84</span>
+<span class="go">/tmp/mach_universal_binary:x86_64 0x100000f24</span>
+<span class="gp">$</span> llvm-symbolizer < addr.txt
+<span class="go">main</span>
+<span class="go">/tmp/a.cc:4</span>
+
+<span class="go">f(int, int)</span>
+<span class="go">/tmp/b.cc:11</span>
+
+<span class="go">h_inlined_into_g</span>
+<span class="go">/tmp/header.h:2</span>
+<span class="go">g_inlined_into_f</span>
+<span class="go">/tmp/header.h:7</span>
+<span class="go">f_inlined_into_main</span>
+<span class="go">/tmp/source.cc:3</span>
+<span class="go">main</span>
+<span class="go">/tmp/source.cc:8</span>
+
+<span class="go">_main</span>
+<span class="go">/tmp/source_i386.cc:8</span>
+
+<span class="go">_main</span>
+<span class="go">/tmp/source_x86_64.cc:8</span>
+<span class="gp">$</span> cat addr2.txt
+<span class="go">0x4004f4</span>
+<span class="go">0x401000</span>
+<span class="gp">$</span> llvm-symbolizer -obj<span class="o">=</span>a.out < addr2.txt
+<span class="go">main</span>
+<span class="go">/tmp/a.cc:4</span>
+
+<span class="go">foo(int)</span>
+<span class="go">/tmp/a.cc:12</span>
+<span class="gp">$</span>cat addr.txt
+<span class="go">0x40054d</span>
+<span class="gp">$</span>llvm-symbolizer -inlining -print-address -pretty-print -obj<span class="o">=</span>addr.exe < addr.txt
+<span class="go">0x40054d: inc at /tmp/x.c:3:3</span>
+<span class="go"> (inlined by) main at /tmp/x.c:9:0</span>
+<span class="gp">$</span>llvm-symbolizer -inlining -pretty-print -obj<span class="o">=</span>addr.exe < addr.txt
+<span class="go">inc at /tmp/x.c:3:3</span>
+<span class="go"> (inlined by) main at /tmp/x.c:9:0</span>
+</pre></div>
+</div>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<dl class="option">
+<dt id="cmdoption-obj">
+<tt class="descname">-obj</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-obj" title="Permalink to this definition">¶</a></dt>
+<dd><p>Path to object file to be symbolized.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-functions">
+<tt class="descname">-functions</tt><tt class="descclassname">=[none|short|linkage]</tt><a class="headerlink" href="#cmdoption-functions" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the way function names are printed (omit function name,
+print short function name, or print full linkage name, respectively).
+Defaults to <tt class="docutils literal"><span class="pre">linkage</span></tt>.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-use-symbol-table">
+<tt class="descname">-use-symbol-table</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-use-symbol-table" title="Permalink to this definition">¶</a></dt>
+<dd><p>Prefer function names stored in symbol table to function names
+in debug info sections. Defaults to true.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-demangle">
+<tt class="descname">-demangle</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-demangle" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print demangled function names. Defaults to true.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-inlining">
+<tt class="descname">-inlining</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-inlining" title="Permalink to this definition">¶</a></dt>
+<dd><p>If a source code location is in an inlined function, prints all the
+inlnied frames. Defaults to true.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-default-arch">
+<tt class="descname">-default-arch</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-default-arch" title="Permalink to this definition">¶</a></dt>
+<dd><p>If a binary contains object files for multiple architectures (e.g. it is a
+Mach-O universal binary), symbolize the object file for a given architecture.
+You can also specify architecture by writing <tt class="docutils literal"><span class="pre">binary_name:arch_name</span></tt> in the
+input (see example above). If architecture is not specified in either way,
+address will not be symbolized. Defaults to empty string.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-dsym-hint">
+<tt class="descname">-dsym-hint</tt><tt class="descclassname">=<path/to/file.dSYM></tt><a class="headerlink" href="#cmdoption-dsym-hint" title="Permalink to this definition">¶</a></dt>
+<dd><p>(Darwin-only flag). If the debug info for a binary isn’t present in the default
+location, look for the debug info at the .dSYM path provided via the
+<tt class="docutils literal"><span class="pre">-dsym-hint</span></tt> flag. This flag can be used multiple times.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-print-address">
+<tt class="descname">-print-address</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-print-address" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print address before the source code location. Defaults to false.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-pretty-print">
+<tt class="descname">-pretty-print</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-pretty-print" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print human readable output. If <tt class="docutils literal"><span class="pre">-inlining</span></tt> is specified, enclosing scope is
+prefixed by (inlined by). Refer to listed examples.</p>
+</dd></dl>
+
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">llvm-symbolizer</strong> returns 0. Other exit codes imply internal program error.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llvm-dwarfdump.html" title="llvm-dwarfdump - print contents of DWARF sections"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-stress.html" title="llvm-stress - generate random .ll files"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/opt.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/opt.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/opt.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/opt.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,240 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>opt - LLVM optimizer — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="llc - LLVM static compiler" href="llc.html" />
+ <link rel="prev" title="llvm-dis - LLVM disassembler" href="llvm-dis.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="llc.html" title="llc - LLVM static compiler"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="llvm-dis.html" title="llvm-dis - LLVM disassembler"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="opt-llvm-optimizer">
+<h1>opt - LLVM optimizer<a class="headerlink" href="#opt-llvm-optimizer" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">opt</strong> [<em>options</em>] [<em>filename</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p>The <strong class="program">opt</strong> command is the modular LLVM optimizer and analyzer. It
+takes LLVM source files as input, runs the specified optimizations or analyses
+on it, and then outputs the optimized file or the analysis results. The
+function of <strong class="program">opt</strong> depends on whether the <cite>-analyze</cite> option is
+given.</p>
+<p>When <cite>-analyze</cite> is specified, <strong class="program">opt</strong> performs various analyses
+of the input source. It will usually print the results on standard output, but
+in a few cases, it will print output to standard error or generate a file with
+the analysis output, which is usually done when the output is meant for another
+program.</p>
+<p>While <cite>-analyze</cite> is <em>not</em> given, <strong class="program">opt</strong> attempts to produce an
+optimized output file. The optimizations available via <strong class="program">opt</strong> depend
+upon what libraries were linked into it as well as any additional libraries
+that have been loaded with the <a class="reference internal" href="#cmdoption-load"><em class="xref std std-option">-load</em></a> option. Use the <a class="reference internal" href="#cmdoption-help"><em class="xref std std-option">-help</em></a>
+option to determine what optimizations you can use.</p>
+<p>If <tt class="docutils literal"><span class="pre">filename</span></tt> is omitted from the command line or is “<tt class="docutils literal"><span class="pre">-</span></tt>”, <strong class="program">opt</strong>
+reads its input from standard input. Inputs can be in either the LLVM assembly
+language format (<tt class="docutils literal"><span class="pre">.ll</span></tt>) or the LLVM bitcode format (<tt class="docutils literal"><span class="pre">.bc</span></tt>).</p>
+<p>If an output filename is not specified with the <a class="reference internal" href="#cmdoption-o"><em class="xref std std-option">-o</em></a> option,
+<strong class="program">opt</strong> writes its output to the standard output.</p>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<dl class="option">
+<dt id="cmdoption-f">
+<tt class="descname">-f</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-f" title="Permalink to this definition">¶</a></dt>
+<dd><p>Enable binary output on terminals. Normally, <strong class="program">opt</strong> will refuse to
+write raw bitcode output if the output stream is a terminal. With this option,
+<strong class="program">opt</strong> will write raw bitcode regardless of the output device.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-help">
+<tt class="descname">-help</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-help" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print a summary of command line options.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-o">
+<tt class="descname">-o</tt><tt class="descclassname"> <filename></tt><a class="headerlink" href="#cmdoption-o" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the output filename.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-S">
+<tt class="descname">-S</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-S" title="Permalink to this definition">¶</a></dt>
+<dd><p>Write output in LLVM intermediate language (instead of bitcode).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-">
+<tt class="descname">-</tt><tt class="descclassname">{passname}</tt><a class="headerlink" href="#cmdoption-" title="Permalink to this definition">¶</a></dt>
+<dd><p><strong class="program">opt</strong> provides the ability to run any of LLVM’s optimization or
+analysis passes in any order. The <a class="reference internal" href="#cmdoption-help"><em class="xref std std-option">-help</em></a> option lists all the passes
+available. The order in which the options occur on the command line are the
+order in which they are executed (within pass constraints).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-disable-inlining">
+<tt class="descname">-disable-inlining</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-disable-inlining" title="Permalink to this definition">¶</a></dt>
+<dd><p>This option simply removes the inlining pass from the standard list.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-disable-opt">
+<tt class="descname">-disable-opt</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-disable-opt" title="Permalink to this definition">¶</a></dt>
+<dd><p>This option is only meaningful when <cite>-std-link-opts</cite> is given. It
+disables most passes.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-strip-debug">
+<tt class="descname">-strip-debug</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-strip-debug" title="Permalink to this definition">¶</a></dt>
+<dd><p>This option causes opt to strip debug information from the module before
+applying other optimizations. It is essentially the same as <cite>-strip</cite>
+but it ensures that stripping of debug information is done first.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-verify-each">
+<tt class="descname">-verify-each</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-verify-each" title="Permalink to this definition">¶</a></dt>
+<dd><p>This option causes opt to add a verify pass after every pass otherwise
+specified on the command line (including <cite>-verify</cite>). This is useful
+for cases where it is suspected that a pass is creating an invalid module but
+it is not clear which pass is doing it.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-stats">
+<tt class="descname">-stats</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-stats" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print statistics.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-time-passes">
+<tt class="descname">-time-passes</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-time-passes" title="Permalink to this definition">¶</a></dt>
+<dd><p>Record the amount of time needed for each pass and print it to standard
+error.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-debug">
+<tt class="descname">-debug</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-debug" title="Permalink to this definition">¶</a></dt>
+<dd><p>If this is a debug build, this option will enable debug printouts from passes
+which use the <tt class="docutils literal"><span class="pre">DEBUG()</span></tt> macro. See the <a class="reference external" href="../ProgrammersManual.html">LLVM Programmer’s Manual</a>, section <tt class="docutils literal"><span class="pre">#DEBUG</span></tt> for more information.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-load">
+<tt class="descname">-load</tt><tt class="descclassname">=<plugin></tt><a class="headerlink" href="#cmdoption-load" title="Permalink to this definition">¶</a></dt>
+<dd><p>Load the dynamic object <tt class="docutils literal"><span class="pre">plugin</span></tt>. This object should register new
+optimization or analysis passes. Once loaded, the object will add new command
+line options to enable various optimizations or analyses. To see the new
+complete list of optimizations, use the <a class="reference internal" href="#cmdoption-help"><em class="xref std std-option">-help</em></a> and <a class="reference internal" href="#cmdoption-load"><em class="xref std std-option">-load</em></a>
+options together. For example:</p>
+<div class="highlight-sh"><div class="highlight"><pre>opt -load<span class="o">=</span>plugin.so -help
+</pre></div>
+</div>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-p">
+<tt class="descname">-p</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-p" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print module after each transformation.</p>
+</dd></dl>
+
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p>If <strong class="program">opt</strong> succeeds, it will exit with 0. Otherwise, if an error
+occurs, it will exit with a non-zero value.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="llc.html" title="llc - LLVM static compiler"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="llvm-dis.html" title="llvm-dis - LLVM disassembler"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandGuide/tblgen.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandGuide/tblgen.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandGuide/tblgen.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandGuide/tblgen.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,271 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>tblgen - Target Description To C++ Code Generator — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="up" title="LLVM Command Guide" href="index.html" />
+ <link rel="next" title="lit - LLVM Integrated Tester" href="lit.html" />
+ <link rel="prev" title="FileCheck - Flexible pattern matching file verifier" href="FileCheck.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="lit.html" title="lit - LLVM Integrated Tester"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="FileCheck.html" title="FileCheck - Flexible pattern matching file verifier"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" accesskey="U">LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="tblgen-target-description-to-c-code-generator">
+<h1>tblgen - Target Description To C++ Code Generator<a class="headerlink" href="#tblgen-target-description-to-c-code-generator" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="synopsis">
+<h2>SYNOPSIS<a class="headerlink" href="#synopsis" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">tblgen</strong> [<em>options</em>] [<em>filename</em>]</p>
+</div>
+<div class="section" id="description">
+<h2>DESCRIPTION<a class="headerlink" href="#description" title="Permalink to this headline">¶</a></h2>
+<p><strong class="program">tblgen</strong> translates from target description (<tt class="docutils literal"><span class="pre">.td</span></tt>) files into C++
+code that can be included in the definition of an LLVM target library. Most
+users of LLVM will not need to use this program. It is only for assisting with
+writing an LLVM target backend.</p>
+<p>The input and output of <strong class="program">tblgen</strong> is beyond the scope of this short
+introduction; please see the <a class="reference internal" href="../TableGen/index.html"><em>introduction to TableGen</em></a>.</p>
+<p>The <em>filename</em> argument specifies the name of a Target Description (<tt class="docutils literal"><span class="pre">.td</span></tt>)
+file to read as input.</p>
+</div>
+<div class="section" id="options">
+<h2>OPTIONS<a class="headerlink" href="#options" title="Permalink to this headline">¶</a></h2>
+<dl class="option">
+<dt id="cmdoption-tblgen-help">
+<tt class="descname">-help</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-help" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print a summary of command line options.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-o">
+<tt class="descname">-o</tt><tt class="descclassname"> filename</tt><a class="headerlink" href="#cmdoption-tblgen-o" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify the output file name. If <tt class="docutils literal"><span class="pre">filename</span></tt> is <tt class="docutils literal"><span class="pre">-</span></tt>, then
+<strong class="program">tblgen</strong> sends its output to standard output.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-I">
+<tt class="descname">-I</tt><tt class="descclassname"> directory</tt><a class="headerlink" href="#cmdoption-tblgen-I" title="Permalink to this definition">¶</a></dt>
+<dd><p>Specify where to find other target description files for inclusion. The
+<tt class="docutils literal"><span class="pre">directory</span></tt> value should be a full or partial path to a directory that
+contains target description files.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-asmparsernum">
+<tt class="descname">-asmparsernum</tt><tt class="descclassname"> N</tt><a class="headerlink" href="#cmdoption-tblgen-asmparsernum" title="Permalink to this definition">¶</a></dt>
+<dd><p>Make -gen-asm-parser emit assembly writer number <tt class="docutils literal"><span class="pre">N</span></tt>.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-asmwriternum">
+<tt class="descname">-asmwriternum</tt><tt class="descclassname"> N</tt><a class="headerlink" href="#cmdoption-tblgen-asmwriternum" title="Permalink to this definition">¶</a></dt>
+<dd><p>Make -gen-asm-writer emit assembly writer number <tt class="docutils literal"><span class="pre">N</span></tt>.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-class">
+<tt class="descname">-class</tt><tt class="descclassname"> className</tt><a class="headerlink" href="#cmdoption-tblgen-class" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print the enumeration list for this class.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-print-records">
+<tt class="descname">-print-records</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-print-records" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print all records to standard output (default).</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-print-enums">
+<tt class="descname">-print-enums</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-print-enums" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print enumeration values for a class.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-print-sets">
+<tt class="descname">-print-sets</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-print-sets" title="Permalink to this definition">¶</a></dt>
+<dd><p>Print expanded sets for testing DAG exprs.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-emitter">
+<tt class="descname">-gen-emitter</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-emitter" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate machine code emitter.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-register-info">
+<tt class="descname">-gen-register-info</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-register-info" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate registers and register classes info.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-instr-info">
+<tt class="descname">-gen-instr-info</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-instr-info" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate instruction descriptions.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-asm-writer">
+<tt class="descname">-gen-asm-writer</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-asm-writer" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate the assembly writer.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-disassembler">
+<tt class="descname">-gen-disassembler</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-disassembler" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate disassembler.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-pseudo-lowering">
+<tt class="descname">-gen-pseudo-lowering</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-pseudo-lowering" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate pseudo instruction lowering.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-dag-isel">
+<tt class="descname">-gen-dag-isel</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-dag-isel" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate a DAG (Directed Acycle Graph) instruction selector.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-asm-matcher">
+<tt class="descname">-gen-asm-matcher</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-asm-matcher" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate assembly instruction matcher.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-dfa-packetizer">
+<tt class="descname">-gen-dfa-packetizer</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-dfa-packetizer" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate DFA Packetizer for VLIW targets.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-fast-isel">
+<tt class="descname">-gen-fast-isel</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-fast-isel" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate a “fast” instruction selector.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-subtarget">
+<tt class="descname">-gen-subtarget</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-subtarget" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate subtarget enumerations.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-intrinsic">
+<tt class="descname">-gen-intrinsic</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-intrinsic" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate intrinsic information.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-tgt-intrinsic">
+<tt class="descname">-gen-tgt-intrinsic</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-tgt-intrinsic" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate target intrinsic information.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-gen-enhanced-disassembly-info">
+<tt class="descname">-gen-enhanced-disassembly-info</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-gen-enhanced-disassembly-info" title="Permalink to this definition">¶</a></dt>
+<dd><p>Generate enhanced disassembly info.</p>
+</dd></dl>
+
+<dl class="option">
+<dt id="cmdoption-tblgen-version">
+<tt class="descname">-version</tt><tt class="descclassname"></tt><a class="headerlink" href="#cmdoption-tblgen-version" title="Permalink to this definition">¶</a></dt>
+<dd><p>Show the version number of this program.</p>
+</dd></dl>
+
+</div>
+<div class="section" id="exit-status">
+<h2>EXIT STATUS<a class="headerlink" href="#exit-status" title="Permalink to this headline">¶</a></h2>
+<p>If <strong class="program">tblgen</strong> succeeds, it will exit with 0. Otherwise, if an error
+occurs, it will exit with a non-zero value.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="lit.html" title="lit - LLVM Integrated Tester"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="FileCheck.html" title="FileCheck - Flexible pattern matching file verifier"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ <li><a href="index.html" >LLVM Command Guide</a> »</li>
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CommandLine.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CommandLine.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CommandLine.html (added)
+++ www-releases/trunk/4.0.0/docs/CommandLine.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,1571 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>CommandLine 2.0 Library Manual — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="Architecture & Platform Information for Compiler Writers" href="CompilerWriterInfo.html" />
+ <link rel="prev" title="LLVM Coding Standards" href="CodingStandards.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="CompilerWriterInfo.html" title="Architecture & Platform Information for Compiler Writers"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="CodingStandards.html" title="LLVM Coding Standards"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="commandline-2-0-library-manual">
+<h1>CommandLine 2.0 Library Manual<a class="headerlink" href="#commandline-2-0-library-manual" title="Permalink to this headline">¶</a></h1>
+<div class="contents local topic" id="contents">
+<ul class="simple">
+<li><a class="reference internal" href="#introduction" id="id8">Introduction</a></li>
+<li><a class="reference internal" href="#quick-start-guide" id="id9">Quick Start Guide</a><ul>
+<li><a class="reference internal" href="#boolean-arguments" id="id10">Boolean Arguments</a></li>
+<li><a class="reference internal" href="#argument-aliases" id="id11">Argument Aliases</a></li>
+<li><a class="reference internal" href="#selecting-an-alternative-from-a-set-of-possibilities" id="id12">Selecting an alternative from a set of possibilities</a></li>
+<li><a class="reference internal" href="#named-alternatives" id="id13">Named Alternatives</a></li>
+<li><a class="reference internal" href="#parsing-a-list-of-options" id="id14">Parsing a list of options</a></li>
+<li><a class="reference internal" href="#collecting-options-as-a-set-of-flags" id="id15">Collecting options as a set of flags</a></li>
+<li><a class="reference internal" href="#adding-freeform-text-to-help-output" id="id16">Adding freeform text to help output</a></li>
+<li><a class="reference internal" href="#grouping-options-into-categories" id="id17">Grouping options into categories</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#reference-guide" id="id18">Reference Guide</a><ul>
+<li><a class="reference internal" href="#positional-options" id="id19">Positional Arguments</a><ul>
+<li><a class="reference internal" href="#specifying-positional-options-with-hyphens" id="id20">Specifying positional options with hyphens</a></li>
+<li><a class="reference internal" href="#determining-absolute-position-with-getposition" id="id21">Determining absolute position with getPosition()</a></li>
+<li><a class="reference internal" href="#the-cl-consumeafter-modifier" id="id22">The <tt class="docutils literal"><span class="pre">cl::ConsumeAfter</span></tt> modifier</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#internal-vs-external-storage" id="id23">Internal vs External Storage</a></li>
+<li><a class="reference internal" href="#option-attributes" id="id24">Option Attributes</a></li>
+<li><a class="reference internal" href="#option-modifiers" id="id25">Option Modifiers</a><ul>
+<li><a class="reference internal" href="#hiding-an-option-from-help-output" id="id26">Hiding an option from <tt class="docutils literal"><span class="pre">-help</span></tt> output</a></li>
+<li><a class="reference internal" href="#controlling-the-number-of-occurrences-required-and-allowed" id="id27">Controlling the number of occurrences required and allowed</a></li>
+<li><a class="reference internal" href="#controlling-whether-or-not-a-value-must-be-specified" id="id28">Controlling whether or not a value must be specified</a></li>
+<li><a class="reference internal" href="#controlling-other-formatting-options" id="id29">Controlling other formatting options</a></li>
+<li><a class="reference internal" href="#miscellaneous-option-modifiers" id="id30">Miscellaneous option modifiers</a></li>
+<li><a class="reference internal" href="#response-files" id="id31">Response files</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#top-level-classes-and-functions" id="id32">Top-Level Classes and Functions</a><ul>
+<li><a class="reference internal" href="#the-cl-getregisteredoptions-function" id="id33">The <tt class="docutils literal"><span class="pre">cl::getRegisteredOptions</span></tt> function</a></li>
+<li><a class="reference internal" href="#the-cl-parsecommandlineoptions-function" id="id34">The <tt class="docutils literal"><span class="pre">cl::ParseCommandLineOptions</span></tt> function</a></li>
+<li><a class="reference internal" href="#the-cl-parseenvironmentoptions-function" id="id35">The <tt class="docutils literal"><span class="pre">cl::ParseEnvironmentOptions</span></tt> function</a></li>
+<li><a class="reference internal" href="#the-cl-setversionprinter-function" id="id36">The <tt class="docutils literal"><span class="pre">cl::SetVersionPrinter</span></tt> function</a></li>
+<li><a class="reference internal" href="#the-cl-opt-class" id="id37">The <tt class="docutils literal"><span class="pre">cl::opt</span></tt> class</a></li>
+<li><a class="reference internal" href="#the-cl-list-class" id="id38">The <tt class="docutils literal"><span class="pre">cl::list</span></tt> class</a></li>
+<li><a class="reference internal" href="#the-cl-bits-class" id="id39">The <tt class="docutils literal"><span class="pre">cl::bits</span></tt> class</a></li>
+<li><a class="reference internal" href="#the-cl-alias-class" id="id40">The <tt class="docutils literal"><span class="pre">cl::alias</span></tt> class</a></li>
+<li><a class="reference internal" href="#the-cl-extrahelp-class" id="id41">The <tt class="docutils literal"><span class="pre">cl::extrahelp</span></tt> class</a></li>
+<li><a class="reference internal" href="#the-cl-optioncategory-class" id="id42">The <tt class="docutils literal"><span class="pre">cl::OptionCategory</span></tt> class</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#builtin-parsers" id="id43">Builtin parsers</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#extending-the-library" id="id44">Extension Guide</a><ul>
+<li><a class="reference internal" href="#writing-a-custom-parser" id="id45">Writing a custom parser</a></li>
+<li><a class="reference internal" href="#exploiting-external-storage" id="id46">Exploiting external storage</a></li>
+<li><a class="reference internal" href="#dynamically-adding-command-line-options" id="id47">Dynamically adding command line options</a></li>
+</ul>
+</li>
+</ul>
+</div>
+<div class="section" id="introduction">
+<h2><a class="toc-backref" href="#id8">Introduction</a><a class="headerlink" href="#introduction" title="Permalink to this headline">¶</a></h2>
+<p>This document describes the CommandLine argument processing library. It will
+show you how to use it, and what it can do. The CommandLine library uses a
+declarative approach to specifying the command line options that your program
+takes. By default, these options declarations implicitly hold the value parsed
+for the option declared (of course this <a class="reference internal" href="#can-be-changed">can be changed</a>).</p>
+<p>Although there are a <strong>lot</strong> of command line argument parsing libraries out
+there in many different languages, none of them fit well with what I needed. By
+looking at the features and problems of other libraries, I designed the
+CommandLine library to have the following features:</p>
+<ol class="arabic simple">
+<li>Speed: The CommandLine library is very quick and uses little resources. The
+parsing time of the library is directly proportional to the number of
+arguments parsed, not the number of options recognized. Additionally,
+command line argument values are captured transparently into user defined
+global variables, which can be accessed like any other variable (and with the
+same performance).</li>
+<li>Type Safe: As a user of CommandLine, you don’t have to worry about
+remembering the type of arguments that you want (is it an int? a string? a
+bool? an enum?) and keep casting it around. Not only does this help prevent
+error prone constructs, it also leads to dramatically cleaner source code.</li>
+<li>No subclasses required: To use CommandLine, you instantiate variables that
+correspond to the arguments that you would like to capture, you don’t
+subclass a parser. This means that you don’t have to write <strong>any</strong>
+boilerplate code.</li>
+<li>Globally accessible: Libraries can specify command line arguments that are
+automatically enabled in any tool that links to the library. This is
+possible because the application doesn’t have to keep a list of arguments to
+pass to the parser. This also makes supporting <a class="reference internal" href="#dynamically-loaded-options">dynamically loaded options</a>
+trivial.</li>
+<li>Cleaner: CommandLine supports enum and other types directly, meaning that
+there is less error and more security built into the library. You don’t have
+to worry about whether your integral command line argument accidentally got
+assigned a value that is not valid for your enum type.</li>
+<li>Powerful: The CommandLine library supports many different types of arguments,
+from simple <a class="reference internal" href="#boolean-flags">boolean flags</a> to <a class="reference internal" href="#scalars-arguments">scalars arguments</a> (<a class="reference internal" href="#strings">strings</a>,
+<a class="reference internal" href="#integers">integers</a>, <a class="reference internal" href="#enums">enums</a>, <a class="reference internal" href="#doubles">doubles</a>), to <a class="reference internal" href="#lists-of-arguments">lists of arguments</a>. This is
+possible because CommandLine is...</li>
+<li>Extensible: It is very simple to add a new argument type to CommandLine.
+Simply specify the parser that you want to use with the command line option
+when you declare it. <a class="reference internal" href="#custom-parsers">Custom parsers</a> are no problem.</li>
+<li>Labor Saving: The CommandLine library cuts down on the amount of grunt work
+that you, the user, have to do. For example, it automatically provides a
+<tt class="docutils literal"><span class="pre">-help</span></tt> option that shows the available command line options for your tool.
+Additionally, it does most of the basic correctness checking for you.</li>
+<li>Capable: The CommandLine library can handle lots of different forms of
+options often found in real programs. For example, <a class="reference internal" href="#positional">positional</a> arguments,
+<tt class="docutils literal"><span class="pre">ls</span></tt> style <a class="reference internal" href="#grouping">grouping</a> options (to allow processing ‘<tt class="docutils literal"><span class="pre">ls</span> <span class="pre">-lad</span></tt>‘
+naturally), <tt class="docutils literal"><span class="pre">ld</span></tt> style <a class="reference internal" href="#prefix">prefix</a> options (to parse ‘<tt class="docutils literal"><span class="pre">-lmalloc</span>
+<span class="pre">-L/usr/lib</span></tt>‘), and interpreter style options.</li>
+</ol>
+<p>This document will hopefully let you jump in and start using CommandLine in your
+utility quickly and painlessly. Additionally it should be a simple reference
+manual to figure out how stuff works.</p>
+</div>
+<div class="section" id="quick-start-guide">
+<h2><a class="toc-backref" href="#id9">Quick Start Guide</a><a class="headerlink" href="#quick-start-guide" title="Permalink to this headline">¶</a></h2>
+<p>This section of the manual runs through a simple CommandLine’ification of a
+basic compiler tool. This is intended to show you how to jump into using the
+CommandLine library in your own program, and show you some of the cool things it
+can do.</p>
+<p>To start out, you need to include the CommandLine header file into your program:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="cp">#include "llvm/Support/CommandLine.h"</span>
+</pre></div>
+</div>
+<p>Additionally, you need to add this as the first line of your main program:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="kt">int</span> <span class="nf">main</span><span class="p">(</span><span class="kt">int</span> <span class="n">argc</span><span class="p">,</span> <span class="kt">char</span> <span class="o">**</span><span class="n">argv</span><span class="p">)</span> <span class="p">{</span>
+ <span class="n">cl</span><span class="o">::</span><span class="n">ParseCommandLineOptions</span><span class="p">(</span><span class="n">argc</span><span class="p">,</span> <span class="n">argv</span><span class="p">);</span>
+ <span class="p">...</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>... which actually parses the arguments and fills in the variable declarations.</p>
+<p>Now that you are ready to support command line arguments, we need to tell the
+system which ones we want, and what type of arguments they are. The CommandLine
+library uses a declarative syntax to model command line arguments with the
+global variable declarations that capture the parsed values. This means that
+for every command line option that you would like to support, there should be a
+global variable declaration to capture the result. For example, in a compiler,
+we would like to support the Unix-standard ‘<tt class="docutils literal"><span class="pre">-o</span> <span class="pre"><filename></span></tt>‘ option to specify
+where to put the output. With the CommandLine library, this is represented like
+this:</p>
+<div class="highlight-c++" id="here"><span id="scalars-arguments"></span><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="n">string</span><span class="o">></span> <span class="n">OutputFilename</span><span class="p">(</span><span class="s">"o"</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Specify output filename"</span><span class="p">),</span> <span class="n">cl</span><span class="o">::</span><span class="n">value_desc</span><span class="p">(</span><span class="s">"filename"</span><span class="p">));</span>
+</pre></div>
+</div>
+<p>This declares a global variable “<tt class="docutils literal"><span class="pre">OutputFilename</span></tt>” that is used to capture the
+result of the “<tt class="docutils literal"><span class="pre">o</span></tt>” argument (first parameter). We specify that this is a
+simple scalar option by using the “<tt class="docutils literal"><span class="pre">cl::opt</span></tt>” template (as opposed to the
+“<tt class="docutils literal"><span class="pre">cl::list</span></tt>” template), and tell the CommandLine library that the data
+type that we are parsing is a string.</p>
+<p>The second and third parameters (which are optional) are used to specify what to
+output for the “<tt class="docutils literal"><span class="pre">-help</span></tt>” option. In this case, we get a line that looks like
+this:</p>
+<div class="highlight-python"><div class="highlight"><pre>USAGE: compiler [options]
+
+OPTIONS:
+ -help - display available options (-help-hidden for more)
+ -o <filename> - Specify output filename
+</pre></div>
+</div>
+<p>Because we specified that the command line option should parse using the
+<tt class="docutils literal"><span class="pre">string</span></tt> data type, the variable declared is automatically usable as a real
+string in all contexts that a normal C++ string object may be used. For
+example:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="p">...</span>
+<span class="n">std</span><span class="o">::</span><span class="n">ofstream</span> <span class="n">Output</span><span class="p">(</span><span class="n">OutputFilename</span><span class="p">.</span><span class="n">c_str</span><span class="p">());</span>
+<span class="k">if</span> <span class="p">(</span><span class="n">Output</span><span class="p">.</span><span class="n">good</span><span class="p">())</span> <span class="p">...</span>
+<span class="p">...</span>
+</pre></div>
+</div>
+<p>There are many different options that you can use to customize the command line
+option handling library, but the above example shows the general interface to
+these options. The options can be specified in any order, and are specified
+with helper functions like <a class="reference internal" href="#cl-desc">cl::desc(...)</a>, so there are no positional
+dependencies to remember. The available options are discussed in detail in the
+<a class="reference internal" href="#reference-guide">Reference Guide</a>.</p>
+<p>Continuing the example, we would like to have our compiler take an input
+filename as well as an output filename, but we do not want the input filename to
+be specified with a hyphen (ie, not <tt class="docutils literal"><span class="pre">-filename.c</span></tt>). To support this style of
+argument, the CommandLine library allows for <a class="reference internal" href="#positional">positional</a> arguments to be
+specified for the program. These positional arguments are filled with command
+line parameters that are not in option form. We use this feature like this:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="n">string</span><span class="o">></span> <span class="n">InputFilename</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">Positional</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"<input file>"</span><span class="p">),</span> <span class="n">cl</span><span class="o">::</span><span class="n">init</span><span class="p">(</span><span class="s">"-"</span><span class="p">));</span>
+</pre></div>
+</div>
+<p>This declaration indicates that the first positional argument should be treated
+as the input filename. Here we use the <a class="reference internal" href="#cl-init">cl::init</a> option to specify an initial
+value for the command line option, which is used if the option is not specified
+(if you do not specify a <a class="reference internal" href="#cl-init">cl::init</a> modifier for an option, then the default
+constructor for the data type is used to initialize the value). Command line
+options default to being optional, so if we would like to require that the user
+always specify an input filename, we would add the <a class="reference internal" href="#cl-required">cl::Required</a> flag, and we
+could eliminate the <a class="reference internal" href="#cl-init">cl::init</a> modifier, like this:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="n">string</span><span class="o">></span> <span class="n">InputFilename</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">Positional</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"<input file>"</span><span class="p">),</span> <span class="n">cl</span><span class="o">::</span><span class="n">Required</span><span class="p">);</span>
+</pre></div>
+</div>
+<p>Again, the CommandLine library does not require the options to be specified in
+any particular order, so the above declaration is equivalent to:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="n">string</span><span class="o">></span> <span class="n">InputFilename</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">Positional</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">Required</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"<input file>"</span><span class="p">));</span>
+</pre></div>
+</div>
+<p>By simply adding the <a class="reference internal" href="#cl-required">cl::Required</a> flag, the CommandLine library will
+automatically issue an error if the argument is not specified, which shifts all
+of the command line option verification code out of your application into the
+library. This is just one example of how using flags can alter the default
+behaviour of the library, on a per-option basis. By adding one of the
+declarations above, the <tt class="docutils literal"><span class="pre">-help</span></tt> option synopsis is now extended to:</p>
+<div class="highlight-python"><div class="highlight"><pre>USAGE: compiler [options] <input file>
+
+OPTIONS:
+ -help - display available options (-help-hidden for more)
+ -o <filename> - Specify output filename
+</pre></div>
+</div>
+<p>... indicating that an input filename is expected.</p>
+<div class="section" id="boolean-arguments">
+<h3><a class="toc-backref" href="#id10">Boolean Arguments</a><a class="headerlink" href="#boolean-arguments" title="Permalink to this headline">¶</a></h3>
+<p>In addition to input and output filenames, we would like the compiler example to
+support three boolean flags: “<tt class="docutils literal"><span class="pre">-f</span></tt>” to force writing binary output to a
+terminal, “<tt class="docutils literal"><span class="pre">--quiet</span></tt>” to enable quiet mode, and “<tt class="docutils literal"><span class="pre">-q</span></tt>” for backwards
+compatibility with some of our users. We can support these by declaring options
+of boolean type like this:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="kt">bool</span><span class="o">></span> <span class="n">Force</span> <span class="p">(</span><span class="s">"f"</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Enable binary output on terminals"</span><span class="p">));</span>
+<span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="kt">bool</span><span class="o">></span> <span class="n">Quiet</span> <span class="p">(</span><span class="s">"quiet"</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Don't print informational messages"</span><span class="p">));</span>
+<span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="kt">bool</span><span class="o">></span> <span class="n">Quiet2</span><span class="p">(</span><span class="s">"q"</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Don't print informational messages"</span><span class="p">),</span> <span class="n">cl</span><span class="o">::</span><span class="n">Hidden</span><span class="p">);</span>
+</pre></div>
+</div>
+<p>This does what you would expect: it declares three boolean variables
+(“<tt class="docutils literal"><span class="pre">Force</span></tt>”, “<tt class="docutils literal"><span class="pre">Quiet</span></tt>”, and “<tt class="docutils literal"><span class="pre">Quiet2</span></tt>”) to recognize these options. Note
+that the “<tt class="docutils literal"><span class="pre">-q</span></tt>” option is specified with the “<a class="reference internal" href="#cl-hidden">cl::Hidden</a>” flag. This
+modifier prevents it from being shown by the standard “<tt class="docutils literal"><span class="pre">-help</span></tt>” output (note
+that it is still shown in the “<tt class="docutils literal"><span class="pre">-help-hidden</span></tt>” output).</p>
+<p>The CommandLine library uses a <a class="reference internal" href="#different-parser">different parser</a> for different data types.
+For example, in the string case, the argument passed to the option is copied
+literally into the content of the string variable... we obviously cannot do that
+in the boolean case, however, so we must use a smarter parser. In the case of
+the boolean parser, it allows no options (in which case it assigns the value of
+true to the variable), or it allows the values “<tt class="docutils literal"><span class="pre">true</span></tt>” or “<tt class="docutils literal"><span class="pre">false</span></tt>” to be
+specified, allowing any of the following inputs:</p>
+<div class="highlight-python"><div class="highlight"><pre><span class="n">compiler</span> <span class="o">-</span><span class="n">f</span> <span class="c"># No value, 'Force' == true</span>
+<span class="n">compiler</span> <span class="o">-</span><span class="n">f</span><span class="o">=</span><span class="n">true</span> <span class="c"># Value specified, 'Force' == true</span>
+<span class="n">compiler</span> <span class="o">-</span><span class="n">f</span><span class="o">=</span><span class="n">TRUE</span> <span class="c"># Value specified, 'Force' == true</span>
+<span class="n">compiler</span> <span class="o">-</span><span class="n">f</span><span class="o">=</span><span class="n">FALSE</span> <span class="c"># Value specified, 'Force' == false</span>
+</pre></div>
+</div>
+<p>... you get the idea. The <a class="reference internal" href="#bool-parser">bool parser</a> just turns the string values into
+boolean values, and rejects things like ‘<tt class="docutils literal"><span class="pre">compiler</span> <span class="pre">-f=foo</span></tt>‘. Similarly, the
+<a class="reference internal" href="#float">float</a>, <a class="reference internal" href="#double">double</a>, and <a class="reference internal" href="#int">int</a> parsers work like you would expect, using the
+‘<tt class="docutils literal"><span class="pre">strtol</span></tt>‘ and ‘<tt class="docutils literal"><span class="pre">strtod</span></tt>‘ C library calls to parse the string value into the
+specified data type.</p>
+<p>With the declarations above, “<tt class="docutils literal"><span class="pre">compiler</span> <span class="pre">-help</span></tt>” emits this:</p>
+<div class="highlight-python"><div class="highlight"><pre>USAGE: compiler [options] <input file>
+
+OPTIONS:
+ -f - Enable binary output on terminals
+ -o - Override output filename
+ -quiet - Don't print informational messages
+ -help - display available options (-help-hidden for more)
+</pre></div>
+</div>
+<p>and “<tt class="docutils literal"><span class="pre">compiler</span> <span class="pre">-help-hidden</span></tt>” prints this:</p>
+<div class="highlight-python"><div class="highlight"><pre>USAGE: compiler [options] <input file>
+
+OPTIONS:
+ -f - Enable binary output on terminals
+ -o - Override output filename
+ -q - Don't print informational messages
+ -quiet - Don't print informational messages
+ -help - display available options (-help-hidden for more)
+</pre></div>
+</div>
+<p>This brief example has shown you how to use the ‘<a class="reference internal" href="#cl-opt">cl::opt</a>‘ class to parse
+simple scalar command line arguments. In addition to simple scalar arguments,
+the CommandLine library also provides primitives to support CommandLine option
+<a class="reference internal" href="#aliases">aliases</a>, and <a class="reference internal" href="#lists">lists</a> of options.</p>
+</div>
+<div class="section" id="argument-aliases">
+<span id="aliases"></span><h3><a class="toc-backref" href="#id11">Argument Aliases</a><a class="headerlink" href="#argument-aliases" title="Permalink to this headline">¶</a></h3>
+<p>So far, the example works well, except for the fact that we need to check the
+quiet condition like this now:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="p">...</span>
+ <span class="k">if</span> <span class="p">(</span><span class="o">!</span><span class="n">Quiet</span> <span class="o">&&</span> <span class="o">!</span><span class="n">Quiet2</span><span class="p">)</span> <span class="n">printInformationalMessage</span><span class="p">(...);</span>
+<span class="p">...</span>
+</pre></div>
+</div>
+<p>... which is a real pain! Instead of defining two values for the same
+condition, we can use the “<a class="reference internal" href="#cl-alias">cl::alias</a>” class to make the “<tt class="docutils literal"><span class="pre">-q</span></tt>” option an
+<strong>alias</strong> for the “<tt class="docutils literal"><span class="pre">-quiet</span></tt>” option, instead of providing a value itself:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="kt">bool</span><span class="o">></span> <span class="n">Force</span> <span class="p">(</span><span class="s">"f"</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Overwrite output files"</span><span class="p">));</span>
+<span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="kt">bool</span><span class="o">></span> <span class="n">Quiet</span> <span class="p">(</span><span class="s">"quiet"</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Don't print informational messages"</span><span class="p">));</span>
+<span class="n">cl</span><span class="o">::</span><span class="n">alias</span> <span class="n">QuietA</span><span class="p">(</span><span class="s">"q"</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Alias for -quiet"</span><span class="p">),</span> <span class="n">cl</span><span class="o">::</span><span class="n">aliasopt</span><span class="p">(</span><span class="n">Quiet</span><span class="p">));</span>
+</pre></div>
+</div>
+<p>The third line (which is the only one we modified from above) defines a “<tt class="docutils literal"><span class="pre">-q</span></tt>”
+alias that updates the “<tt class="docutils literal"><span class="pre">Quiet</span></tt>” variable (as specified by the <a class="reference internal" href="#cl-aliasopt">cl::aliasopt</a>
+modifier) whenever it is specified. Because aliases do not hold state, the only
+thing the program has to query is the <tt class="docutils literal"><span class="pre">Quiet</span></tt> variable now. Another nice
+feature of aliases is that they automatically hide themselves from the <tt class="docutils literal"><span class="pre">-help</span></tt>
+output (although, again, they are still visible in the <tt class="docutils literal"><span class="pre">-help-hidden</span> <span class="pre">output</span></tt>).</p>
+<p>Now the application code can simply use:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="p">...</span>
+ <span class="k">if</span> <span class="p">(</span><span class="o">!</span><span class="n">Quiet</span><span class="p">)</span> <span class="n">printInformationalMessage</span><span class="p">(...);</span>
+<span class="p">...</span>
+</pre></div>
+</div>
+<p>... which is much nicer! The “<a class="reference internal" href="#cl-alias">cl::alias</a>” can be used to specify an
+alternative name for any variable type, and has many uses.</p>
+</div>
+<div class="section" id="selecting-an-alternative-from-a-set-of-possibilities">
+<span id="unnamed-alternatives-using-the-generic-parser"></span><h3><a class="toc-backref" href="#id12">Selecting an alternative from a set of possibilities</a><a class="headerlink" href="#selecting-an-alternative-from-a-set-of-possibilities" title="Permalink to this headline">¶</a></h3>
+<p>So far we have seen how the CommandLine library handles builtin types like
+<tt class="docutils literal"><span class="pre">std::string</span></tt>, <tt class="docutils literal"><span class="pre">bool</span></tt> and <tt class="docutils literal"><span class="pre">int</span></tt>, but how does it handle things it doesn’t
+know about, like enums or ‘<tt class="docutils literal"><span class="pre">int*</span></tt>‘s?</p>
+<p>The answer is that it uses a table-driven generic parser (unless you specify
+your own parser, as described in the <a class="reference internal" href="#extension-guide">Extension Guide</a>). This parser maps
+literal strings to whatever type is required, and requires you to tell it what
+this mapping should be.</p>
+<p>Let’s say that we would like to add four optimization levels to our optimizer,
+using the standard flags “<tt class="docutils literal"><span class="pre">-g</span></tt>”, “<tt class="docutils literal"><span class="pre">-O0</span></tt>”, “<tt class="docutils literal"><span class="pre">-O1</span></tt>”, and “<tt class="docutils literal"><span class="pre">-O2</span></tt>”. We
+could easily implement this with boolean options like above, but there are
+several problems with this strategy:</p>
+<ol class="arabic simple">
+<li>A user could specify more than one of the options at a time, for example,
+“<tt class="docutils literal"><span class="pre">compiler</span> <span class="pre">-O3</span> <span class="pre">-O2</span></tt>”. The CommandLine library would not be able to catch
+this erroneous input for us.</li>
+<li>We would have to test 4 different variables to see which ones are set.</li>
+<li>This doesn’t map to the numeric levels that we want... so we cannot easily
+see if some level >= “<tt class="docutils literal"><span class="pre">-O1</span></tt>” is enabled.</li>
+</ol>
+<p>To cope with these problems, we can use an enum value, and have the CommandLine
+library fill it in with the appropriate level directly, which is used like this:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">enum</span> <span class="n">OptLevel</span> <span class="p">{</span>
+ <span class="n">g</span><span class="p">,</span> <span class="n">O1</span><span class="p">,</span> <span class="n">O2</span><span class="p">,</span> <span class="n">O3</span>
+<span class="p">};</span>
+
+<span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="n">OptLevel</span><span class="o">></span> <span class="n">OptimizationLevel</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Choose optimization level:"</span><span class="p">),</span>
+ <span class="n">cl</span><span class="o">::</span><span class="n">values</span><span class="p">(</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">g</span> <span class="p">,</span> <span class="s">"No optimizations, enable debugging"</span><span class="p">),</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">O1</span><span class="p">,</span> <span class="s">"Enable trivial optimizations"</span><span class="p">),</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">O2</span><span class="p">,</span> <span class="s">"Enable default optimizations"</span><span class="p">),</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">O3</span><span class="p">,</span> <span class="s">"Enable expensive optimizations"</span><span class="p">)));</span>
+
+<span class="p">...</span>
+ <span class="k">if</span> <span class="p">(</span><span class="n">OptimizationLevel</span> <span class="o">>=</span> <span class="n">O2</span><span class="p">)</span> <span class="n">doPartialRedundancyElimination</span><span class="p">(...);</span>
+<span class="p">...</span>
+</pre></div>
+</div>
+<p>This declaration defines a variable “<tt class="docutils literal"><span class="pre">OptimizationLevel</span></tt>” of the
+“<tt class="docutils literal"><span class="pre">OptLevel</span></tt>” enum type. This variable can be assigned any of the values that
+are listed in the declaration. The CommandLine library enforces that
+the user can only specify one of the options, and it ensure that only valid enum
+values can be specified. The “<tt class="docutils literal"><span class="pre">clEnumVal</span></tt>” macros ensure that the command
+line arguments matched the enum values. With this option added, our help output
+now is:</p>
+<div class="highlight-python"><div class="highlight"><pre>USAGE: compiler [options] <input file>
+
+OPTIONS:
+ Choose optimization level:
+ -g - No optimizations, enable debugging
+ -O1 - Enable trivial optimizations
+ -O2 - Enable default optimizations
+ -O3 - Enable expensive optimizations
+ -f - Enable binary output on terminals
+ -help - display available options (-help-hidden for more)
+ -o <filename> - Specify output filename
+ -quiet - Don't print informational messages
+</pre></div>
+</div>
+<p>In this case, it is sort of awkward that flag names correspond directly to enum
+names, because we probably don’t want a enum definition named “<tt class="docutils literal"><span class="pre">g</span></tt>” in our
+program. Because of this, we can alternatively write this example like this:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">enum</span> <span class="n">OptLevel</span> <span class="p">{</span>
+ <span class="n">Debug</span><span class="p">,</span> <span class="n">O1</span><span class="p">,</span> <span class="n">O2</span><span class="p">,</span> <span class="n">O3</span>
+<span class="p">};</span>
+
+<span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="n">OptLevel</span><span class="o">></span> <span class="n">OptimizationLevel</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Choose optimization level:"</span><span class="p">),</span>
+ <span class="n">cl</span><span class="o">::</span><span class="n">values</span><span class="p">(</span>
+ <span class="n">clEnumValN</span><span class="p">(</span><span class="n">Debug</span><span class="p">,</span> <span class="s">"g"</span><span class="p">,</span> <span class="s">"No optimizations, enable debugging"</span><span class="p">),</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">O1</span> <span class="p">,</span> <span class="s">"Enable trivial optimizations"</span><span class="p">),</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">O2</span> <span class="p">,</span> <span class="s">"Enable default optimizations"</span><span class="p">),</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">O3</span> <span class="p">,</span> <span class="s">"Enable expensive optimizations"</span><span class="p">)));</span>
+
+<span class="p">...</span>
+ <span class="k">if</span> <span class="p">(</span><span class="n">OptimizationLevel</span> <span class="o">==</span> <span class="n">Debug</span><span class="p">)</span> <span class="n">outputDebugInfo</span><span class="p">(...);</span>
+<span class="p">...</span>
+</pre></div>
+</div>
+<p>By using the “<tt class="docutils literal"><span class="pre">clEnumValN</span></tt>” macro instead of “<tt class="docutils literal"><span class="pre">clEnumVal</span></tt>”, we can directly
+specify the name that the flag should get. In general a direct mapping is nice,
+but sometimes you can’t or don’t want to preserve the mapping, which is when you
+would use it.</p>
+</div>
+<div class="section" id="named-alternatives">
+<h3><a class="toc-backref" href="#id13">Named Alternatives</a><a class="headerlink" href="#named-alternatives" title="Permalink to this headline">¶</a></h3>
+<p>Another useful argument form is a named alternative style. We shall use this
+style in our compiler to specify different debug levels that can be used.
+Instead of each debug level being its own switch, we want to support the
+following options, of which only one can be specified at a time:
+“<tt class="docutils literal"><span class="pre">--debug-level=none</span></tt>”, “<tt class="docutils literal"><span class="pre">--debug-level=quick</span></tt>”,
+“<tt class="docutils literal"><span class="pre">--debug-level=detailed</span></tt>”. To do this, we use the exact same format as our
+optimization level flags, but we also specify an option name. For this case,
+the code looks like this:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">enum</span> <span class="n">DebugLev</span> <span class="p">{</span>
+ <span class="n">nodebuginfo</span><span class="p">,</span> <span class="n">quick</span><span class="p">,</span> <span class="n">detailed</span>
+<span class="p">};</span>
+
+<span class="c1">// Enable Debug Options to be specified on the command line</span>
+<span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="n">DebugLev</span><span class="o">></span> <span class="n">DebugLevel</span><span class="p">(</span><span class="s">"debug_level"</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Set the debugging level:"</span><span class="p">),</span>
+ <span class="n">cl</span><span class="o">::</span><span class="n">values</span><span class="p">(</span>
+ <span class="n">clEnumValN</span><span class="p">(</span><span class="n">nodebuginfo</span><span class="p">,</span> <span class="s">"none"</span><span class="p">,</span> <span class="s">"disable debug information"</span><span class="p">),</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">quick</span><span class="p">,</span> <span class="s">"enable quick debug information"</span><span class="p">),</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">detailed</span><span class="p">,</span> <span class="s">"enable detailed debug information"</span><span class="p">)));</span>
+</pre></div>
+</div>
+<p>This definition defines an enumerated command line variable of type “<tt class="docutils literal"><span class="pre">enum</span>
+<span class="pre">DebugLev</span></tt>”, which works exactly the same way as before. The difference here is
+just the interface exposed to the user of your program and the help output by
+the “<tt class="docutils literal"><span class="pre">-help</span></tt>” option:</p>
+<div class="highlight-python"><div class="highlight"><pre>USAGE: compiler [options] <input file>
+
+OPTIONS:
+ Choose optimization level:
+ -g - No optimizations, enable debugging
+ -O1 - Enable trivial optimizations
+ -O2 - Enable default optimizations
+ -O3 - Enable expensive optimizations
+ -debug_level - Set the debugging level:
+ =none - disable debug information
+ =quick - enable quick debug information
+ =detailed - enable detailed debug information
+ -f - Enable binary output on terminals
+ -help - display available options (-help-hidden for more)
+ -o <filename> - Specify output filename
+ -quiet - Don't print informational messages
+</pre></div>
+</div>
+<p>Again, the only structural difference between the debug level declaration and
+the optimization level declaration is that the debug level declaration includes
+an option name (<tt class="docutils literal"><span class="pre">"debug_level"</span></tt>), which automatically changes how the library
+processes the argument. The CommandLine library supports both forms so that you
+can choose the form most appropriate for your application.</p>
+</div>
+<div class="section" id="parsing-a-list-of-options">
+<span id="lists"></span><h3><a class="toc-backref" href="#id14">Parsing a list of options</a><a class="headerlink" href="#parsing-a-list-of-options" title="Permalink to this headline">¶</a></h3>
+<p>Now that we have the standard run-of-the-mill argument types out of the way,
+lets get a little wild and crazy. Lets say that we want our optimizer to accept
+a <strong>list</strong> of optimizations to perform, allowing duplicates. For example, we
+might want to run: “<tt class="docutils literal"><span class="pre">compiler</span> <span class="pre">-dce</span> <span class="pre">-constprop</span> <span class="pre">-inline</span> <span class="pre">-dce</span> <span class="pre">-strip</span></tt>”. In this
+case, the order of the arguments and the number of appearances is very
+important. This is what the “<tt class="docutils literal"><span class="pre">cl::list</span></tt>” template is for. First, start by
+defining an enum of the optimizations that you would like to perform:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">enum</span> <span class="n">Opts</span> <span class="p">{</span>
+ <span class="c1">// 'inline' is a C++ keyword, so name it 'inlining'</span>
+ <span class="n">dce</span><span class="p">,</span> <span class="n">constprop</span><span class="p">,</span> <span class="n">inlining</span><span class="p">,</span> <span class="n">strip</span>
+<span class="p">};</span>
+</pre></div>
+</div>
+<p>Then define your “<tt class="docutils literal"><span class="pre">cl::list</span></tt>” variable:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">list</span><span class="o"><</span><span class="n">Opts</span><span class="o">></span> <span class="n">OptimizationList</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Available Optimizations:"</span><span class="p">),</span>
+ <span class="n">cl</span><span class="o">::</span><span class="n">values</span><span class="p">(</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">dce</span> <span class="p">,</span> <span class="s">"Dead Code Elimination"</span><span class="p">),</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">constprop</span> <span class="p">,</span> <span class="s">"Constant Propagation"</span><span class="p">),</span>
+ <span class="n">clEnumValN</span><span class="p">(</span><span class="n">inlining</span><span class="p">,</span> <span class="s">"inline"</span><span class="p">,</span> <span class="s">"Procedure Integration"</span><span class="p">),</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">strip</span> <span class="p">,</span> <span class="s">"Strip Symbols"</span><span class="p">)));</span>
+</pre></div>
+</div>
+<p>This defines a variable that is conceptually of the type
+“<tt class="docutils literal"><span class="pre">std::vector<enum</span> <span class="pre">Opts></span></tt>”. Thus, you can access it with standard vector
+methods:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">for</span> <span class="p">(</span><span class="kt">unsigned</span> <span class="n">i</span> <span class="o">=</span> <span class="mi">0</span><span class="p">;</span> <span class="n">i</span> <span class="o">!=</span> <span class="n">OptimizationList</span><span class="p">.</span><span class="n">size</span><span class="p">();</span> <span class="o">++</span><span class="n">i</span><span class="p">)</span>
+ <span class="k">switch</span> <span class="p">(</span><span class="n">OptimizationList</span><span class="p">[</span><span class="n">i</span><span class="p">])</span>
+ <span class="p">...</span>
+</pre></div>
+</div>
+<p>... to iterate through the list of options specified.</p>
+<p>Note that the “<tt class="docutils literal"><span class="pre">cl::list</span></tt>” template is completely general and may be used with
+any data types or other arguments that you can use with the “<tt class="docutils literal"><span class="pre">cl::opt</span></tt>”
+template. One especially useful way to use a list is to capture all of the
+positional arguments together if there may be more than one specified. In the
+case of a linker, for example, the linker takes several ‘<tt class="docutils literal"><span class="pre">.o</span></tt>‘ files, and
+needs to capture them into a list. This is naturally specified as:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="p">...</span>
+<span class="n">cl</span><span class="o">::</span><span class="n">list</span><span class="o"><</span><span class="n">std</span><span class="o">::</span><span class="n">string</span><span class="o">></span> <span class="n">InputFilenames</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">Positional</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"<Input files>"</span><span class="p">),</span> <span class="n">cl</span><span class="o">::</span><span class="n">OneOrMore</span><span class="p">);</span>
+<span class="p">...</span>
+</pre></div>
+</div>
+<p>This variable works just like a “<tt class="docutils literal"><span class="pre">vector<string></span></tt>” object. As such, accessing
+the list is simple, just like above. In this example, we used the
+<a class="reference internal" href="#cl-oneormore">cl::OneOrMore</a> modifier to inform the CommandLine library that it is an error
+if the user does not specify any <tt class="docutils literal"><span class="pre">.o</span></tt> files on our command line. Again, this
+just reduces the amount of checking we have to do.</p>
+</div>
+<div class="section" id="collecting-options-as-a-set-of-flags">
+<h3><a class="toc-backref" href="#id15">Collecting options as a set of flags</a><a class="headerlink" href="#collecting-options-as-a-set-of-flags" title="Permalink to this headline">¶</a></h3>
+<p>Instead of collecting sets of options in a list, it is also possible to gather
+information for enum values in a <strong>bit vector</strong>. The representation used by the
+<a class="reference internal" href="#cl-bits">cl::bits</a> class is an <tt class="docutils literal"><span class="pre">unsigned</span></tt> integer. An enum value is represented by a
+0/1 in the enum’s ordinal value bit position. 1 indicating that the enum was
+specified, 0 otherwise. As each specified value is parsed, the resulting enum’s
+bit is set in the option’s bit vector:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">bits</span> <span class="o">|=</span> <span class="mi">1</span> <span class="o"><<</span> <span class="p">(</span><span class="kt">unsigned</span><span class="p">)</span><span class="k">enum</span><span class="p">;</span>
+</pre></div>
+</div>
+<p>Options that are specified multiple times are redundant. Any instances after
+the first are discarded.</p>
+<p>Reworking the above list example, we could replace <a class="reference internal" href="#cl-list">cl::list</a> with <a class="reference internal" href="#cl-bits">cl::bits</a>:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">bits</span><span class="o"><</span><span class="n">Opts</span><span class="o">></span> <span class="n">OptimizationBits</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Available Optimizations:"</span><span class="p">),</span>
+ <span class="n">cl</span><span class="o">::</span><span class="n">values</span><span class="p">(</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">dce</span> <span class="p">,</span> <span class="s">"Dead Code Elimination"</span><span class="p">),</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">constprop</span> <span class="p">,</span> <span class="s">"Constant Propagation"</span><span class="p">),</span>
+ <span class="n">clEnumValN</span><span class="p">(</span><span class="n">inlining</span><span class="p">,</span> <span class="s">"inline"</span><span class="p">,</span> <span class="s">"Procedure Integration"</span><span class="p">),</span>
+ <span class="n">clEnumVal</span><span class="p">(</span><span class="n">strip</span> <span class="p">,</span> <span class="s">"Strip Symbols"</span><span class="p">)));</span>
+</pre></div>
+</div>
+<p>To test to see if <tt class="docutils literal"><span class="pre">constprop</span></tt> was specified, we can use the <tt class="docutils literal"><span class="pre">cl:bits::isSet</span></tt>
+function:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">if</span> <span class="p">(</span><span class="n">OptimizationBits</span><span class="p">.</span><span class="n">isSet</span><span class="p">(</span><span class="n">constprop</span><span class="p">))</span> <span class="p">{</span>
+ <span class="p">...</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>It’s also possible to get the raw bit vector using the <tt class="docutils literal"><span class="pre">cl::bits::getBits</span></tt>
+function:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="kt">unsigned</span> <span class="n">bits</span> <span class="o">=</span> <span class="n">OptimizationBits</span><span class="p">.</span><span class="n">getBits</span><span class="p">();</span>
+</pre></div>
+</div>
+<p>Finally, if external storage is used, then the location specified must be of
+<strong>type</strong> <tt class="docutils literal"><span class="pre">unsigned</span></tt>. In all other ways a <a class="reference internal" href="#cl-bits">cl::bits</a> option is equivalent to a
+<a class="reference internal" href="#cl-list">cl::list</a> option.</p>
+</div>
+<div class="section" id="adding-freeform-text-to-help-output">
+<span id="additional-extra-text"></span><h3><a class="toc-backref" href="#id16">Adding freeform text to help output</a><a class="headerlink" href="#adding-freeform-text-to-help-output" title="Permalink to this headline">¶</a></h3>
+<p>As our program grows and becomes more mature, we may decide to put summary
+information about what it does into the help output. The help output is styled
+to look similar to a Unix <tt class="docutils literal"><span class="pre">man</span></tt> page, providing concise information about a
+program. Unix <tt class="docutils literal"><span class="pre">man</span></tt> pages, however often have a description about what the
+program does. To add this to your CommandLine program, simply pass a third
+argument to the <a class="reference internal" href="#cl-parsecommandlineoptions">cl::ParseCommandLineOptions</a> call in main. This additional
+argument is then printed as the overview information for your program, allowing
+you to include any additional information that you want. For example:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="kt">int</span> <span class="nf">main</span><span class="p">(</span><span class="kt">int</span> <span class="n">argc</span><span class="p">,</span> <span class="kt">char</span> <span class="o">**</span><span class="n">argv</span><span class="p">)</span> <span class="p">{</span>
+ <span class="n">cl</span><span class="o">::</span><span class="n">ParseCommandLineOptions</span><span class="p">(</span><span class="n">argc</span><span class="p">,</span> <span class="n">argv</span><span class="p">,</span> <span class="s">" CommandLine compiler example</span><span class="se">\n\n</span><span class="s">"</span>
+ <span class="s">" This program blah blah blah...</span><span class="se">\n</span><span class="s">"</span><span class="p">);</span>
+ <span class="p">...</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>would yield the help output:</p>
+<div class="highlight-python"><div class="highlight"><pre>**OVERVIEW: CommandLine compiler example
+
+ This program blah blah blah...**
+
+USAGE: compiler [options] <input file>
+
+OPTIONS:
+ ...
+ -help - display available options (-help-hidden for more)
+ -o <filename> - Specify output filename
+</pre></div>
+</div>
+</div>
+<div class="section" id="grouping-options-into-categories">
+<span id="id1"></span><h3><a class="toc-backref" href="#id17">Grouping options into categories</a><a class="headerlink" href="#grouping-options-into-categories" title="Permalink to this headline">¶</a></h3>
+<p>If our program has a large number of options it may become difficult for users
+of our tool to navigate the output of <tt class="docutils literal"><span class="pre">-help</span></tt>. To alleviate this problem we
+can put our options into categories. This can be done by declaring option
+categories (<a class="reference internal" href="#cl-optioncategory">cl::OptionCategory</a> objects) and then placing our options into
+these categories using the <a class="reference internal" href="#cl-cat">cl::cat</a> option attribute. For example:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">OptionCategory</span> <span class="n">StageSelectionCat</span><span class="p">(</span><span class="s">"Stage Selection Options"</span><span class="p">,</span>
+ <span class="s">"These control which stages are run."</span><span class="p">);</span>
+
+<span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="kt">bool</span><span class="o">></span> <span class="n">Preprocessor</span><span class="p">(</span><span class="s">"E"</span><span class="p">,</span><span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Run preprocessor stage."</span><span class="p">),</span>
+ <span class="n">cl</span><span class="o">::</span><span class="n">cat</span><span class="p">(</span><span class="n">StageSelectionCat</span><span class="p">));</span>
+
+<span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="kt">bool</span><span class="o">></span> <span class="n">NoLink</span><span class="p">(</span><span class="s">"c"</span><span class="p">,</span><span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Run all stages except linking."</span><span class="p">),</span>
+ <span class="n">cl</span><span class="o">::</span><span class="n">cat</span><span class="p">(</span><span class="n">StageSelectionCat</span><span class="p">));</span>
+</pre></div>
+</div>
+<p>The output of <tt class="docutils literal"><span class="pre">-help</span></tt> will become categorized if an option category is
+declared. The output looks something like</p>
+<div class="highlight-python"><div class="highlight"><pre>OVERVIEW: This is a small program to demo the LLVM CommandLine API
+USAGE: Sample [options]
+
+OPTIONS:
+
+ General options:
+
+ -help - Display available options (-help-hidden for more)
+ -help-list - Display list of available options (-help-list-hidden for more)
+
+
+ Stage Selection Options:
+ These control which stages are run.
+
+ -E - Run preprocessor stage.
+ -c - Run all stages except linking.
+</pre></div>
+</div>
+<p>In addition to the behaviour of <tt class="docutils literal"><span class="pre">-help</span></tt> changing when an option category is
+declared, the command line option <tt class="docutils literal"><span class="pre">-help-list</span></tt> becomes visible which will
+print the command line options as uncategorized list.</p>
+<p>Note that Options that are not explicitly categorized will be placed in the
+<tt class="docutils literal"><span class="pre">cl::GeneralCategory</span></tt> category.</p>
+</div>
+</div>
+<div class="section" id="reference-guide">
+<span id="id2"></span><h2><a class="toc-backref" href="#id18">Reference Guide</a><a class="headerlink" href="#reference-guide" title="Permalink to this headline">¶</a></h2>
+<p>Now that you know the basics of how to use the CommandLine library, this section
+will give you the detailed information you need to tune how command line options
+work, as well as information on more “advanced” command line option processing
+capabilities.</p>
+<div class="section" id="positional-options">
+<span id="positional-arguments-section"></span><span id="positional-arguments"></span><span id="positional-argument"></span><span id="positional"></span><span id="id3"></span><h3><a class="toc-backref" href="#id19">Positional Arguments</a><a class="headerlink" href="#positional-options" title="Permalink to this headline">¶</a></h3>
+<p>Positional arguments are those arguments that are not named, and are not
+specified with a hyphen. Positional arguments should be used when an option is
+specified by its position alone. For example, the standard Unix <tt class="docutils literal"><span class="pre">grep</span></tt> tool
+takes a regular expression argument, and an optional filename to search through
+(which defaults to standard input if a filename is not specified). Using the
+CommandLine library, this would be specified as:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="n">string</span><span class="o">></span> <span class="n">Regex</span> <span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">Positional</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"<regular expression>"</span><span class="p">),</span> <span class="n">cl</span><span class="o">::</span><span class="n">Required</span><span class="p">);</span>
+<span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="n">string</span><span class="o">></span> <span class="n">Filename</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">Positional</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"<input file>"</span><span class="p">),</span> <span class="n">cl</span><span class="o">::</span><span class="n">init</span><span class="p">(</span><span class="s">"-"</span><span class="p">));</span>
+</pre></div>
+</div>
+<p>Given these two option declarations, the <tt class="docutils literal"><span class="pre">-help</span></tt> output for our grep
+replacement would look like this:</p>
+<div class="highlight-python"><div class="highlight"><pre>USAGE: spiffygrep [options] <regular expression> <input file>
+
+OPTIONS:
+ -help - display available options (-help-hidden for more)
+</pre></div>
+</div>
+<p>... and the resultant program could be used just like the standard <tt class="docutils literal"><span class="pre">grep</span></tt>
+tool.</p>
+<p>Positional arguments are sorted by their order of construction. This means that
+command line options will be ordered according to how they are listed in a .cpp
+file, but will not have an ordering defined if the positional arguments are
+defined in multiple .cpp files. The fix for this problem is simply to define
+all of your positional arguments in one .cpp file.</p>
+<div class="section" id="specifying-positional-options-with-hyphens">
+<h4><a class="toc-backref" href="#id20">Specifying positional options with hyphens</a><a class="headerlink" href="#specifying-positional-options-with-hyphens" title="Permalink to this headline">¶</a></h4>
+<p>Sometimes you may want to specify a value to your positional argument that
+starts with a hyphen (for example, searching for ‘<tt class="docutils literal"><span class="pre">-foo</span></tt>‘ in a file). At
+first, you will have trouble doing this, because it will try to find an argument
+named ‘<tt class="docutils literal"><span class="pre">-foo</span></tt>‘, and will fail (and single quotes will not save you). Note
+that the system <tt class="docutils literal"><span class="pre">grep</span></tt> has the same problem:</p>
+<div class="highlight-python"><div class="highlight"><pre>$ spiffygrep '-foo' test.txt
+Unknown command line argument '-foo'. Try: spiffygrep -help'
+
+$ grep '-foo' test.txt
+grep: illegal option -- f
+grep: illegal option -- o
+grep: illegal option -- o
+Usage: grep -hblcnsviw pattern file . . .
+</pre></div>
+</div>
+<p>The solution for this problem is the same for both your tool and the system
+version: use the ‘<tt class="docutils literal"><span class="pre">--</span></tt>‘ marker. When the user specifies ‘<tt class="docutils literal"><span class="pre">--</span></tt>‘ on the
+command line, it is telling the program that all options after the ‘<tt class="docutils literal"><span class="pre">--</span></tt>‘
+should be treated as positional arguments, not options. Thus, we can use it
+like this:</p>
+<div class="highlight-python"><div class="highlight"><pre>$ spiffygrep -- -foo test.txt
+ ...output...
+</pre></div>
+</div>
+</div>
+<div class="section" id="determining-absolute-position-with-getposition">
+<h4><a class="toc-backref" href="#id21">Determining absolute position with getPosition()</a><a class="headerlink" href="#determining-absolute-position-with-getposition" title="Permalink to this headline">¶</a></h4>
+<p>Sometimes an option can affect or modify the meaning of another option. For
+example, consider <tt class="docutils literal"><span class="pre">gcc</span></tt>‘s <tt class="docutils literal"><span class="pre">-x</span> <span class="pre">LANG</span></tt> option. This tells <tt class="docutils literal"><span class="pre">gcc</span></tt> to ignore the
+suffix of subsequent positional arguments and force the file to be interpreted
+as if it contained source code in language <tt class="docutils literal"><span class="pre">LANG</span></tt>. In order to handle this
+properly, you need to know the absolute position of each argument, especially
+those in lists, so their interaction(s) can be applied correctly. This is also
+useful for options like <tt class="docutils literal"><span class="pre">-llibname</span></tt> which is actually a positional argument
+that starts with a dash.</p>
+<p>So, generally, the problem is that you have two <tt class="docutils literal"><span class="pre">cl::list</span></tt> variables that
+interact in some way. To ensure the correct interaction, you can use the
+<tt class="docutils literal"><span class="pre">cl::list::getPosition(optnum)</span></tt> method. This method returns the absolute
+position (as found on the command line) of the <tt class="docutils literal"><span class="pre">optnum</span></tt> item in the
+<tt class="docutils literal"><span class="pre">cl::list</span></tt>.</p>
+<p>The idiom for usage is like this:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">static</span> <span class="n">cl</span><span class="o">::</span><span class="n">list</span><span class="o"><</span><span class="n">std</span><span class="o">::</span><span class="n">string</span><span class="o">></span> <span class="n">Files</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">Positional</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">OneOrMore</span><span class="p">);</span>
+<span class="k">static</span> <span class="n">cl</span><span class="o">::</span><span class="n">list</span><span class="o"><</span><span class="n">std</span><span class="o">::</span><span class="n">string</span><span class="o">></span> <span class="n">Libraries</span><span class="p">(</span><span class="s">"l"</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">ZeroOrMore</span><span class="p">);</span>
+
+<span class="kt">int</span> <span class="nf">main</span><span class="p">(</span><span class="kt">int</span> <span class="n">argc</span><span class="p">,</span> <span class="kt">char</span><span class="o">**</span><span class="n">argv</span><span class="p">)</span> <span class="p">{</span>
+ <span class="c1">// ...</span>
+ <span class="n">std</span><span class="o">::</span><span class="n">vector</span><span class="o"><</span><span class="n">std</span><span class="o">::</span><span class="n">string</span><span class="o">>::</span><span class="n">iterator</span> <span class="n">fileIt</span> <span class="o">=</span> <span class="n">Files</span><span class="p">.</span><span class="n">begin</span><span class="p">();</span>
+ <span class="n">std</span><span class="o">::</span><span class="n">vector</span><span class="o"><</span><span class="n">std</span><span class="o">::</span><span class="n">string</span><span class="o">>::</span><span class="n">iterator</span> <span class="n">libIt</span> <span class="o">=</span> <span class="n">Libraries</span><span class="p">.</span><span class="n">begin</span><span class="p">();</span>
+ <span class="kt">unsigned</span> <span class="n">libPos</span> <span class="o">=</span> <span class="mi">0</span><span class="p">,</span> <span class="n">filePos</span> <span class="o">=</span> <span class="mi">0</span><span class="p">;</span>
+ <span class="k">while</span> <span class="p">(</span> <span class="mi">1</span> <span class="p">)</span> <span class="p">{</span>
+ <span class="k">if</span> <span class="p">(</span> <span class="n">libIt</span> <span class="o">!=</span> <span class="n">Libraries</span><span class="p">.</span><span class="n">end</span><span class="p">()</span> <span class="p">)</span>
+ <span class="n">libPos</span> <span class="o">=</span> <span class="n">Libraries</span><span class="p">.</span><span class="n">getPosition</span><span class="p">(</span> <span class="n">libIt</span> <span class="o">-</span> <span class="n">Libraries</span><span class="p">.</span><span class="n">begin</span><span class="p">()</span> <span class="p">);</span>
+ <span class="k">else</span>
+ <span class="n">libPos</span> <span class="o">=</span> <span class="mi">0</span><span class="p">;</span>
+ <span class="k">if</span> <span class="p">(</span> <span class="n">fileIt</span> <span class="o">!=</span> <span class="n">Files</span><span class="p">.</span><span class="n">end</span><span class="p">()</span> <span class="p">)</span>
+ <span class="n">filePos</span> <span class="o">=</span> <span class="n">Files</span><span class="p">.</span><span class="n">getPosition</span><span class="p">(</span> <span class="n">fileIt</span> <span class="o">-</span> <span class="n">Files</span><span class="p">.</span><span class="n">begin</span><span class="p">()</span> <span class="p">);</span>
+ <span class="k">else</span>
+ <span class="n">filePos</span> <span class="o">=</span> <span class="mi">0</span><span class="p">;</span>
+
+ <span class="k">if</span> <span class="p">(</span> <span class="n">filePos</span> <span class="o">!=</span> <span class="mi">0</span> <span class="o">&&</span> <span class="p">(</span><span class="n">libPos</span> <span class="o">==</span> <span class="mi">0</span> <span class="o">||</span> <span class="n">filePos</span> <span class="o"><</span> <span class="n">libPos</span><span class="p">)</span> <span class="p">)</span> <span class="p">{</span>
+ <span class="c1">// Source File Is next</span>
+ <span class="o">++</span><span class="n">fileIt</span><span class="p">;</span>
+ <span class="p">}</span>
+ <span class="k">else</span> <span class="k">if</span> <span class="p">(</span> <span class="n">libPos</span> <span class="o">!=</span> <span class="mi">0</span> <span class="o">&&</span> <span class="p">(</span><span class="n">filePos</span> <span class="o">==</span> <span class="mi">0</span> <span class="o">||</span> <span class="n">libPos</span> <span class="o"><</span> <span class="n">filePos</span><span class="p">)</span> <span class="p">)</span> <span class="p">{</span>
+ <span class="c1">// Library is next</span>
+ <span class="o">++</span><span class="n">libIt</span><span class="p">;</span>
+ <span class="p">}</span>
+ <span class="k">else</span>
+ <span class="k">break</span><span class="p">;</span> <span class="c1">// we're done with the list</span>
+ <span class="p">}</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>Note that, for compatibility reasons, the <tt class="docutils literal"><span class="pre">cl::opt</span></tt> also supports an
+<tt class="docutils literal"><span class="pre">unsigned</span> <span class="pre">getPosition()</span></tt> option that will provide the absolute position of
+that option. You can apply the same approach as above with a <tt class="docutils literal"><span class="pre">cl::opt</span></tt> and a
+<tt class="docutils literal"><span class="pre">cl::list</span></tt> option as you can with two lists.</p>
+</div>
+<div class="section" id="the-cl-consumeafter-modifier">
+<span id="this-section-for-more-information"></span><span id="cl-consumeafter"></span><span id="interpreter-style-options"></span><h4><a class="toc-backref" href="#id22">The <tt class="docutils literal"><span class="pre">cl::ConsumeAfter</span></tt> modifier</a><a class="headerlink" href="#the-cl-consumeafter-modifier" title="Permalink to this headline">¶</a></h4>
+<p>The <tt class="docutils literal"><span class="pre">cl::ConsumeAfter</span></tt> <a class="reference internal" href="#formatting-option">formatting option</a> is used to construct programs that
+use “interpreter style” option processing. With this style of option
+processing, all arguments specified after the last positional argument are
+treated as special interpreter arguments that are not interpreted by the command
+line argument.</p>
+<p>As a concrete example, lets say we are developing a replacement for the standard
+Unix Bourne shell (<tt class="docutils literal"><span class="pre">/bin/sh</span></tt>). To run <tt class="docutils literal"><span class="pre">/bin/sh</span></tt>, first you specify options
+to the shell itself (like <tt class="docutils literal"><span class="pre">-x</span></tt> which turns on trace output), then you specify
+the name of the script to run, then you specify arguments to the script. These
+arguments to the script are parsed by the Bourne shell command line option
+processor, but are not interpreted as options to the shell itself. Using the
+CommandLine library, we would specify this as:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="n">string</span><span class="o">></span> <span class="n">Script</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">Positional</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"<input script>"</span><span class="p">),</span> <span class="n">cl</span><span class="o">::</span><span class="n">init</span><span class="p">(</span><span class="s">"-"</span><span class="p">));</span>
+<span class="n">cl</span><span class="o">::</span><span class="n">list</span><span class="o"><</span><span class="n">string</span><span class="o">></span> <span class="n">Argv</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">ConsumeAfter</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"<program arguments>..."</span><span class="p">));</span>
+<span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="kt">bool</span><span class="o">></span> <span class="n">Trace</span><span class="p">(</span><span class="s">"x"</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Enable trace output"</span><span class="p">));</span>
+</pre></div>
+</div>
+<p>which automatically provides the help output:</p>
+<div class="highlight-python"><div class="highlight"><pre>USAGE: spiffysh [options] <input script> <program arguments>...
+
+OPTIONS:
+ -help - display available options (-help-hidden for more)
+ -x - Enable trace output
+</pre></div>
+</div>
+<p>At runtime, if we run our new shell replacement as <tt class="docutils literal"><span class="pre">`spiffysh</span> <span class="pre">-x</span> <span class="pre">test.sh</span> <span class="pre">-a</span> <span class="pre">-x</span>
+<span class="pre">-y</span> <span class="pre">bar</span></tt>‘, the <tt class="docutils literal"><span class="pre">Trace</span></tt> variable will be set to true, the <tt class="docutils literal"><span class="pre">Script</span></tt> variable
+will be set to “<tt class="docutils literal"><span class="pre">test.sh</span></tt>”, and the <tt class="docutils literal"><span class="pre">Argv</span></tt> list will contain <tt class="docutils literal"><span class="pre">["-a",</span> <span class="pre">"-x",</span>
+<span class="pre">"-y",</span> <span class="pre">"bar"]</span></tt>, because they were specified after the last positional argument
+(which is the script name).</p>
+<p>There are several limitations to when <tt class="docutils literal"><span class="pre">cl::ConsumeAfter</span></tt> options can be
+specified. For example, only one <tt class="docutils literal"><span class="pre">cl::ConsumeAfter</span></tt> can be specified per
+program, there must be at least one <a class="reference internal" href="#positional-argument">positional argument</a> specified, there must
+not be any <a class="reference internal" href="#cl-list">cl::list</a> positional arguments, and the <tt class="docutils literal"><span class="pre">cl::ConsumeAfter</span></tt> option
+should be a <a class="reference internal" href="#cl-list">cl::list</a> option.</p>
+</div>
+</div>
+<div class="section" id="internal-vs-external-storage">
+<span id="can-be-changed"></span><span id="id4"></span><h3><a class="toc-backref" href="#id23">Internal vs External Storage</a><a class="headerlink" href="#internal-vs-external-storage" title="Permalink to this headline">¶</a></h3>
+<p>By default, all command line options automatically hold the value that they
+parse from the command line. This is very convenient in the common case,
+especially when combined with the ability to define command line options in the
+files that use them. This is called the internal storage model.</p>
+<p>Sometimes, however, it is nice to separate the command line option processing
+code from the storage of the value parsed. For example, lets say that we have a
+‘<tt class="docutils literal"><span class="pre">-debug</span></tt>‘ option that we would like to use to enable debug information across
+the entire body of our program. In this case, the boolean value controlling the
+debug code should be globally accessible (in a header file, for example) yet the
+command line option processing code should not be exposed to all of these
+clients (requiring lots of .cpp files to <tt class="docutils literal"><span class="pre">#include</span> <span class="pre">CommandLine.h</span></tt>).</p>
+<p>To do this, set up your .h file with your option, like this for example:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="c1">// DebugFlag.h - Get access to the '-debug' command line option</span>
+<span class="c1">//</span>
+
+<span class="c1">// DebugFlag - This boolean is set to true if the '-debug' command line option</span>
+<span class="c1">// is specified. This should probably not be referenced directly, instead, use</span>
+<span class="c1">// the DEBUG macro below.</span>
+<span class="c1">//</span>
+<span class="k">extern</span> <span class="kt">bool</span> <span class="n">DebugFlag</span><span class="p">;</span>
+
+<span class="c1">// DEBUG macro - This macro should be used by code to emit debug information.</span>
+<span class="c1">// In the '-debug' option is specified on the command line, and if this is a</span>
+<span class="c1">// debug build, then the code specified as the option to the macro will be</span>
+<span class="c1">// executed. Otherwise it will not be.</span>
+<span class="cp">#ifdef NDEBUG</span>
+<span class="cp">#define DEBUG(X)</span>
+<span class="cp">#else</span>
+<span class="cp">#define DEBUG(X) do { if (DebugFlag) { X; } } while (0)</span>
+<span class="cp">#endif</span>
+</pre></div>
+</div>
+<p>This allows clients to blissfully use the <tt class="docutils literal"><span class="pre">DEBUG()</span></tt> macro, or the
+<tt class="docutils literal"><span class="pre">DebugFlag</span></tt> explicitly if they want to. Now we just need to be able to set
+the <tt class="docutils literal"><span class="pre">DebugFlag</span></tt> boolean when the option is set. To do this, we pass an
+additional argument to our command line argument processor, and we specify where
+to fill in with the <a class="reference internal" href="#cl-location">cl::location</a> attribute:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="kt">bool</span> <span class="n">DebugFlag</span><span class="p">;</span> <span class="c1">// the actual value</span>
+<span class="k">static</span> <span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="kt">bool</span><span class="p">,</span> <span class="nb">true</span><span class="o">></span> <span class="c1">// The parser</span>
+<span class="n">Debug</span><span class="p">(</span><span class="s">"debug"</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Enable debug output"</span><span class="p">),</span> <span class="n">cl</span><span class="o">::</span><span class="n">Hidden</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">location</span><span class="p">(</span><span class="n">DebugFlag</span><span class="p">));</span>
+</pre></div>
+</div>
+<p>In the above example, we specify “<tt class="docutils literal"><span class="pre">true</span></tt>” as the second argument to the
+<a class="reference internal" href="#cl-opt">cl::opt</a> template, indicating that the template should not maintain a copy of
+the value itself. In addition to this, we specify the <a class="reference internal" href="#cl-location">cl::location</a>
+attribute, so that <tt class="docutils literal"><span class="pre">DebugFlag</span></tt> is automatically set.</p>
+</div>
+<div class="section" id="option-attributes">
+<h3><a class="toc-backref" href="#id24">Option Attributes</a><a class="headerlink" href="#option-attributes" title="Permalink to this headline">¶</a></h3>
+<p>This section describes the basic attributes that you can specify on options.</p>
+<ul>
+<li><p class="first">The option name attribute (which is required for all options, except
+<a class="reference internal" href="#positional-options">positional options</a>) specifies what the option name is. This option is
+specified in simple double quotes:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="kt">bool</span><span class="o">></span> <span class="n">Quiet</span><span class="p">(</span><span class="s">"quiet"</span><span class="p">);</span>
+</pre></div>
+</div>
+</li>
+</ul>
+<ul class="simple" id="cl-desc">
+<li>The <strong>cl::desc</strong> attribute specifies a description for the option to be
+shown in the <tt class="docutils literal"><span class="pre">-help</span></tt> output for the program. This attribute supports
+multi-line descriptions with lines separated by ‘n’.</li>
+</ul>
+<ul class="simple" id="cl-value-desc">
+<li>The <strong>cl::value_desc</strong> attribute specifies a string that can be used to
+fine tune the <tt class="docutils literal"><span class="pre">-help</span></tt> output for a command line option. Look <a class="reference internal" href="#here">here</a> for an
+example.</li>
+</ul>
+<ul id="cl-init">
+<li><p class="first">The <strong>cl::init</strong> attribute specifies an initial value for a <a class="reference internal" href="#scalar">scalar</a>
+option. If this attribute is not specified then the command line option value
+defaults to the value created by the default constructor for the
+type.</p>
+<div class="admonition warning">
+<p class="first admonition-title">Warning</p>
+<p class="last">If you specify both <strong>cl::init</strong> and <strong>cl::location</strong> for an option, you
+must specify <strong>cl::location</strong> first, so that when the command-line parser
+sees <strong>cl::init</strong>, it knows where to put the initial value. (You will get an
+error at runtime if you don’t put them in the right order.)</p>
+</div>
+</li>
+</ul>
+<ul class="simple" id="cl-location">
+<li>The <strong>cl::location</strong> attribute where to store the value for a parsed command
+line option if using external storage. See the section on <a class="reference internal" href="#internal-vs-external-storage">Internal vs
+External Storage</a> for more information.</li>
+</ul>
+<ul class="simple" id="cl-aliasopt">
+<li>The <strong>cl::aliasopt</strong> attribute specifies which option a <a class="reference internal" href="#cl-alias">cl::alias</a> option is
+an alias for.</li>
+</ul>
+<ul id="cl-values">
+<li><p class="first">The <strong>cl::values</strong> attribute specifies the string-to-value mapping to be used
+by the generic parser. It takes a list of (option, value, description)
+triplets that specify the option name, the value mapped to, and the
+description shown in the <tt class="docutils literal"><span class="pre">-help</span></tt> for the tool. Because the generic parser
+is used most frequently with enum values, two macros are often useful:</p>
+<ol class="arabic simple">
+<li>The <strong>clEnumVal</strong> macro is used as a nice simple way to specify a triplet
+for an enum. This macro automatically makes the option name be the same as
+the enum name. The first option to the macro is the enum, the second is
+the description for the command line option.</li>
+<li>The <strong>clEnumValN</strong> macro is used to specify macro options where the option
+name doesn’t equal the enum name. For this macro, the first argument is
+the enum value, the second is the flag name, and the second is the
+description.</li>
+</ol>
+<p>You will get a compile time error if you try to use cl::values with a parser
+that does not support it.</p>
+</li>
+</ul>
+<ul class="simple" id="cl-multi-val">
+<li>The <strong>cl::multi_val</strong> attribute specifies that this option takes has multiple
+values (example: <tt class="docutils literal"><span class="pre">-sectalign</span> <span class="pre">segname</span> <span class="pre">sectname</span> <span class="pre">sectvalue</span></tt>). This attribute
+takes one unsigned argument - the number of values for the option. This
+attribute is valid only on <tt class="docutils literal"><span class="pre">cl::list</span></tt> options (and will fail with compile
+error if you try to use it with other option types). It is allowed to use all
+of the usual modifiers on multi-valued options (besides
+<tt class="docutils literal"><span class="pre">cl::ValueDisallowed</span></tt>, obviously).</li>
+</ul>
+<ul class="simple" id="cl-cat">
+<li>The <strong>cl::cat</strong> attribute specifies the option category that the option
+belongs to. The category should be a <a class="reference internal" href="#cl-optioncategory">cl::OptionCategory</a> object.</li>
+</ul>
+</div>
+<div class="section" id="option-modifiers">
+<h3><a class="toc-backref" href="#id25">Option Modifiers</a><a class="headerlink" href="#option-modifiers" title="Permalink to this headline">¶</a></h3>
+<p>Option modifiers are the flags and expressions that you pass into the
+constructors for <a class="reference internal" href="#cl-opt">cl::opt</a> and <a class="reference internal" href="#cl-list">cl::list</a>. These modifiers give you the
+ability to tweak how options are parsed and how <tt class="docutils literal"><span class="pre">-help</span></tt> output is generated to
+fit your application well.</p>
+<p>These options fall into five main categories:</p>
+<ol class="arabic simple">
+<li>Hiding an option from <tt class="docutils literal"><span class="pre">-help</span></tt> output</li>
+<li>Controlling the number of occurrences required and allowed</li>
+<li>Controlling whether or not a value must be specified</li>
+<li>Controlling other formatting options</li>
+<li>Miscellaneous option modifiers</li>
+</ol>
+<p>It is not possible to specify two options from the same category (you’ll get a
+runtime error) to a single option, except for options in the miscellaneous
+category. The CommandLine library specifies defaults for all of these settings
+that are the most useful in practice and the most common, which mean that you
+usually shouldn’t have to worry about these.</p>
+<div class="section" id="hiding-an-option-from-help-output">
+<h4><a class="toc-backref" href="#id26">Hiding an option from <tt class="docutils literal"><span class="pre">-help</span></tt> output</a><a class="headerlink" href="#hiding-an-option-from-help-output" title="Permalink to this headline">¶</a></h4>
+<p>The <tt class="docutils literal"><span class="pre">cl::NotHidden</span></tt>, <tt class="docutils literal"><span class="pre">cl::Hidden</span></tt>, and <tt class="docutils literal"><span class="pre">cl::ReallyHidden</span></tt> modifiers are
+used to control whether or not an option appears in the <tt class="docutils literal"><span class="pre">-help</span></tt> and
+<tt class="docutils literal"><span class="pre">-help-hidden</span></tt> output for the compiled program:</p>
+<ul class="simple" id="cl-nothidden">
+<li>The <strong>cl::NotHidden</strong> modifier (which is the default for <a class="reference internal" href="#cl-opt">cl::opt</a> and
+<a class="reference internal" href="#cl-list">cl::list</a> options) indicates the option is to appear in both help
+listings.</li>
+</ul>
+<ul class="simple" id="cl-hidden">
+<li>The <strong>cl::Hidden</strong> modifier (which is the default for <a class="reference internal" href="#cl-alias">cl::alias</a> options)
+indicates that the option should not appear in the <tt class="docutils literal"><span class="pre">-help</span></tt> output, but
+should appear in the <tt class="docutils literal"><span class="pre">-help-hidden</span></tt> output.</li>
+</ul>
+<ul class="simple" id="cl-reallyhidden">
+<li>The <strong>cl::ReallyHidden</strong> modifier indicates that the option should not appear
+in any help output.</li>
+</ul>
+</div>
+<div class="section" id="controlling-the-number-of-occurrences-required-and-allowed">
+<h4><a class="toc-backref" href="#id27">Controlling the number of occurrences required and allowed</a><a class="headerlink" href="#controlling-the-number-of-occurrences-required-and-allowed" title="Permalink to this headline">¶</a></h4>
+<p>This group of options is used to control how many time an option is allowed (or
+required) to be specified on the command line of your program. Specifying a
+value for this setting allows the CommandLine library to do error checking for
+you.</p>
+<p>The allowed values for this option group are:</p>
+<ul class="simple" id="cl-optional">
+<li>The <strong>cl::Optional</strong> modifier (which is the default for the <a class="reference internal" href="#cl-opt">cl::opt</a> and
+<a class="reference internal" href="#cl-alias">cl::alias</a> classes) indicates that your program will allow either zero or
+one occurrence of the option to be specified.</li>
+</ul>
+<ul class="simple" id="cl-zeroormore">
+<li>The <strong>cl::ZeroOrMore</strong> modifier (which is the default for the <a class="reference internal" href="#cl-list">cl::list</a>
+class) indicates that your program will allow the option to be specified zero
+or more times.</li>
+</ul>
+<ul class="simple" id="cl-required">
+<li>The <strong>cl::Required</strong> modifier indicates that the specified option must be
+specified exactly one time.</li>
+</ul>
+<ul class="simple" id="cl-oneormore">
+<li>The <strong>cl::OneOrMore</strong> modifier indicates that the option must be specified at
+least one time.</li>
+<li>The <strong>cl::ConsumeAfter</strong> modifier is described in the <a class="reference internal" href="#positional-arguments-section">Positional arguments
+section</a>.</li>
+</ul>
+<p>If an option is not specified, then the value of the option is equal to the
+value specified by the <a class="reference internal" href="#cl-init">cl::init</a> attribute. If the <tt class="docutils literal"><span class="pre">cl::init</span></tt> attribute is
+not specified, the option value is initialized with the default constructor for
+the data type.</p>
+<p>If an option is specified multiple times for an option of the <a class="reference internal" href="#cl-opt">cl::opt</a> class,
+only the last value will be retained.</p>
+</div>
+<div class="section" id="controlling-whether-or-not-a-value-must-be-specified">
+<h4><a class="toc-backref" href="#id28">Controlling whether or not a value must be specified</a><a class="headerlink" href="#controlling-whether-or-not-a-value-must-be-specified" title="Permalink to this headline">¶</a></h4>
+<p>This group of options is used to control whether or not the option allows a
+value to be present. In the case of the CommandLine library, a value is either
+specified with an equal sign (e.g. ‘<tt class="docutils literal"><span class="pre">-index-depth=17</span></tt>‘) or as a trailing
+string (e.g. ‘<tt class="docutils literal"><span class="pre">-o</span> <span class="pre">a.out</span></tt>‘).</p>
+<p>The allowed values for this option group are:</p>
+<ul class="simple" id="cl-valueoptional">
+<li>The <strong>cl::ValueOptional</strong> modifier (which is the default for <tt class="docutils literal"><span class="pre">bool</span></tt> typed
+options) specifies that it is acceptable to have a value, or not. A boolean
+argument can be enabled just by appearing on the command line, or it can have
+an explicit ‘<tt class="docutils literal"><span class="pre">-foo=true</span></tt>‘. If an option is specified with this mode, it is
+illegal for the value to be provided without the equal sign. Therefore
+‘<tt class="docutils literal"><span class="pre">-foo</span> <span class="pre">true</span></tt>‘ is illegal. To get this behavior, you must use
+the <a class="reference internal" href="#cl-valuerequired">cl::ValueRequired</a> modifier.</li>
+</ul>
+<ul class="simple" id="cl-valuerequired">
+<li>The <strong>cl::ValueRequired</strong> modifier (which is the default for all other types
+except for <a class="reference internal" href="#unnamed-alternatives-using-the-generic-parser">unnamed alternatives using the generic parser</a>) specifies that a
+value must be provided. This mode informs the command line library that if an
+option is not provides with an equal sign, that the next argument provided
+must be the value. This allows things like ‘<tt class="docutils literal"><span class="pre">-o</span> <span class="pre">a.out</span></tt>‘ to work.</li>
+</ul>
+<ul class="simple" id="cl-valuedisallowed">
+<li>The <strong>cl::ValueDisallowed</strong> modifier (which is the default for <a class="reference internal" href="#unnamed-alternatives-using-the-generic-parser">unnamed
+alternatives using the generic parser</a>) indicates that it is a runtime error
+for the user to specify a value. This can be provided to disallow users from
+providing options to boolean options (like ‘<tt class="docutils literal"><span class="pre">-foo=true</span></tt>‘).</li>
+</ul>
+<p>In general, the default values for this option group work just like you would
+want them to. As mentioned above, you can specify the <a class="reference internal" href="#cl-valuedisallowed">cl::ValueDisallowed</a>
+modifier to a boolean argument to restrict your command line parser. These
+options are mostly useful when <a class="reference internal" href="#extending-the-library">extending the library</a>.</p>
+</div>
+<div class="section" id="controlling-other-formatting-options">
+<span id="formatting-option"></span><h4><a class="toc-backref" href="#id29">Controlling other formatting options</a><a class="headerlink" href="#controlling-other-formatting-options" title="Permalink to this headline">¶</a></h4>
+<p>The formatting option group is used to specify that the command line option has
+special abilities and is otherwise different from other command line arguments.
+As usual, you can only specify one of these arguments at most.</p>
+<ul class="simple" id="cl-normalformatting">
+<li>The <strong>cl::NormalFormatting</strong> modifier (which is the default all options)
+specifies that this option is “normal”.</li>
+</ul>
+<ul class="simple" id="cl-positional">
+<li>The <strong>cl::Positional</strong> modifier specifies that this is a positional argument
+that does not have a command line option associated with it. See the
+<a class="reference internal" href="#positional-arguments">Positional Arguments</a> section for more information.</li>
+<li>The <strong>cl::ConsumeAfter</strong> modifier specifies that this option is used to
+capture “interpreter style” arguments. See <a class="reference internal" href="#this-section-for-more-information">this section for more
+information</a>.</li>
+</ul>
+<ul class="simple" id="cl-prefix">
+<span id="prefix"></span><li>The <strong>cl::Prefix</strong> modifier specifies that this option prefixes its value.
+With ‘Prefix’ options, the equal sign does not separate the value from the
+option name specified. Instead, the value is everything after the prefix,
+including any equal sign if present. This is useful for processing odd
+arguments like <tt class="docutils literal"><span class="pre">-lmalloc</span></tt> and <tt class="docutils literal"><span class="pre">-L/usr/lib</span></tt> in a linker tool or
+<tt class="docutils literal"><span class="pre">-DNAME=value</span></tt> in a compiler tool. Here, the ‘<tt class="docutils literal"><span class="pre">l</span></tt>‘, ‘<tt class="docutils literal"><span class="pre">D</span></tt>‘ and ‘<tt class="docutils literal"><span class="pre">L</span></tt>‘
+options are normal string (or list) options, that have the <strong>cl::Prefix</strong>
+modifier added to allow the CommandLine library to recognize them. Note that
+<strong>cl::Prefix</strong> options must not have the <strong>cl::ValueDisallowed</strong> modifier
+specified.</li>
+</ul>
+<ul class="simple" id="cl-grouping">
+<span id="grouping"></span><li>The <strong>cl::Grouping</strong> modifier is used to implement Unix-style tools (like
+<tt class="docutils literal"><span class="pre">ls</span></tt>) that have lots of single letter arguments, but only require a single
+dash. For example, the ‘<tt class="docutils literal"><span class="pre">ls</span> <span class="pre">-labF</span></tt>‘ command actually enables four different
+options, all of which are single letters. Note that <strong>cl::Grouping</strong> options
+cannot have values.</li>
+</ul>
+<p>The CommandLine library does not restrict how you use the <strong>cl::Prefix</strong> or
+<strong>cl::Grouping</strong> modifiers, but it is possible to specify ambiguous argument
+settings. Thus, it is possible to have multiple letter options that are prefix
+or grouping options, and they will still work as designed.</p>
+<p>To do this, the CommandLine library uses a greedy algorithm to parse the input
+option into (potentially multiple) prefix and grouping options. The strategy
+basically looks like this:</p>
+<div class="highlight-python"><div class="highlight"><pre>parse(string OrigInput) {
+
+1. string input = OrigInput;
+2. if (isOption(input)) return getOption(input).parse(); // Normal option
+3. while (!isOption(input) && !input.empty()) input.pop_back(); // Remove the last letter
+4. if (input.empty()) return error(); // No matching option
+5. if (getOption(input).isPrefix())
+ return getOption(input).parse(input);
+6. while (!input.empty()) { // Must be grouping options
+ getOption(input).parse();
+ OrigInput.erase(OrigInput.begin(), OrigInput.begin()+input.length());
+ input = OrigInput;
+ while (!isOption(input) && !input.empty()) input.pop_back();
+ }
+7. if (!OrigInput.empty()) error();
+
+}
+</pre></div>
+</div>
+</div>
+<div class="section" id="miscellaneous-option-modifiers">
+<h4><a class="toc-backref" href="#id30">Miscellaneous option modifiers</a><a class="headerlink" href="#miscellaneous-option-modifiers" title="Permalink to this headline">¶</a></h4>
+<p>The miscellaneous option modifiers are the only flags where you can specify more
+than one flag from the set: they are not mutually exclusive. These flags
+specify boolean properties that modify the option.</p>
+<ul class="simple" id="cl-commaseparated">
+<li>The <strong>cl::CommaSeparated</strong> modifier indicates that any commas specified for an
+option’s value should be used to split the value up into multiple values for
+the option. For example, these two options are equivalent when
+<tt class="docutils literal"><span class="pre">cl::CommaSeparated</span></tt> is specified: “<tt class="docutils literal"><span class="pre">-foo=a</span> <span class="pre">-foo=b</span> <span class="pre">-foo=c</span></tt>” and
+“<tt class="docutils literal"><span class="pre">-foo=a,b,c</span></tt>”. This option only makes sense to be used in a case where the
+option is allowed to accept one or more values (i.e. it is a <a class="reference internal" href="#cl-list">cl::list</a>
+option).</li>
+</ul>
+<ul class="simple" id="cl-positionaleatsargs">
+<li>The <strong>cl::PositionalEatsArgs</strong> modifier (which only applies to positional
+arguments, and only makes sense for lists) indicates that positional argument
+should consume any strings after it (including strings that start with a “-”)
+up until another recognized positional argument. For example, if you have two
+“eating” positional arguments, “<tt class="docutils literal"><span class="pre">pos1</span></tt>” and “<tt class="docutils literal"><span class="pre">pos2</span></tt>”, the string “<tt class="docutils literal"><span class="pre">-pos1</span>
+<span class="pre">-foo</span> <span class="pre">-bar</span> <span class="pre">baz</span> <span class="pre">-pos2</span> <span class="pre">-bork</span></tt>” would cause the “<tt class="docutils literal"><span class="pre">-foo</span> <span class="pre">-bar</span> <span class="pre">-baz</span></tt>” strings to
+be applied to the “<tt class="docutils literal"><span class="pre">-pos1</span></tt>” option and the “<tt class="docutils literal"><span class="pre">-bork</span></tt>” string to be applied
+to the “<tt class="docutils literal"><span class="pre">-pos2</span></tt>” option.</li>
+</ul>
+<ul class="simple" id="cl-sink">
+<li>The <strong>cl::Sink</strong> modifier is used to handle unknown options. If there is at
+least one option with <tt class="docutils literal"><span class="pre">cl::Sink</span></tt> modifier specified, the parser passes
+unrecognized option strings to it as values instead of signaling an error. As
+with <tt class="docutils literal"><span class="pre">cl::CommaSeparated</span></tt>, this modifier only makes sense with a <a class="reference internal" href="#cl-list">cl::list</a>
+option.</li>
+</ul>
+<p>So far, these are the only three miscellaneous option modifiers.</p>
+</div>
+<div class="section" id="response-files">
+<span id="id5"></span><h4><a class="toc-backref" href="#id31">Response files</a><a class="headerlink" href="#response-files" title="Permalink to this headline">¶</a></h4>
+<p>Some systems, such as certain variants of Microsoft Windows and some older
+Unices have a relatively low limit on command-line length. It is therefore
+customary to use the so-called ‘response files’ to circumvent this
+restriction. These files are mentioned on the command-line (using the “@file”)
+syntax. The program reads these files and inserts the contents into argv,
+thereby working around the command-line length limits. Response files are
+enabled by an optional fourth argument to <a class="reference internal" href="#cl-parseenvironmentoptions">cl::ParseEnvironmentOptions</a> and
+<a class="reference internal" href="#cl-parsecommandlineoptions">cl::ParseCommandLineOptions</a>.</p>
+</div>
+</div>
+<div class="section" id="top-level-classes-and-functions">
+<h3><a class="toc-backref" href="#id32">Top-Level Classes and Functions</a><a class="headerlink" href="#top-level-classes-and-functions" title="Permalink to this headline">¶</a></h3>
+<p>Despite all of the built-in flexibility, the CommandLine option library really
+only consists of one function <a class="reference internal" href="#cl-parsecommandlineoptions">cl::ParseCommandLineOptions</a>) and three main
+classes: <a class="reference internal" href="#cl-opt">cl::opt</a>, <a class="reference internal" href="#cl-list">cl::list</a>, and <a class="reference internal" href="#cl-alias">cl::alias</a>. This section describes
+these three classes in detail.</p>
+<div class="section" id="the-cl-getregisteredoptions-function">
+<span id="cl-getregisteredoptions"></span><h4><a class="toc-backref" href="#id33">The <tt class="docutils literal"><span class="pre">cl::getRegisteredOptions</span></tt> function</a><a class="headerlink" href="#the-cl-getregisteredoptions-function" title="Permalink to this headline">¶</a></h4>
+<p>The <tt class="docutils literal"><span class="pre">cl::getRegisteredOptions</span></tt> function is designed to give a programmer
+access to declared non-positional command line options so that how they appear
+in <tt class="docutils literal"><span class="pre">-help</span></tt> can be modified prior to calling <a class="reference internal" href="#cl-parsecommandlineoptions">cl::ParseCommandLineOptions</a>.
+Note this method should not be called during any static initialisation because
+it cannot be guaranteed that all options will have been initialised. Hence it
+should be called from <tt class="docutils literal"><span class="pre">main</span></tt>.</p>
+<p>This function can be used to gain access to options declared in libraries that
+the tool writter may not have direct access to.</p>
+<p>The function retrieves a <a class="reference internal" href="ProgrammersManual.html#dss-stringmap"><em>StringMap</em></a> that maps the option
+string (e.g. <tt class="docutils literal"><span class="pre">-help</span></tt>) to an <tt class="docutils literal"><span class="pre">Option*</span></tt>.</p>
+<p>Here is an example of how the function could be used:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">using</span> <span class="k">namespace</span> <span class="n">llvm</span><span class="p">;</span>
+<span class="kt">int</span> <span class="nf">main</span><span class="p">(</span><span class="kt">int</span> <span class="n">argc</span><span class="p">,</span> <span class="kt">char</span> <span class="o">**</span><span class="n">argv</span><span class="p">)</span> <span class="p">{</span>
+ <span class="n">cl</span><span class="o">::</span><span class="n">OptionCategory</span> <span class="n">AnotherCategory</span><span class="p">(</span><span class="s">"Some options"</span><span class="p">);</span>
+
+ <span class="n">StringMap</span><span class="o"><</span><span class="n">cl</span><span class="o">::</span><span class="n">Option</span><span class="o">*></span> <span class="o">&</span><span class="n">Map</span> <span class="o">=</span> <span class="n">cl</span><span class="o">::</span><span class="n">getRegisteredOptions</span><span class="p">();</span>
+
+ <span class="c1">//Unhide useful option and put it in a different category</span>
+ <span class="n">assert</span><span class="p">(</span><span class="n">Map</span><span class="p">.</span><span class="n">count</span><span class="p">(</span><span class="s">"print-all-options"</span><span class="p">)</span> <span class="o">></span> <span class="mi">0</span><span class="p">);</span>
+ <span class="n">Map</span><span class="p">[</span><span class="s">"print-all-options"</span><span class="p">]</span><span class="o">-></span><span class="n">setHiddenFlag</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">NotHidden</span><span class="p">);</span>
+ <span class="n">Map</span><span class="p">[</span><span class="s">"print-all-options"</span><span class="p">]</span><span class="o">-></span><span class="n">setCategory</span><span class="p">(</span><span class="n">AnotherCategory</span><span class="p">);</span>
+
+ <span class="c1">//Hide an option we don't want to see</span>
+ <span class="n">assert</span><span class="p">(</span><span class="n">Map</span><span class="p">.</span><span class="n">count</span><span class="p">(</span><span class="s">"enable-no-infs-fp-math"</span><span class="p">)</span> <span class="o">></span> <span class="mi">0</span><span class="p">);</span>
+ <span class="n">Map</span><span class="p">[</span><span class="s">"enable-no-infs-fp-math"</span><span class="p">]</span><span class="o">-></span><span class="n">setHiddenFlag</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">Hidden</span><span class="p">);</span>
+
+ <span class="c1">//Change --version to --show-version</span>
+ <span class="n">assert</span><span class="p">(</span><span class="n">Map</span><span class="p">.</span><span class="n">count</span><span class="p">(</span><span class="s">"version"</span><span class="p">)</span> <span class="o">></span> <span class="mi">0</span><span class="p">);</span>
+ <span class="n">Map</span><span class="p">[</span><span class="s">"version"</span><span class="p">]</span><span class="o">-></span><span class="n">setArgStr</span><span class="p">(</span><span class="s">"show-version"</span><span class="p">);</span>
+
+ <span class="c1">//Change --help description</span>
+ <span class="n">assert</span><span class="p">(</span><span class="n">Map</span><span class="p">.</span><span class="n">count</span><span class="p">(</span><span class="s">"help"</span><span class="p">)</span> <span class="o">></span> <span class="mi">0</span><span class="p">);</span>
+ <span class="n">Map</span><span class="p">[</span><span class="s">"help"</span><span class="p">]</span><span class="o">-></span><span class="n">setDescription</span><span class="p">(</span><span class="s">"Shows help"</span><span class="p">);</span>
+
+ <span class="n">cl</span><span class="o">::</span><span class="n">ParseCommandLineOptions</span><span class="p">(</span><span class="n">argc</span><span class="p">,</span> <span class="n">argv</span><span class="p">,</span> <span class="s">"This is a small program to demo the LLVM CommandLine API"</span><span class="p">);</span>
+ <span class="p">...</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+</div>
+<div class="section" id="the-cl-parsecommandlineoptions-function">
+<span id="cl-parsecommandlineoptions"></span><h4><a class="toc-backref" href="#id34">The <tt class="docutils literal"><span class="pre">cl::ParseCommandLineOptions</span></tt> function</a><a class="headerlink" href="#the-cl-parsecommandlineoptions-function" title="Permalink to this headline">¶</a></h4>
+<p>The <tt class="docutils literal"><span class="pre">cl::ParseCommandLineOptions</span></tt> function is designed to be called directly
+from <tt class="docutils literal"><span class="pre">main</span></tt>, and is used to fill in the values of all of the command line
+option variables once <tt class="docutils literal"><span class="pre">argc</span></tt> and <tt class="docutils literal"><span class="pre">argv</span></tt> are available.</p>
+<p>The <tt class="docutils literal"><span class="pre">cl::ParseCommandLineOptions</span></tt> function requires two parameters (<tt class="docutils literal"><span class="pre">argc</span></tt>
+and <tt class="docutils literal"><span class="pre">argv</span></tt>), but may also take an optional third parameter which holds
+<a class="reference internal" href="#additional-extra-text">additional extra text</a> to emit when the <tt class="docutils literal"><span class="pre">-help</span></tt> option is invoked, and a
+fourth boolean parameter that enables <a class="reference internal" href="#response-files">response files</a>.</p>
+</div>
+<div class="section" id="the-cl-parseenvironmentoptions-function">
+<span id="cl-parseenvironmentoptions"></span><h4><a class="toc-backref" href="#id35">The <tt class="docutils literal"><span class="pre">cl::ParseEnvironmentOptions</span></tt> function</a><a class="headerlink" href="#the-cl-parseenvironmentoptions-function" title="Permalink to this headline">¶</a></h4>
+<p>The <tt class="docutils literal"><span class="pre">cl::ParseEnvironmentOptions</span></tt> function has mostly the same effects as
+<a class="reference internal" href="#cl-parsecommandlineoptions">cl::ParseCommandLineOptions</a>, except that it is designed to take values for
+options from an environment variable, for those cases in which reading the
+command line is not convenient or desired. It fills in the values of all the
+command line option variables just like <a class="reference internal" href="#cl-parsecommandlineoptions">cl::ParseCommandLineOptions</a> does.</p>
+<p>It takes four parameters: the name of the program (since <tt class="docutils literal"><span class="pre">argv</span></tt> may not be
+available, it can’t just look in <tt class="docutils literal"><span class="pre">argv[0]</span></tt>), the name of the environment
+variable to examine, the optional <a class="reference internal" href="#additional-extra-text">additional extra text</a> to emit when the
+<tt class="docutils literal"><span class="pre">-help</span></tt> option is invoked, and the boolean switch that controls whether
+<a class="reference internal" href="#response-files">response files</a> should be read.</p>
+<p><tt class="docutils literal"><span class="pre">cl::ParseEnvironmentOptions</span></tt> will break the environment variable’s value up
+into words and then process them using <a class="reference internal" href="#cl-parsecommandlineoptions">cl::ParseCommandLineOptions</a>.
+<strong>Note:</strong> Currently <tt class="docutils literal"><span class="pre">cl::ParseEnvironmentOptions</span></tt> does not support quoting, so
+an environment variable containing <tt class="docutils literal"><span class="pre">-option</span> <span class="pre">"foo</span> <span class="pre">bar"</span></tt> will be parsed as three
+words, <tt class="docutils literal"><span class="pre">-option</span></tt>, <tt class="docutils literal"><span class="pre">"foo</span></tt>, and <tt class="docutils literal"><span class="pre">bar"</span></tt>, which is different from what you
+would get from the shell with the same input.</p>
+</div>
+<div class="section" id="the-cl-setversionprinter-function">
+<h4><a class="toc-backref" href="#id36">The <tt class="docutils literal"><span class="pre">cl::SetVersionPrinter</span></tt> function</a><a class="headerlink" href="#the-cl-setversionprinter-function" title="Permalink to this headline">¶</a></h4>
+<p>The <tt class="docutils literal"><span class="pre">cl::SetVersionPrinter</span></tt> function is designed to be called directly from
+<tt class="docutils literal"><span class="pre">main</span></tt> and <em>before</em> <tt class="docutils literal"><span class="pre">cl::ParseCommandLineOptions</span></tt>. Its use is optional. It
+simply arranges for a function to be called in response to the <tt class="docutils literal"><span class="pre">--version</span></tt>
+option instead of having the <tt class="docutils literal"><span class="pre">CommandLine</span></tt> library print out the usual version
+string for LLVM. This is useful for programs that are not part of LLVM but wish
+to use the <tt class="docutils literal"><span class="pre">CommandLine</span></tt> facilities. Such programs should just define a small
+function that takes no arguments and returns <tt class="docutils literal"><span class="pre">void</span></tt> and that prints out
+whatever version information is appropriate for the program. Pass the address of
+that function to <tt class="docutils literal"><span class="pre">cl::SetVersionPrinter</span></tt> to arrange for it to be called when
+the <tt class="docutils literal"><span class="pre">--version</span></tt> option is given by the user.</p>
+</div>
+<div class="section" id="the-cl-opt-class">
+<span id="scalar"></span><span id="cl-opt"></span><h4><a class="toc-backref" href="#id37">The <tt class="docutils literal"><span class="pre">cl::opt</span></tt> class</a><a class="headerlink" href="#the-cl-opt-class" title="Permalink to this headline">¶</a></h4>
+<p>The <tt class="docutils literal"><span class="pre">cl::opt</span></tt> class is the class used to represent scalar command line
+options, and is the one used most of the time. It is a templated class which
+can take up to three arguments (all except for the first have default values
+though):</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">namespace</span> <span class="n">cl</span> <span class="p">{</span>
+ <span class="k">template</span> <span class="o"><</span><span class="k">class</span> <span class="nc">DataType</span><span class="p">,</span> <span class="kt">bool</span> <span class="n">ExternalStorage</span> <span class="o">=</span> <span class="nb">false</span><span class="p">,</span>
+ <span class="k">class</span> <span class="nc">ParserClass</span> <span class="o">=</span> <span class="n">parser</span><span class="o"><</span><span class="n">DataType</span><span class="o">></span> <span class="o">></span>
+ <span class="k">class</span> <span class="nc">opt</span><span class="p">;</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>The first template argument specifies what underlying data type the command line
+argument is, and is used to select a default parser implementation. The second
+template argument is used to specify whether the option should contain the
+storage for the option (the default) or whether external storage should be used
+to contain the value parsed for the option (see <a class="reference internal" href="#internal-vs-external-storage">Internal vs External Storage</a>
+for more information).</p>
+<p>The third template argument specifies which parser to use. The default value
+selects an instantiation of the <tt class="docutils literal"><span class="pre">parser</span></tt> class based on the underlying data
+type of the option. In general, this default works well for most applications,
+so this option is only used when using a <a class="reference internal" href="#custom-parser">custom parser</a>.</p>
+</div>
+<div class="section" id="the-cl-list-class">
+<span id="cl-list"></span><span id="lists-of-arguments"></span><h4><a class="toc-backref" href="#id38">The <tt class="docutils literal"><span class="pre">cl::list</span></tt> class</a><a class="headerlink" href="#the-cl-list-class" title="Permalink to this headline">¶</a></h4>
+<p>The <tt class="docutils literal"><span class="pre">cl::list</span></tt> class is the class used to represent a list of command line
+options. It too is a templated class which can take up to three arguments:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">namespace</span> <span class="n">cl</span> <span class="p">{</span>
+ <span class="k">template</span> <span class="o"><</span><span class="k">class</span> <span class="nc">DataType</span><span class="p">,</span> <span class="k">class</span> <span class="nc">Storage</span> <span class="o">=</span> <span class="kt">bool</span><span class="p">,</span>
+ <span class="k">class</span> <span class="nc">ParserClass</span> <span class="o">=</span> <span class="n">parser</span><span class="o"><</span><span class="n">DataType</span><span class="o">></span> <span class="o">></span>
+ <span class="k">class</span> <span class="nc">list</span><span class="p">;</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>This class works the exact same as the <a class="reference internal" href="#cl-opt">cl::opt</a> class, except that the second
+argument is the <strong>type</strong> of the external storage, not a boolean value. For this
+class, the marker type ‘<tt class="docutils literal"><span class="pre">bool</span></tt>‘ is used to indicate that internal storage
+should be used.</p>
+</div>
+<div class="section" id="the-cl-bits-class">
+<span id="cl-bits"></span><h4><a class="toc-backref" href="#id39">The <tt class="docutils literal"><span class="pre">cl::bits</span></tt> class</a><a class="headerlink" href="#the-cl-bits-class" title="Permalink to this headline">¶</a></h4>
+<p>The <tt class="docutils literal"><span class="pre">cl::bits</span></tt> class is the class used to represent a list of command line
+options in the form of a bit vector. It is also a templated class which can
+take up to three arguments:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">namespace</span> <span class="n">cl</span> <span class="p">{</span>
+ <span class="k">template</span> <span class="o"><</span><span class="k">class</span> <span class="nc">DataType</span><span class="p">,</span> <span class="k">class</span> <span class="nc">Storage</span> <span class="o">=</span> <span class="kt">bool</span><span class="p">,</span>
+ <span class="k">class</span> <span class="nc">ParserClass</span> <span class="o">=</span> <span class="n">parser</span><span class="o"><</span><span class="n">DataType</span><span class="o">></span> <span class="o">></span>
+ <span class="k">class</span> <span class="nc">bits</span><span class="p">;</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>This class works the exact same as the <a class="reference internal" href="#cl-list">cl::list</a> class, except that the second
+argument must be of <strong>type</strong> <tt class="docutils literal"><span class="pre">unsigned</span></tt> if external storage is used.</p>
+</div>
+<div class="section" id="the-cl-alias-class">
+<span id="cl-alias"></span><h4><a class="toc-backref" href="#id40">The <tt class="docutils literal"><span class="pre">cl::alias</span></tt> class</a><a class="headerlink" href="#the-cl-alias-class" title="Permalink to this headline">¶</a></h4>
+<p>The <tt class="docutils literal"><span class="pre">cl::alias</span></tt> class is a nontemplated class that is used to form aliases for
+other arguments.</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">namespace</span> <span class="n">cl</span> <span class="p">{</span>
+ <span class="k">class</span> <span class="nc">alias</span><span class="p">;</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>The <a class="reference internal" href="#cl-aliasopt">cl::aliasopt</a> attribute should be used to specify which option this is an
+alias for. Alias arguments default to being <a class="reference internal" href="#cl-hidden">cl::Hidden</a>, and use the aliased
+options parser to do the conversion from string to data.</p>
+</div>
+<div class="section" id="the-cl-extrahelp-class">
+<span id="cl-extrahelp"></span><h4><a class="toc-backref" href="#id41">The <tt class="docutils literal"><span class="pre">cl::extrahelp</span></tt> class</a><a class="headerlink" href="#the-cl-extrahelp-class" title="Permalink to this headline">¶</a></h4>
+<p>The <tt class="docutils literal"><span class="pre">cl::extrahelp</span></tt> class is a nontemplated class that allows extra help text
+to be printed out for the <tt class="docutils literal"><span class="pre">-help</span></tt> option.</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">namespace</span> <span class="n">cl</span> <span class="p">{</span>
+ <span class="k">struct</span> <span class="n">extrahelp</span><span class="p">;</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>To use the extrahelp, simply construct one with a <tt class="docutils literal"><span class="pre">const</span> <span class="pre">char*</span></tt> parameter to
+the constructor. The text passed to the constructor will be printed at the
+bottom of the help message, verbatim. Note that multiple <tt class="docutils literal"><span class="pre">cl::extrahelp</span></tt>
+<strong>can</strong> be used, but this practice is discouraged. If your tool needs to print
+additional help information, put all that help into a single <tt class="docutils literal"><span class="pre">cl::extrahelp</span></tt>
+instance.</p>
+<p>For example:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">cl</span><span class="o">::</span><span class="n">extrahelp</span><span class="p">(</span><span class="s">"</span><span class="se">\n</span><span class="s">ADDITIONAL HELP:</span><span class="se">\n\n</span><span class="s"> This is the extra help</span><span class="se">\n</span><span class="s">"</span><span class="p">);</span>
+</pre></div>
+</div>
+</div>
+<div class="section" id="the-cl-optioncategory-class">
+<span id="cl-optioncategory"></span><h4><a class="toc-backref" href="#id42">The <tt class="docutils literal"><span class="pre">cl::OptionCategory</span></tt> class</a><a class="headerlink" href="#the-cl-optioncategory-class" title="Permalink to this headline">¶</a></h4>
+<p>The <tt class="docutils literal"><span class="pre">cl::OptionCategory</span></tt> class is a simple class for declaring
+option categories.</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">namespace</span> <span class="n">cl</span> <span class="p">{</span>
+ <span class="k">class</span> <span class="nc">OptionCategory</span><span class="p">;</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>An option category must have a name and optionally a description which are
+passed to the constructor as <tt class="docutils literal"><span class="pre">const</span> <span class="pre">char*</span></tt>.</p>
+<p>Note that declaring an option category and associating it with an option before
+parsing options (e.g. statically) will change the output of <tt class="docutils literal"><span class="pre">-help</span></tt> from
+uncategorized to categorized. If an option category is declared but not
+associated with an option then it will be hidden from the output of <tt class="docutils literal"><span class="pre">-help</span></tt>
+but will be shown in the output of <tt class="docutils literal"><span class="pre">-help-hidden</span></tt>.</p>
+</div>
+</div>
+<div class="section" id="builtin-parsers">
+<span id="discussed-previously"></span><span id="different-parser"></span><h3><a class="toc-backref" href="#id43">Builtin parsers</a><a class="headerlink" href="#builtin-parsers" title="Permalink to this headline">¶</a></h3>
+<p>Parsers control how the string value taken from the command line is translated
+into a typed value, suitable for use in a C++ program. By default, the
+CommandLine library uses an instance of <tt class="docutils literal"><span class="pre">parser<type></span></tt> if the command line
+option specifies that it uses values of type ‘<tt class="docutils literal"><span class="pre">type</span></tt>‘. Because of this,
+custom option processing is specified with specializations of the ‘<tt class="docutils literal"><span class="pre">parser</span></tt>‘
+class.</p>
+<p>The CommandLine library provides the following builtin parser specializations,
+which are sufficient for most applications. It can, however, also be extended to
+work with new data types and new ways of interpreting the same data. See the
+<a class="reference internal" href="#writing-a-custom-parser">Writing a Custom Parser</a> for more details on this type of library extension.</p>
+<ul class="simple" id="cl-parser">
+<span id="enums"></span><li>The generic <tt class="docutils literal"><span class="pre">parser<t></span></tt> parser can be used to map strings values to any data
+type, through the use of the <a class="reference internal" href="#cl-values">cl::values</a> property, which specifies the
+mapping information. The most common use of this parser is for parsing enum
+values, which allows you to use the CommandLine library for all of the error
+checking to make sure that only valid enum values are specified (as opposed to
+accepting arbitrary strings). Despite this, however, the generic parser class
+can be used for any data type.</li>
+</ul>
+<ul class="simple" id="bool-parser">
+<span id="boolean-flags"></span><li>The <strong>parser<bool> specialization</strong> is used to convert boolean strings to a
+boolean value. Currently accepted strings are “<tt class="docutils literal"><span class="pre">true</span></tt>”, “<tt class="docutils literal"><span class="pre">TRUE</span></tt>”,
+“<tt class="docutils literal"><span class="pre">True</span></tt>”, “<tt class="docutils literal"><span class="pre">1</span></tt>”, “<tt class="docutils literal"><span class="pre">false</span></tt>”, “<tt class="docutils literal"><span class="pre">FALSE</span></tt>”, “<tt class="docutils literal"><span class="pre">False</span></tt>”, and “<tt class="docutils literal"><span class="pre">0</span></tt>”.</li>
+<li>The <strong>parser<boolOrDefault> specialization</strong> is used for cases where the value
+is boolean, but we also need to know whether the option was specified at all.
+boolOrDefault is an enum with 3 values, BOU_UNSET, BOU_TRUE and BOU_FALSE.
+This parser accepts the same strings as <strong>``parser<bool>``</strong>.</li>
+</ul>
+<ul class="simple" id="strings">
+<li>The <strong>parser<string> specialization</strong> simply stores the parsed string into the
+string value specified. No conversion or modification of the data is
+performed.</li>
+</ul>
+<ul class="simple" id="int">
+<span id="integers"></span><li>The <strong>parser<int> specialization</strong> uses the C <tt class="docutils literal"><span class="pre">strtol</span></tt> function to parse the
+string input. As such, it will accept a decimal number (with an optional ‘+’
+or ‘-‘ prefix) which must start with a non-zero digit. It accepts octal
+numbers, which are identified with a ‘<tt class="docutils literal"><span class="pre">0</span></tt>‘ prefix digit, and hexadecimal
+numbers with a prefix of ‘<tt class="docutils literal"><span class="pre">0x</span></tt>‘ or ‘<tt class="docutils literal"><span class="pre">0X</span></tt>‘.</li>
+</ul>
+<ul class="simple" id="double">
+<span id="float"></span><span id="doubles"></span><li>The <strong>parser<double></strong> and <strong>parser<float> specializations</strong> use the standard
+C <tt class="docutils literal"><span class="pre">strtod</span></tt> function to convert floating point strings into floating point
+values. As such, a broad range of string formats is supported, including
+exponential notation (ex: <tt class="docutils literal"><span class="pre">1.7e15</span></tt>) and properly supports locales.</li>
+</ul>
+</div>
+</div>
+<div class="section" id="extending-the-library">
+<span id="extension-guide"></span><span id="id6"></span><h2><a class="toc-backref" href="#id44">Extension Guide</a><a class="headerlink" href="#extending-the-library" title="Permalink to this headline">¶</a></h2>
+<p>Although the CommandLine library has a lot of functionality built into it
+already (as discussed previously), one of its true strengths lie in its
+extensibility. This section discusses how the CommandLine library works under
+the covers and illustrates how to do some simple, common, extensions.</p>
+<div class="section" id="writing-a-custom-parser">
+<span id="custom-parser"></span><span id="custom-parsers"></span><span id="id7"></span><h3><a class="toc-backref" href="#id45">Writing a custom parser</a><a class="headerlink" href="#writing-a-custom-parser" title="Permalink to this headline">¶</a></h3>
+<p>One of the simplest and most common extensions is the use of a custom parser.
+As <a class="reference internal" href="#discussed-previously">discussed previously</a>, parsers are the portion of the CommandLine library
+that turns string input from the user into a particular parsed data type,
+validating the input in the process.</p>
+<p>There are two ways to use a new parser:</p>
+<ol class="arabic">
+<li><p class="first">Specialize the <a class="reference internal" href="#cl-parser">cl::parser</a> template for your custom data type.</p>
+<p>This approach has the advantage that users of your custom data type will
+automatically use your custom parser whenever they define an option with a
+value type of your data type. The disadvantage of this approach is that it
+doesn’t work if your fundamental data type is something that is already
+supported.</p>
+</li>
+<li><p class="first">Write an independent class, using it explicitly from options that need it.</p>
+<p>This approach works well in situations where you would line to parse an
+option using special syntax for a not-very-special data-type. The drawback
+of this approach is that users of your parser have to be aware that they are
+using your parser instead of the builtin ones.</p>
+</li>
+</ol>
+<p>To guide the discussion, we will discuss a custom parser that accepts file
+sizes, specified with an optional unit after the numeric size. For example, we
+would like to parse “102kb”, “41M”, “1G” into the appropriate integer value. In
+this case, the underlying data type we want to parse into is ‘<tt class="docutils literal"><span class="pre">unsigned</span></tt>‘. We
+choose approach #2 above because we don’t want to make this the default for all
+<tt class="docutils literal"><span class="pre">unsigned</span></tt> options.</p>
+<p>To start out, we declare our new <tt class="docutils literal"><span class="pre">FileSizeParser</span></tt> class:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">struct</span> <span class="n">FileSizeParser</span> <span class="o">:</span> <span class="k">public</span> <span class="n">cl</span><span class="o">::</span><span class="n">parser</span><span class="o"><</span><span class="kt">unsigned</span><span class="o">></span> <span class="p">{</span>
+ <span class="c1">// parse - Return true on error.</span>
+ <span class="kt">bool</span> <span class="n">parse</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">Option</span> <span class="o">&</span><span class="n">O</span><span class="p">,</span> <span class="n">StringRef</span> <span class="n">ArgName</span><span class="p">,</span> <span class="k">const</span> <span class="n">std</span><span class="o">::</span><span class="n">string</span> <span class="o">&</span><span class="n">ArgValue</span><span class="p">,</span>
+ <span class="kt">unsigned</span> <span class="o">&</span><span class="n">Val</span><span class="p">);</span>
+<span class="p">};</span>
+</pre></div>
+</div>
+<p>Our new class inherits from the <tt class="docutils literal"><span class="pre">cl::parser</span></tt> template class to fill in
+the default, boiler plate code for us. We give it the data type that we parse
+into, the last argument to the <tt class="docutils literal"><span class="pre">parse</span></tt> method, so that clients of our custom
+parser know what object type to pass in to the parse method. (Here we declare
+that we parse into ‘<tt class="docutils literal"><span class="pre">unsigned</span></tt>‘ variables.)</p>
+<p>For most purposes, the only method that must be implemented in a custom parser
+is the <tt class="docutils literal"><span class="pre">parse</span></tt> method. The <tt class="docutils literal"><span class="pre">parse</span></tt> method is called whenever the option is
+invoked, passing in the option itself, the option name, the string to parse, and
+a reference to a return value. If the string to parse is not well-formed, the
+parser should output an error message and return true. Otherwise it should
+return false and set ‘<tt class="docutils literal"><span class="pre">Val</span></tt>‘ to the parsed value. In our example, we
+implement <tt class="docutils literal"><span class="pre">parse</span></tt> as:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="kt">bool</span> <span class="n">FileSizeParser</span><span class="o">::</span><span class="n">parse</span><span class="p">(</span><span class="n">cl</span><span class="o">::</span><span class="n">Option</span> <span class="o">&</span><span class="n">O</span><span class="p">,</span> <span class="n">StringRef</span> <span class="n">ArgName</span><span class="p">,</span>
+ <span class="k">const</span> <span class="n">std</span><span class="o">::</span><span class="n">string</span> <span class="o">&</span><span class="n">Arg</span><span class="p">,</span> <span class="kt">unsigned</span> <span class="o">&</span><span class="n">Val</span><span class="p">)</span> <span class="p">{</span>
+ <span class="k">const</span> <span class="kt">char</span> <span class="o">*</span><span class="n">ArgStart</span> <span class="o">=</span> <span class="n">Arg</span><span class="p">.</span><span class="n">c_str</span><span class="p">();</span>
+ <span class="kt">char</span> <span class="o">*</span><span class="n">End</span><span class="p">;</span>
+
+ <span class="c1">// Parse integer part, leaving 'End' pointing to the first non-integer char</span>
+ <span class="n">Val</span> <span class="o">=</span> <span class="p">(</span><span class="kt">unsigned</span><span class="p">)</span><span class="n">strtol</span><span class="p">(</span><span class="n">ArgStart</span><span class="p">,</span> <span class="o">&</span><span class="n">End</span><span class="p">,</span> <span class="mi">0</span><span class="p">);</span>
+
+ <span class="k">while</span> <span class="p">(</span><span class="mi">1</span><span class="p">)</span> <span class="p">{</span>
+ <span class="k">switch</span> <span class="p">(</span><span class="o">*</span><span class="n">End</span><span class="o">++</span><span class="p">)</span> <span class="p">{</span>
+ <span class="k">case</span> <span class="mi">0</span>: <span class="k">return</span> <span class="nb">false</span><span class="p">;</span> <span class="c1">// No error</span>
+ <span class="k">case</span> <span class="sc">'i'</span>: <span class="c1">// Ignore the 'i' in KiB if people use that</span>
+ <span class="k">case</span> <span class="sc">'b'</span>: <span class="k">case</span> <span class="sc">'B'</span>: <span class="c1">// Ignore B suffix</span>
+ <span class="k">break</span><span class="p">;</span>
+
+ <span class="k">case</span> <span class="sc">'g'</span>: <span class="k">case</span> <span class="sc">'G'</span>: <span class="n">Val</span> <span class="o">*=</span> <span class="mi">1024</span><span class="o">*</span><span class="mi">1024</span><span class="o">*</span><span class="mi">1024</span><span class="p">;</span> <span class="k">break</span><span class="p">;</span>
+ <span class="k">case</span> <span class="sc">'m'</span>: <span class="k">case</span> <span class="sc">'M'</span>: <span class="n">Val</span> <span class="o">*=</span> <span class="mi">1024</span><span class="o">*</span><span class="mi">1024</span><span class="p">;</span> <span class="k">break</span><span class="p">;</span>
+ <span class="k">case</span> <span class="sc">'k'</span>: <span class="k">case</span> <span class="sc">'K'</span>: <span class="n">Val</span> <span class="o">*=</span> <span class="mi">1024</span><span class="p">;</span> <span class="k">break</span><span class="p">;</span>
+
+ <span class="nl">default:</span>
+ <span class="c1">// Print an error message if unrecognized character!</span>
+ <span class="k">return</span> <span class="n">O</span><span class="p">.</span><span class="n">error</span><span class="p">(</span><span class="s">"'"</span> <span class="o">+</span> <span class="n">Arg</span> <span class="o">+</span> <span class="s">"' value invalid for file size argument!"</span><span class="p">);</span>
+ <span class="p">}</span>
+ <span class="p">}</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>This function implements a very simple parser for the kinds of strings we are
+interested in. Although it has some holes (it allows “<tt class="docutils literal"><span class="pre">123KKK</span></tt>” for example),
+it is good enough for this example. Note that we use the option itself to print
+out the error message (the <tt class="docutils literal"><span class="pre">error</span></tt> method always returns true) in order to get
+a nice error message (shown below). Now that we have our parser class, we can
+use it like this:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">static</span> <span class="n">cl</span><span class="o">::</span><span class="n">opt</span><span class="o"><</span><span class="kt">unsigned</span><span class="p">,</span> <span class="nb">false</span><span class="p">,</span> <span class="n">FileSizeParser</span><span class="o">></span>
+<span class="n">MFS</span><span class="p">(</span><span class="s">"max-file-size"</span><span class="p">,</span> <span class="n">cl</span><span class="o">::</span><span class="n">desc</span><span class="p">(</span><span class="s">"Maximum file size to accept"</span><span class="p">),</span>
+ <span class="n">cl</span><span class="o">::</span><span class="n">value_desc</span><span class="p">(</span><span class="s">"size"</span><span class="p">));</span>
+</pre></div>
+</div>
+<p>Which adds this to the output of our program:</p>
+<div class="highlight-python"><div class="highlight"><pre>OPTIONS:
+ -help - display available options (-help-hidden for more)
+ ...
+ -max-file-size=<size> - Maximum file size to accept
+</pre></div>
+</div>
+<p>And we can test that our parse works correctly now (the test program just prints
+out the max-file-size argument value):</p>
+<div class="highlight-python"><div class="highlight"><pre>$ ./test
+MFS: 0
+$ ./test -max-file-size=123MB
+MFS: 128974848
+$ ./test -max-file-size=3G
+MFS: 3221225472
+$ ./test -max-file-size=dog
+-max-file-size option: 'dog' value invalid for file size argument!
+</pre></div>
+</div>
+<p>It looks like it works. The error message that we get is nice and helpful, and
+we seem to accept reasonable file sizes. This wraps up the “custom parser”
+tutorial.</p>
+</div>
+<div class="section" id="exploiting-external-storage">
+<h3><a class="toc-backref" href="#id46">Exploiting external storage</a><a class="headerlink" href="#exploiting-external-storage" title="Permalink to this headline">¶</a></h3>
+<p>Several of the LLVM libraries define static <tt class="docutils literal"><span class="pre">cl::opt</span></tt> instances that will
+automatically be included in any program that links with that library. This is
+a feature. However, sometimes it is necessary to know the value of the command
+line option outside of the library. In these cases the library does or should
+provide an external storage location that is accessible to users of the
+library. Examples of this include the <tt class="docutils literal"><span class="pre">llvm::DebugFlag</span></tt> exported by the
+<tt class="docutils literal"><span class="pre">lib/Support/Debug.cpp</span></tt> file and the <tt class="docutils literal"><span class="pre">llvm::TimePassesIsEnabled</span></tt> flag
+exported by the <tt class="docutils literal"><span class="pre">lib/VMCore/PassManager.cpp</span></tt> file.</p>
+</div>
+<div class="section" id="dynamically-adding-command-line-options">
+<span id="dynamically-loaded-options"></span><h3><a class="toc-backref" href="#id47">Dynamically adding command line options</a><a class="headerlink" href="#dynamically-adding-command-line-options" title="Permalink to this headline">¶</a></h3>
+</div>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="CompilerWriterInfo.html" title="Architecture & Platform Information for Compiler Writers"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="CodingStandards.html" title="LLVM Coding Standards"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CompileCudaWithLLVM.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CompileCudaWithLLVM.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CompileCudaWithLLVM.html (added)
+++ www-releases/trunk/4.0.0/docs/CompileCudaWithLLVM.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,610 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>Compiling CUDA with clang — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="Reporting Guide" href="ReportingGuide.html" />
+ <link rel="prev" title="LLVM Community Code of Conduct" href="CodeOfConduct.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="ReportingGuide.html" title="Reporting Guide"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="CodeOfConduct.html" title="LLVM Community Code of Conduct"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="compiling-cuda-with-clang">
+<h1>Compiling CUDA with clang<a class="headerlink" href="#compiling-cuda-with-clang" title="Permalink to this headline">¶</a></h1>
+<div class="contents local topic" id="contents">
+<ul class="simple">
+<li><a class="reference internal" href="#introduction" id="id1">Introduction</a></li>
+<li><a class="reference internal" href="#compiling-cuda-code" id="id2">Compiling CUDA Code</a><ul>
+<li><a class="reference internal" href="#prerequisites" id="id3">Prerequisites</a></li>
+<li><a class="reference internal" href="#invoking-clang" id="id4">Invoking clang</a></li>
+<li><a class="reference internal" href="#flags-that-control-numerical-code" id="id5">Flags that control numerical code</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#standard-library-support" id="id6">Standard library support</a><ul>
+<li><a class="reference internal" href="#math-h-and-cmath" id="id7"><tt class="docutils literal"><span class="pre"><math.h></span></tt> and <tt class="docutils literal"><span class="pre"><cmath></span></tt></a></li>
+<li><a class="reference internal" href="#std-complex" id="id8"><tt class="docutils literal"><span class="pre"><std::complex></span></tt></a></li>
+<li><a class="reference internal" href="#algorithm" id="id9"><tt class="docutils literal"><span class="pre"><algorithm></span></tt></a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#detecting-clang-vs-nvcc-from-code" id="id10">Detecting clang vs NVCC from code</a></li>
+<li><a class="reference internal" href="#dialect-differences-between-clang-and-nvcc" id="id11">Dialect Differences Between clang and nvcc</a><ul>
+<li><a class="reference internal" href="#compilation-models" id="id12">Compilation Models</a></li>
+<li><a class="reference internal" href="#overloading-based-on-host-and-device-attributes" id="id13">Overloading Based on <tt class="docutils literal"><span class="pre">__host__</span></tt> and <tt class="docutils literal"><span class="pre">__device__</span></tt> Attributes</a></li>
+<li><a class="reference internal" href="#using-a-different-class-on-host-device" id="id14">Using a Different Class on Host/Device</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#optimizations" id="id15">Optimizations</a></li>
+<li><a class="reference internal" href="#publication" id="id16">Publication</a></li>
+<li><a class="reference internal" href="#obtaining-help" id="id17">Obtaining Help</a></li>
+</ul>
+</div>
+<div class="section" id="introduction">
+<h2><a class="toc-backref" href="#id1">Introduction</a><a class="headerlink" href="#introduction" title="Permalink to this headline">¶</a></h2>
+<p>This document describes how to compile CUDA code with clang, and gives some
+details about LLVM and clang’s CUDA implementations.</p>
+<p>This document assumes a basic familiarity with CUDA. Information about CUDA
+programming can be found in the
+<a class="reference external" href="http://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html">CUDA programming guide</a>.</p>
+</div>
+<div class="section" id="compiling-cuda-code">
+<h2><a class="toc-backref" href="#id2">Compiling CUDA Code</a><a class="headerlink" href="#compiling-cuda-code" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="prerequisites">
+<h3><a class="toc-backref" href="#id3">Prerequisites</a><a class="headerlink" href="#prerequisites" title="Permalink to this headline">¶</a></h3>
+<p>CUDA is supported in llvm 3.9, but it’s still in active development, so we
+recommend you <a class="reference external" href="http://llvm.org/docs/GettingStarted.html">compile clang/LLVM from HEAD</a>.</p>
+<p>Before you build CUDA code, you’ll need to have installed the appropriate
+driver for your nvidia GPU and the CUDA SDK. See <a class="reference external" href="https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html">NVIDIA’s CUDA installation
+guide</a>
+for details. Note that clang <a class="reference external" href="https://llvm.org/bugs/show_bug.cgi?id=26966">does not support</a> the CUDA toolkit as installed
+by many Linux package managers; you probably need to install nvidia’s package.</p>
+<p>You will need CUDA 7.0, 7.5, or 8.0 to compile with clang.</p>
+<p>CUDA compilation is supported on Linux, on MacOS as of 2016-11-18, and on
+Windows as of 2017-01-05.</p>
+</div>
+<div class="section" id="invoking-clang">
+<h3><a class="toc-backref" href="#id4">Invoking clang</a><a class="headerlink" href="#invoking-clang" title="Permalink to this headline">¶</a></h3>
+<p>Invoking clang for CUDA compilation works similarly to compiling regular C++.
+You just need to be aware of a few additional flags.</p>
+<p>You can use <a class="reference external" href="https://gist.github.com/855e277884eb6b388cd2f00d956c2fd4">this</a>
+program as a toy example. Save it as <tt class="docutils literal"><span class="pre">axpy.cu</span></tt>. (Clang detects that you’re
+compiling CUDA code by noticing that your filename ends with <tt class="docutils literal"><span class="pre">.cu</span></tt>.
+Alternatively, you can pass <tt class="docutils literal"><span class="pre">-x</span> <span class="pre">cuda</span></tt>.)</p>
+<p>To build and run, run the following commands, filling in the parts in angle
+brackets as described below:</p>
+<div class="highlight-console"><div class="highlight"><pre><span class="gp">$</span> clang++ axpy.cu -o axpy --cuda-gpu-arch<span class="o">=</span><GPU arch> <span class="se">\</span>
+<span class="go"> -L<CUDA install path>/<lib64 or lib> \</span>
+<span class="go"> -lcudart_static -ldl -lrt -pthread</span>
+<span class="gp">$</span> ./axpy
+<span class="go">y[0] = 2</span>
+<span class="go">y[1] = 4</span>
+<span class="go">y[2] = 6</span>
+<span class="go">y[3] = 8</span>
+</pre></div>
+</div>
+<p>On MacOS, replace <cite>-lcudart_static</cite> with <cite>-lcudart</cite>; otherwise, you may get
+“CUDA driver version is insufficient for CUDA runtime version” errors when you
+run your program.</p>
+<ul>
+<li><p class="first"><tt class="docutils literal"><span class="pre"><CUDA</span> <span class="pre">install</span> <span class="pre">path></span></tt> – the directory where you installed CUDA SDK.
+Typically, <tt class="docutils literal"><span class="pre">/usr/local/cuda</span></tt>.</p>
+<p>Pass e.g. <tt class="docutils literal"><span class="pre">-L/usr/local/cuda/lib64</span></tt> if compiling in 64-bit mode; otherwise,
+pass e.g. <tt class="docutils literal"><span class="pre">-L/usr/local/cuda/lib</span></tt>. (In CUDA, the device code and host code
+always have the same pointer widths, so if you’re compiling 64-bit code for
+the host, you’re also compiling 64-bit code for the device.)</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre"><GPU</span> <span class="pre">arch></span></tt> – the <a class="reference external" href="https://developer.nvidia.com/cuda-gpus">compute capability</a> of your GPU. For example, if you
+want to run your program on a GPU with compute capability of 3.5, specify
+<tt class="docutils literal"><span class="pre">--cuda-gpu-arch=sm_35</span></tt>.</p>
+<p>Note: You cannot pass <tt class="docutils literal"><span class="pre">compute_XX</span></tt> as an argument to <tt class="docutils literal"><span class="pre">--cuda-gpu-arch</span></tt>;
+only <tt class="docutils literal"><span class="pre">sm_XX</span></tt> is currently supported. However, clang always includes PTX in
+its binaries, so e.g. a binary compiled with <tt class="docutils literal"><span class="pre">--cuda-gpu-arch=sm_30</span></tt> would be
+forwards-compatible with e.g. <tt class="docutils literal"><span class="pre">sm_35</span></tt> GPUs.</p>
+<p>You can pass <tt class="docutils literal"><span class="pre">--cuda-gpu-arch</span></tt> multiple times to compile for multiple archs.</p>
+</li>
+</ul>
+<p>The <cite>-L</cite> and <cite>-l</cite> flags only need to be passed when linking. When compiling,
+you may also need to pass <tt class="docutils literal"><span class="pre">--cuda-path=/path/to/cuda</span></tt> if you didn’t install
+the CUDA SDK into <tt class="docutils literal"><span class="pre">/usr/local/cuda</span></tt>, <tt class="docutils literal"><span class="pre">/usr/local/cuda-7.0</span></tt>, or
+<tt class="docutils literal"><span class="pre">/usr/local/cuda-7.5</span></tt>.</p>
+</div>
+<div class="section" id="flags-that-control-numerical-code">
+<h3><a class="toc-backref" href="#id5">Flags that control numerical code</a><a class="headerlink" href="#flags-that-control-numerical-code" title="Permalink to this headline">¶</a></h3>
+<p>If you’re using GPUs, you probably care about making numerical code run fast.
+GPU hardware allows for more control over numerical operations than most CPUs,
+but this results in more compiler options for you to juggle.</p>
+<p>Flags you may wish to tweak include:</p>
+<ul>
+<li><p class="first"><tt class="docutils literal"><span class="pre">-ffp-contract={on,off,fast}</span></tt> (defaults to <tt class="docutils literal"><span class="pre">fast</span></tt> on host and device when
+compiling CUDA) Controls whether the compiler emits fused multiply-add
+operations.</p>
+<ul class="simple">
+<li><tt class="docutils literal"><span class="pre">off</span></tt>: never emit fma operations, and prevent ptxas from fusing multiply
+and add instructions.</li>
+<li><tt class="docutils literal"><span class="pre">on</span></tt>: fuse multiplies and adds within a single statement, but never
+across statements (C11 semantics). Prevent ptxas from fusing other
+multiplies and adds.</li>
+<li><tt class="docutils literal"><span class="pre">fast</span></tt>: fuse multiplies and adds wherever profitable, even across
+statements. Doesn’t prevent ptxas from fusing additional multiplies and
+adds.</li>
+</ul>
+<p>Fused multiply-add instructions can be much faster than the unfused
+equivalents, but because the intermediate result in an fma is not rounded,
+this flag can affect numerical code.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">-fcuda-flush-denormals-to-zero</span></tt> (default: off) When this is enabled,
+floating point operations may flush <a class="reference external" href="https://en.wikipedia.org/wiki/Denormal_number">denormal</a> inputs and/or outputs to 0.
+Operations on denormal numbers are often much slower than the same operations
+on normal numbers.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">-fcuda-approx-transcendentals</span></tt> (default: off) When this is enabled, the
+compiler may emit calls to faster, approximate versions of transcendental
+functions, instead of using the slower, fully IEEE-compliant versions. For
+example, this flag allows clang to emit the ptx <tt class="docutils literal"><span class="pre">sin.approx.f32</span></tt>
+instruction.</p>
+<p>This is implied by <tt class="docutils literal"><span class="pre">-ffast-math</span></tt>.</p>
+</li>
+</ul>
+</div>
+</div>
+<div class="section" id="standard-library-support">
+<h2><a class="toc-backref" href="#id6">Standard library support</a><a class="headerlink" href="#standard-library-support" title="Permalink to this headline">¶</a></h2>
+<p>In clang and nvcc, most of the C++ standard library is not supported on the
+device side.</p>
+<div class="section" id="math-h-and-cmath">
+<h3><a class="toc-backref" href="#id7"><tt class="docutils literal"><span class="pre"><math.h></span></tt> and <tt class="docutils literal"><span class="pre"><cmath></span></tt></a><a class="headerlink" href="#math-h-and-cmath" title="Permalink to this headline">¶</a></h3>
+<p>In clang, <tt class="docutils literal"><span class="pre">math.h</span></tt> and <tt class="docutils literal"><span class="pre">cmath</span></tt> are available and <a class="reference external" href="https://github.com/llvm-mirror/test-suite/blob/master/External/CUDA/math_h.cu">pass</a>
+<a class="reference external" href="https://github.com/llvm-mirror/test-suite/blob/master/External/CUDA/cmath.cu">tests</a>
+adapted from libc++’s test suite.</p>
+<p>In nvcc <tt class="docutils literal"><span class="pre">math.h</span></tt> and <tt class="docutils literal"><span class="pre">cmath</span></tt> are mostly available. Versions of <tt class="docutils literal"><span class="pre">::foof</span></tt>
+in namespace std (e.g. <tt class="docutils literal"><span class="pre">std::sinf</span></tt>) are not available, and where the standard
+calls for overloads that take integral arguments, these are usually not
+available.</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="cp">#include <math.h></span>
+<span class="cp">#include <cmath.h></span>
+
+<span class="c1">// clang is OK with everything in this function.</span>
+<span class="n">__device__</span> <span class="kt">void</span> <span class="nf">test</span><span class="p">()</span> <span class="p">{</span>
+ <span class="n">std</span><span class="o">::</span><span class="n">sin</span><span class="p">(</span><span class="mf">0.</span><span class="p">);</span> <span class="c1">// nvcc - ok</span>
+ <span class="n">std</span><span class="o">::</span><span class="n">sin</span><span class="p">(</span><span class="mi">0</span><span class="p">);</span> <span class="c1">// nvcc - error, because no std::sin(int) override is available.</span>
+ <span class="n">sin</span><span class="p">(</span><span class="mi">0</span><span class="p">);</span> <span class="c1">// nvcc - same as above.</span>
+
+ <span class="n">sinf</span><span class="p">(</span><span class="mf">0.</span><span class="p">);</span> <span class="c1">// nvcc - ok</span>
+ <span class="n">std</span><span class="o">::</span><span class="n">sinf</span><span class="p">(</span><span class="mf">0.</span><span class="p">);</span> <span class="c1">// nvcc - no such function</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+</div>
+<div class="section" id="std-complex">
+<h3><a class="toc-backref" href="#id8"><tt class="docutils literal"><span class="pre"><std::complex></span></tt></a><a class="headerlink" href="#std-complex" title="Permalink to this headline">¶</a></h3>
+<p>nvcc does not officially support <tt class="docutils literal"><span class="pre">std::complex</span></tt>. It’s an error to use
+<tt class="docutils literal"><span class="pre">std::complex</span></tt> in <tt class="docutils literal"><span class="pre">__device__</span></tt> code, but it often works in <tt class="docutils literal"><span class="pre">__host__</span>
+<span class="pre">__device__</span></tt> code due to nvcc’s interpretation of the “wrong-side rule” (see
+below). However, we have heard from implementers that it’s possible to get
+into situations where nvcc will omit a call to an <tt class="docutils literal"><span class="pre">std::complex</span></tt> function,
+especially when compiling without optimizations.</p>
+<p>As of 2016-11-16, clang supports <tt class="docutils literal"><span class="pre">std::complex</span></tt> without these caveats. It is
+tested with libstdc++ 4.8.5 and newer, but is known to work only with libc++
+newer than 2016-11-16.</p>
+</div>
+<div class="section" id="algorithm">
+<h3><a class="toc-backref" href="#id9"><tt class="docutils literal"><span class="pre"><algorithm></span></tt></a><a class="headerlink" href="#algorithm" title="Permalink to this headline">¶</a></h3>
+<p>In C++14, many useful functions from <tt class="docutils literal"><span class="pre"><algorithm></span></tt> (notably, <tt class="docutils literal"><span class="pre">std::min</span></tt> and
+<tt class="docutils literal"><span class="pre">std::max</span></tt>) become constexpr. You can therefore use these in device code,
+when compiling with clang.</p>
+</div>
+</div>
+<div class="section" id="detecting-clang-vs-nvcc-from-code">
+<h2><a class="toc-backref" href="#id10">Detecting clang vs NVCC from code</a><a class="headerlink" href="#detecting-clang-vs-nvcc-from-code" title="Permalink to this headline">¶</a></h2>
+<p>Although clang’s CUDA implementation is largely compatible with NVCC’s, you may
+still want to detect when you’re compiling CUDA code specifically with clang.</p>
+<p>This is tricky, because NVCC may invoke clang as part of its own compilation
+process! For example, NVCC uses the host compiler’s preprocessor when
+compiling for device code, and that host compiler may in fact be clang.</p>
+<p>When clang is actually compiling CUDA code – rather than being used as a
+subtool of NVCC’s – it defines the <tt class="docutils literal"><span class="pre">__CUDA__</span></tt> macro. <tt class="docutils literal"><span class="pre">__CUDA_ARCH__</span></tt> is
+defined only in device mode (but will be defined if NVCC is using clang as a
+preprocessor). So you can use the following incantations to detect clang CUDA
+compilation, in host and device modes:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="cp">#if defined(__clang__) && defined(__CUDA__) && !defined(__CUDA_ARCH__)</span>
+<span class="c1">// clang compiling CUDA code, host mode.</span>
+<span class="cp">#endif</span>
+
+<span class="cp">#if defined(__clang__) && defined(__CUDA__) && defined(__CUDA_ARCH__)</span>
+<span class="c1">// clang compiling CUDA code, device mode.</span>
+<span class="cp">#endif</span>
+</pre></div>
+</div>
+<p>Both clang and nvcc define <tt class="docutils literal"><span class="pre">__CUDACC__</span></tt> during CUDA compilation. You can
+detect NVCC specifically by looking for <tt class="docutils literal"><span class="pre">__NVCC__</span></tt>.</p>
+</div>
+<div class="section" id="dialect-differences-between-clang-and-nvcc">
+<h2><a class="toc-backref" href="#id11">Dialect Differences Between clang and nvcc</a><a class="headerlink" href="#dialect-differences-between-clang-and-nvcc" title="Permalink to this headline">¶</a></h2>
+<p>There is no formal CUDA spec, and clang and nvcc speak slightly different
+dialects of the language. Below, we describe some of the differences.</p>
+<p>This section is painful; hopefully you can skip this section and live your life
+blissfully unaware.</p>
+<div class="section" id="compilation-models">
+<h3><a class="toc-backref" href="#id12">Compilation Models</a><a class="headerlink" href="#compilation-models" title="Permalink to this headline">¶</a></h3>
+<p>Most of the differences between clang and nvcc stem from the different
+compilation models used by clang and nvcc. nvcc uses <em>split compilation</em>,
+which works roughly as follows:</p>
+<blockquote>
+<div><ul class="simple">
+<li>Run a preprocessor over the input <tt class="docutils literal"><span class="pre">.cu</span></tt> file to split it into two source
+files: <tt class="docutils literal"><span class="pre">H</span></tt>, containing source code for the host, and <tt class="docutils literal"><span class="pre">D</span></tt>, containing
+source code for the device.</li>
+<li>For each GPU architecture <tt class="docutils literal"><span class="pre">arch</span></tt> that we’re compiling for, do:<ul>
+<li>Compile <tt class="docutils literal"><span class="pre">D</span></tt> using nvcc proper. The result of this is a <tt class="docutils literal"><span class="pre">ptx</span></tt> file for
+<tt class="docutils literal"><span class="pre">P_arch</span></tt>.</li>
+<li>Optionally, invoke <tt class="docutils literal"><span class="pre">ptxas</span></tt>, the PTX assembler, to generate a file,
+<tt class="docutils literal"><span class="pre">S_arch</span></tt>, containing GPU machine code (SASS) for <tt class="docutils literal"><span class="pre">arch</span></tt>.</li>
+</ul>
+</li>
+<li>Invoke <tt class="docutils literal"><span class="pre">fatbin</span></tt> to combine all <tt class="docutils literal"><span class="pre">P_arch</span></tt> and <tt class="docutils literal"><span class="pre">S_arch</span></tt> files into a
+single “fat binary” file, <tt class="docutils literal"><span class="pre">F</span></tt>.</li>
+<li>Compile <tt class="docutils literal"><span class="pre">H</span></tt> using an external host compiler (gcc, clang, or whatever you
+like). <tt class="docutils literal"><span class="pre">F</span></tt> is packaged up into a header file which is force-included into
+<tt class="docutils literal"><span class="pre">H</span></tt>; nvcc generates code that calls into this header to e.g. launch
+kernels.</li>
+</ul>
+</div></blockquote>
+<p>clang uses <em>merged parsing</em>. This is similar to split compilation, except all
+of the host and device code is present and must be semantically-correct in both
+compilation steps.</p>
+<blockquote>
+<div><ul>
+<li><p class="first">For each GPU architecture <tt class="docutils literal"><span class="pre">arch</span></tt> that we’re compiling for, do:</p>
+<ul>
+<li><p class="first">Compile the input <tt class="docutils literal"><span class="pre">.cu</span></tt> file for device, using clang. <tt class="docutils literal"><span class="pre">__host__</span></tt> code
+is parsed and must be semantically correct, even though we’re not
+generating code for the host at this time.</p>
+<p>The output of this step is a <tt class="docutils literal"><span class="pre">ptx</span></tt> file <tt class="docutils literal"><span class="pre">P_arch</span></tt>.</p>
+</li>
+<li><p class="first">Invoke <tt class="docutils literal"><span class="pre">ptxas</span></tt> to generate a SASS file, <tt class="docutils literal"><span class="pre">S_arch</span></tt>. Note that, unlike
+nvcc, clang always generates SASS code.</p>
+</li>
+</ul>
+</li>
+<li><p class="first">Invoke <tt class="docutils literal"><span class="pre">fatbin</span></tt> to combine all <tt class="docutils literal"><span class="pre">P_arch</span></tt> and <tt class="docutils literal"><span class="pre">S_arch</span></tt> files into a
+single fat binary file, <tt class="docutils literal"><span class="pre">F</span></tt>.</p>
+</li>
+<li><p class="first">Compile <tt class="docutils literal"><span class="pre">H</span></tt> using clang. <tt class="docutils literal"><span class="pre">__device__</span></tt> code is parsed and must be
+semantically correct, even though we’re not generating code for the device
+at this time.</p>
+<p><tt class="docutils literal"><span class="pre">F</span></tt> is passed to this compilation, and clang includes it in a special ELF
+section, where it can be found by tools like <tt class="docutils literal"><span class="pre">cuobjdump</span></tt>.</p>
+</li>
+</ul>
+</div></blockquote>
+<p>(You may ask at this point, why does clang need to parse the input file
+multiple times? Why not parse it just once, and then use the AST to generate
+code for the host and each device architecture?</p>
+<p>Unfortunately this can’t work because we have to define different macros during
+host compilation and during device compilation for each GPU architecture.)</p>
+<p>clang’s approach allows it to be highly robust to C++ edge cases, as it doesn’t
+need to decide at an early stage which declarations to keep and which to throw
+away. But it has some consequences you should be aware of.</p>
+</div>
+<div class="section" id="overloading-based-on-host-and-device-attributes">
+<h3><a class="toc-backref" href="#id13">Overloading Based on <tt class="docutils literal"><span class="pre">__host__</span></tt> and <tt class="docutils literal"><span class="pre">__device__</span></tt> Attributes</a><a class="headerlink" href="#overloading-based-on-host-and-device-attributes" title="Permalink to this headline">¶</a></h3>
+<p>Let “H”, “D”, and “HD” stand for “<tt class="docutils literal"><span class="pre">__host__</span></tt> functions”, “<tt class="docutils literal"><span class="pre">__device__</span></tt>
+functions”, and “<tt class="docutils literal"><span class="pre">__host__</span> <span class="pre">__device__</span></tt> functions”, respectively. Functions
+with no attributes behave the same as H.</p>
+<p>nvcc does not allow you to create H and D functions with the same signature:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="c1">// nvcc: error - function "foo" has already been defined</span>
+<span class="n">__host__</span> <span class="kt">void</span> <span class="nf">foo</span><span class="p">()</span> <span class="p">{}</span>
+<span class="n">__device__</span> <span class="kt">void</span> <span class="nf">foo</span><span class="p">()</span> <span class="p">{}</span>
+</pre></div>
+</div>
+<p>However, nvcc allows you to “overload” H and D functions with different
+signatures:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="c1">// nvcc: no error</span>
+<span class="n">__host__</span> <span class="kt">void</span> <span class="nf">foo</span><span class="p">(</span><span class="kt">int</span><span class="p">)</span> <span class="p">{}</span>
+<span class="n">__device__</span> <span class="kt">void</span> <span class="nf">foo</span><span class="p">()</span> <span class="p">{}</span>
+</pre></div>
+</div>
+<p>In clang, the <tt class="docutils literal"><span class="pre">__host__</span></tt> and <tt class="docutils literal"><span class="pre">__device__</span></tt> attributes are part of a
+function’s signature, and so it’s legal to have H and D functions with
+(otherwise) the same signature:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="c1">// clang: no error</span>
+<span class="n">__host__</span> <span class="kt">void</span> <span class="nf">foo</span><span class="p">()</span> <span class="p">{}</span>
+<span class="n">__device__</span> <span class="kt">void</span> <span class="nf">foo</span><span class="p">()</span> <span class="p">{}</span>
+</pre></div>
+</div>
+<p>HD functions cannot be overloaded by H or D functions with the same signature:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="c1">// nvcc: error - function "foo" has already been defined</span>
+<span class="c1">// clang: error - redefinition of 'foo'</span>
+<span class="n">__host__</span> <span class="n">__device__</span> <span class="kt">void</span> <span class="nf">foo</span><span class="p">()</span> <span class="p">{}</span>
+<span class="n">__device__</span> <span class="kt">void</span> <span class="nf">foo</span><span class="p">()</span> <span class="p">{}</span>
+
+<span class="c1">// nvcc: no error</span>
+<span class="c1">// clang: no error</span>
+<span class="n">__host__</span> <span class="n">__device__</span> <span class="kt">void</span> <span class="nf">bar</span><span class="p">(</span><span class="kt">int</span><span class="p">)</span> <span class="p">{}</span>
+<span class="n">__device__</span> <span class="kt">void</span> <span class="nf">bar</span><span class="p">()</span> <span class="p">{}</span>
+</pre></div>
+</div>
+<p>When resolving an overloaded function, clang considers the host/device
+attributes of the caller and callee. These are used as a tiebreaker during
+overload resolution. See <a class="reference external" href="http://clang.llvm.org/doxygen/SemaCUDA_8cpp.html">IdentifyCUDAPreference</a> for the full set of rules,
+but at a high level they are:</p>
+<blockquote>
+<div><ul>
+<li><p class="first">D functions prefer to call other Ds. HDs are given lower priority.</p>
+</li>
+<li><p class="first">Similarly, H functions prefer to call other Hs, or <tt class="docutils literal"><span class="pre">__global__</span></tt> functions
+(with equal priority). HDs are given lower priority.</p>
+</li>
+<li><p class="first">HD functions prefer to call other HDs.</p>
+<p>When compiling for device, HDs will call Ds with lower priority than HD, and
+will call Hs with still lower priority. If it’s forced to call an H, the
+program is malformed if we emit code for this HD function. We call this the
+“wrong-side rule”, see example below.</p>
+<p>The rules are symmetrical when compiling for host.</p>
+</li>
+</ul>
+</div></blockquote>
+<p>Some examples:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">__host__</span> <span class="kt">void</span> <span class="nf">foo</span><span class="p">();</span>
+<span class="n">__device__</span> <span class="kt">void</span> <span class="nf">foo</span><span class="p">();</span>
+
+<span class="n">__host__</span> <span class="kt">void</span> <span class="nf">bar</span><span class="p">();</span>
+<span class="n">__host__</span> <span class="n">__device__</span> <span class="kt">void</span> <span class="nf">bar</span><span class="p">();</span>
+
+<span class="n">__host__</span> <span class="kt">void</span> <span class="nf">test_host</span><span class="p">()</span> <span class="p">{</span>
+ <span class="n">foo</span><span class="p">();</span> <span class="c1">// calls H overload</span>
+ <span class="n">bar</span><span class="p">();</span> <span class="c1">// calls H overload</span>
+<span class="p">}</span>
+
+<span class="n">__device__</span> <span class="kt">void</span> <span class="nf">test_device</span><span class="p">()</span> <span class="p">{</span>
+ <span class="n">foo</span><span class="p">();</span> <span class="c1">// calls D overload</span>
+ <span class="n">bar</span><span class="p">();</span> <span class="c1">// calls HD overload</span>
+<span class="p">}</span>
+
+<span class="n">__host__</span> <span class="n">__device__</span> <span class="kt">void</span> <span class="nf">test_hd</span><span class="p">()</span> <span class="p">{</span>
+ <span class="n">foo</span><span class="p">();</span> <span class="c1">// calls H overload when compiling for host, otherwise D overload</span>
+ <span class="n">bar</span><span class="p">();</span> <span class="c1">// always calls HD overload</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>Wrong-side rule example:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">__host__</span> <span class="kt">void</span> <span class="nf">host_only</span><span class="p">();</span>
+
+<span class="c1">// We don't codegen inline functions unless they're referenced by a</span>
+<span class="c1">// non-inline function. inline_hd1() is called only from the host side, so</span>
+<span class="c1">// does not generate an error. inline_hd2() is called from the device side,</span>
+<span class="c1">// so it generates an error.</span>
+<span class="kr">inline</span> <span class="n">__host__</span> <span class="n">__device__</span> <span class="kt">void</span> <span class="nf">inline_hd1</span><span class="p">()</span> <span class="p">{</span> <span class="n">host_only</span><span class="p">();</span> <span class="p">}</span> <span class="c1">// no error</span>
+<span class="kr">inline</span> <span class="n">__host__</span> <span class="n">__device__</span> <span class="kt">void</span> <span class="nf">inline_hd2</span><span class="p">()</span> <span class="p">{</span> <span class="n">host_only</span><span class="p">();</span> <span class="p">}</span> <span class="c1">// error</span>
+
+<span class="n">__host__</span> <span class="kt">void</span> <span class="nf">host_fn</span><span class="p">()</span> <span class="p">{</span> <span class="n">inline_hd1</span><span class="p">();</span> <span class="p">}</span>
+<span class="n">__device__</span> <span class="kt">void</span> <span class="nf">device_fn</span><span class="p">()</span> <span class="p">{</span> <span class="n">inline_hd2</span><span class="p">();</span> <span class="p">}</span>
+
+<span class="c1">// This function is not inline, so it's always codegen'ed on both the host</span>
+<span class="c1">// and the device. Therefore, it generates an error.</span>
+<span class="n">__host__</span> <span class="n">__device__</span> <span class="kt">void</span> <span class="nf">not_inline_hd</span><span class="p">()</span> <span class="p">{</span> <span class="n">host_only</span><span class="p">();</span> <span class="p">}</span>
+</pre></div>
+</div>
+<p>For the purposes of the wrong-side rule, templated functions also behave like
+<tt class="docutils literal"><span class="pre">inline</span></tt> functions: They aren’t codegen’ed unless they’re instantiated
+(usually as part of the process of invoking them).</p>
+<p>clang’s behavior with respect to the wrong-side rule matches nvcc’s, except
+nvcc only emits a warning for <tt class="docutils literal"><span class="pre">not_inline_hd</span></tt>; device code is allowed to call
+<tt class="docutils literal"><span class="pre">not_inline_hd</span></tt>. In its generated code, nvcc may omit <tt class="docutils literal"><span class="pre">not_inline_hd</span></tt>‘s
+call to <tt class="docutils literal"><span class="pre">host_only</span></tt> entirely, or it may try to generate code for
+<tt class="docutils literal"><span class="pre">host_only</span></tt> on the device. What you get seems to depend on whether or not
+the compiler chooses to inline <tt class="docutils literal"><span class="pre">host_only</span></tt>.</p>
+<p>Member functions, including constructors, may be overloaded using H and D
+attributes. However, destructors cannot be overloaded.</p>
+</div>
+<div class="section" id="using-a-different-class-on-host-device">
+<h3><a class="toc-backref" href="#id14">Using a Different Class on Host/Device</a><a class="headerlink" href="#using-a-different-class-on-host-device" title="Permalink to this headline">¶</a></h3>
+<p>Occasionally you may want to have a class with different host/device versions.</p>
+<p>If all of the class’s members are the same on the host and device, you can just
+provide overloads for the class’s member functions.</p>
+<p>However, if you want your class to have different members on host/device, you
+won’t be able to provide working H and D overloads in both classes. In this
+case, clang is likely to be unhappy with you.</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="cp">#ifdef __CUDA_ARCH__</span>
+<span class="k">struct</span> <span class="n">S</span> <span class="p">{</span>
+ <span class="n">__device__</span> <span class="kt">void</span> <span class="n">foo</span><span class="p">()</span> <span class="p">{</span> <span class="cm">/* use device_only */</span> <span class="p">}</span>
+ <span class="kt">int</span> <span class="n">device_only</span><span class="p">;</span>
+<span class="p">};</span>
+<span class="cp">#else</span>
+<span class="k">struct</span> <span class="n">S</span> <span class="p">{</span>
+ <span class="n">__host__</span> <span class="kt">void</span> <span class="n">foo</span><span class="p">()</span> <span class="p">{</span> <span class="cm">/* use host_only */</span> <span class="p">}</span>
+ <span class="kt">double</span> <span class="n">host_only</span><span class="p">;</span>
+<span class="p">};</span>
+
+<span class="n">__device__</span> <span class="kt">void</span> <span class="nf">test</span><span class="p">()</span> <span class="p">{</span>
+ <span class="n">S</span> <span class="n">s</span><span class="p">;</span>
+ <span class="c1">// clang generates an error here, because during host compilation, we</span>
+ <span class="c1">// have ifdef'ed away the __device__ overload of S::foo(). The __device__</span>
+ <span class="c1">// overload must be present *even during host compilation*.</span>
+ <span class="n">S</span><span class="p">.</span><span class="n">foo</span><span class="p">();</span>
+<span class="p">}</span>
+<span class="cp">#endif</span>
+</pre></div>
+</div>
+<p>We posit that you don’t really want to have classes with different members on H
+and D. For example, if you were to pass one of these as a parameter to a
+kernel, it would have a different layout on H and D, so would not work
+properly.</p>
+<p>To make code like this compatible with clang, we recommend you separate it out
+into two classes. If you need to write code that works on both host and
+device, consider writing an overloaded wrapper function that returns different
+types on host and device.</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">struct</span> <span class="n">HostS</span> <span class="p">{</span> <span class="p">...</span> <span class="p">};</span>
+<span class="k">struct</span> <span class="n">DeviceS</span> <span class="p">{</span> <span class="p">...</span> <span class="p">};</span>
+
+<span class="n">__host__</span> <span class="n">HostS</span> <span class="nf">MakeStruct</span><span class="p">()</span> <span class="p">{</span> <span class="k">return</span> <span class="n">HostS</span><span class="p">();</span> <span class="p">}</span>
+<span class="n">__device__</span> <span class="n">DeviceS</span> <span class="nf">MakeStruct</span><span class="p">()</span> <span class="p">{</span> <span class="k">return</span> <span class="n">DeviceS</span><span class="p">();</span> <span class="p">}</span>
+
+<span class="c1">// Now host and device code can call MakeStruct().</span>
+</pre></div>
+</div>
+<p>Unfortunately, this idiom isn’t compatible with nvcc, because it doesn’t allow
+you to overload based on the H/D attributes. Here’s an idiom that works with
+both clang and nvcc:</p>
+<div class="highlight-c++"><div class="highlight"><pre>struct HostS { ... };
+struct DeviceS { ... };
+
+#ifdef __NVCC__
+ #ifndef __CUDA_ARCH__
+ __host__ HostS MakeStruct() { return HostS(); }
+ #else
+ __device__ DeviceS MakeStruct() { return DeviceS(); }
+ #endif
+#else
+ __host__ HostS MakeStruct() { return HostS(); }
+ __device__ DeviceS MakeStruct() { return DeviceS(); }
+#endif
+
+// Now host and device code can call MakeStruct().
+</pre></div>
+</div>
+<p>Hopefully you don’t have to do this sort of thing often.</p>
+</div>
+</div>
+<div class="section" id="optimizations">
+<h2><a class="toc-backref" href="#id15">Optimizations</a><a class="headerlink" href="#optimizations" title="Permalink to this headline">¶</a></h2>
+<p>Modern CPUs and GPUs are architecturally quite different, so code that’s fast
+on a CPU isn’t necessarily fast on a GPU. We’ve made a number of changes to
+LLVM to make it generate good GPU code. Among these changes are:</p>
+<ul>
+<li><p class="first"><a class="reference external" href="https://goo.gl/4Rb9As">Straight-line scalar optimizations</a> – These
+reduce redundancy within straight-line code.</p>
+</li>
+<li><p class="first"><a class="reference external" href="http://llvm.org/docs/doxygen/html/SpeculativeExecution_8cpp_source.html">Aggressive speculative execution</a>
+– This is mainly for promoting straight-line scalar optimizations, which are
+most effective on code along dominator paths.</p>
+</li>
+<li><p class="first"><a class="reference external" href="http://llvm.org/doxygen/NVPTXInferAddressSpaces_8cpp_source.html">Memory space inference</a> –
+In PTX, we can operate on pointers that are in a paricular “address space”
+(global, shared, constant, or local), or we can operate on pointers in the
+“generic” address space, which can point to anything. Operations in a
+non-generic address space are faster, but pointers in CUDA are not explicitly
+annotated with their address space, so it’s up to LLVM to infer it where
+possible.</p>
+</li>
+<li><p class="first"><a class="reference external" href="http://llvm.org/docs/doxygen/html/BypassSlowDivision_8cpp_source.html">Bypassing 64-bit divides</a> –
+This was an existing optimization that we enabled for the PTX backend.</p>
+<p>64-bit integer divides are much slower than 32-bit ones on NVIDIA GPUs.
+Many of the 64-bit divides in our benchmarks have a divisor and dividend
+which fit in 32-bits at runtime. This optimization provides a fast path for
+this common case.</p>
+</li>
+<li><p class="first">Aggressive loop unrooling and function inlining – Loop unrolling and
+function inlining need to be more aggressive for GPUs than for CPUs because
+control flow transfer in GPU is more expensive. More aggressive unrolling and
+inlining also promote other optimizations, such as constant propagation and
+SROA, which sometimes speed up code by over 10x.</p>
+<p>(Programmers can force unrolling and inline using clang’s <a class="reference external" href="http://clang.llvm.org/docs/AttributeReference.html#pragma-unroll-pragma-nounroll">loop unrolling pragmas</a>
+and <tt class="docutils literal"><span class="pre">__attribute__((always_inline))</span></tt>.)</p>
+</li>
+</ul>
+</div>
+<div class="section" id="publication">
+<h2><a class="toc-backref" href="#id16">Publication</a><a class="headerlink" href="#publication" title="Permalink to this headline">¶</a></h2>
+<p>The team at Google published a paper in CGO 2016 detailing the optimizations
+they’d made to clang/LLVM. Note that “gpucc” is no longer a meaningful name:
+The relevant tools are now just vanilla clang/LLVM.</p>
+<div class="line-block">
+<div class="line"><a class="reference external" href="http://dl.acm.org/citation.cfm?id=2854041">gpucc: An Open-Source GPGPU Compiler</a></div>
+<div class="line">Jingyue Wu, Artem Belevich, Eli Bendersky, Mark Heffernan, Chris Leary, Jacques Pienaar, Bjarke Roune, Rob Springer, Xuetian Weng, Robert Hundt</div>
+<div class="line"><em>Proceedings of the 2016 International Symposium on Code Generation and Optimization (CGO 2016)</em></div>
+<div class="line"><br /></div>
+<div class="line"><a class="reference external" href="http://wujingyue.com/docs/gpucc-talk.pdf">Slides from the CGO talk</a></div>
+<div class="line"><br /></div>
+<div class="line"><a class="reference external" href="http://wujingyue.com/docs/gpucc-tutorial.pdf">Tutorial given at CGO</a></div>
+</div>
+</div>
+<div class="section" id="obtaining-help">
+<h2><a class="toc-backref" href="#id17">Obtaining Help</a><a class="headerlink" href="#obtaining-help" title="Permalink to this headline">¶</a></h2>
+<p>To obtain help on LLVM in general and its CUDA support, see <a class="reference external" href="http://llvm.org/docs/#mailing-lists">the LLVM
+community</a>.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="ReportingGuide.html" title="Reporting Guide"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="CodeOfConduct.html" title="LLVM Community Code of Conduct"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CompilerWriterInfo.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CompilerWriterInfo.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CompilerWriterInfo.html (added)
+++ www-releases/trunk/4.0.0/docs/CompilerWriterInfo.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,300 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>Architecture & Platform Information for Compiler Writers — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="Extending LLVM: Adding instructions, intrinsics, types, etc." href="ExtendingLLVM.html" />
+ <link rel="prev" title="CommandLine 2.0 Library Manual" href="CommandLine.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="ExtendingLLVM.html" title="Extending LLVM: Adding instructions, intrinsics, types, etc."
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="CommandLine.html" title="CommandLine 2.0 Library Manual"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="architecture-platform-information-for-compiler-writers">
+<h1>Architecture & Platform Information for Compiler Writers<a class="headerlink" href="#architecture-platform-information-for-compiler-writers" title="Permalink to this headline">¶</a></h1>
+<div class="contents local topic" id="contents">
+<ul class="simple">
+<li><a class="reference internal" href="#hardware" id="id1">Hardware</a><ul>
+<li><a class="reference internal" href="#aarch64-arm" id="id2">AArch64 & ARM</a></li>
+<li><a class="reference internal" href="#itanium-ia64" id="id3">Itanium (ia64)</a></li>
+<li><a class="reference internal" href="#lanai" id="id4">Lanai</a></li>
+<li><a class="reference internal" href="#mips" id="id5">MIPS</a></li>
+<li><a class="reference internal" href="#powerpc" id="id6">PowerPC</a><ul>
+<li><a class="reference internal" href="#ibm-official-manuals-and-docs" id="id7">IBM - Official manuals and docs</a></li>
+<li><a class="reference internal" href="#other-documents-collections-notes" id="id8">Other documents, collections, notes</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#amdgpu" id="id9">AMDGPU</a></li>
+<li><a class="reference internal" href="#risc-v" id="id10">RISC-V</a></li>
+<li><a class="reference internal" href="#sparc" id="id11">SPARC</a></li>
+<li><a class="reference internal" href="#systemz" id="id12">SystemZ</a></li>
+<li><a class="reference internal" href="#x86" id="id13">X86</a></li>
+<li><a class="reference internal" href="#xcore" id="id14">XCore</a></li>
+<li><a class="reference internal" href="#hexagon" id="id15">Hexagon</a></li>
+<li><a class="reference internal" href="#other-relevant-lists" id="id16">Other relevant lists</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#abi" id="id17">ABI</a><ul>
+<li><a class="reference internal" href="#linux" id="id18">Linux</a></li>
+<li><a class="reference internal" href="#os-x" id="id19">OS X</a></li>
+<li><a class="reference internal" href="#windows" id="id20">Windows</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#nvptx" id="id21">NVPTX</a></li>
+<li><a class="reference internal" href="#miscellaneous-resources" id="id22">Miscellaneous Resources</a></li>
+</ul>
+</div>
+<div class="admonition note">
+<p class="first admonition-title">Note</p>
+<p class="last">This document is a work-in-progress. Additions and clarifications are
+welcome.</p>
+</div>
+<div class="section" id="hardware">
+<h2><a class="toc-backref" href="#id1">Hardware</a><a class="headerlink" href="#hardware" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="aarch64-arm">
+<h3><a class="toc-backref" href="#id2">AArch64 & ARM</a><a class="headerlink" href="#aarch64-arm" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.ddi0487a.h/index.html">ARMv8-A Architecture Reference Manual</a> (authentication required, free sign-up). This document covers both AArch64 and ARM instructions</li>
+<li><a class="reference external" href="http://infocenter.arm.com/help/topic/com.arm.doc.ddi0403e.b/index.html">ARMv7-M Architecture Reference Manual</a> (authentication required, free sign-up). This covers the Thumb2-only microcontrollers</li>
+<li><a class="reference external" href="http://infocenter.arm.com/help/topic/com.arm.doc.ddi0419c/index.html">ARMv6-M Architecture Reference Manual</a> (authentication required, free sign-up). This covers the Thumb1-only microcontrollers</li>
+<li><a class="reference external" href="http://infocenter.arm.com/help/topic/com.arm.doc.ihi0053c/IHI0053C_acle_2_0.pdf">ARM C Language Extensions</a></li>
+<li>AArch32 <a class="reference external" href="http://infocenter.arm.com/help/topic/com.arm.doc.ihi0045d/IHI0045D_ABI_addenda.pdf">ABI Addenda and Errata</a></li>
+</ul>
+</div>
+<div class="section" id="itanium-ia64">
+<h3><a class="toc-backref" href="#id3">Itanium (ia64)</a><a class="headerlink" href="#itanium-ia64" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="http://developer.intel.com/design/itanium2/documentation.htm">Itanium documentation</a></li>
+</ul>
+</div>
+<div class="section" id="lanai">
+<h3><a class="toc-backref" href="#id4">Lanai</a><a class="headerlink" href="#lanai" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="http://g.co/lanai/isa">Lanai Instruction Set Architecture</a></li>
+</ul>
+</div>
+<div class="section" id="mips">
+<h3><a class="toc-backref" href="#id5">MIPS</a><a class="headerlink" href="#mips" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="https://imgtec.com/mips/architectures/">MIPS Processor Architecture</a></li>
+<li><a class="reference external" href="http://techpubs.sgi.com/library/manuals/4000/007-4658-001/pdf/007-4658-001.pdf">MIPS 64-bit ELF Object File Specification</a></li>
+</ul>
+</div>
+<div class="section" id="powerpc">
+<h3><a class="toc-backref" href="#id6">PowerPC</a><a class="headerlink" href="#powerpc" title="Permalink to this headline">¶</a></h3>
+<div class="section" id="ibm-official-manuals-and-docs">
+<h4><a class="toc-backref" href="#id7">IBM - Official manuals and docs</a><a class="headerlink" href="#ibm-official-manuals-and-docs" title="Permalink to this headline">¶</a></h4>
+<ul class="simple">
+<li><a class="reference external" href="https://www.power.org/technology-introduction/standards-specifications">Power Instruction Set Architecture, Versions 2.03 through 2.06 (authentication required, free sign-up)</a></li>
+<li><a class="reference external" href="http://www.ibm.com/chips/techlib/techlib.nsf/techdocs/852569B20050FF7785256996007558C6">PowerPC Compiler Writer’s Guide</a></li>
+<li><a class="reference external" href="http://www.ibm.com/developerworks/linux/library/l-powarch/">Intro to PowerPC Architecture</a></li>
+<li><a class="reference external" href="http://www.ibm.com/chips/techlib/techlib.nsf/products/PowerPC">PowerPC Processor Manuals (embedded)</a></li>
+<li><a class="reference external" href="https://www.power.org/documentation/?document_company=105&document_category=all&publish_year=all&grid_order=DESC&grid_sort=title">Various IBM specifications and white papers</a></li>
+<li><a class="reference external" href="http://publibn.boulder.ibm.com/doc_link/en_US/a_doc_lib/aixassem/alangref/alangreftfrm.htm">IBM AIX/5L for POWER Assembly Reference</a></li>
+</ul>
+</div>
+<div class="section" id="other-documents-collections-notes">
+<h4><a class="toc-backref" href="#id8">Other documents, collections, notes</a><a class="headerlink" href="#other-documents-collections-notes" title="Permalink to this headline">¶</a></h4>
+<ul class="simple">
+<li><a class="reference external" href="http://penguinppc.org/dev/#library">PowerPC ABI documents</a></li>
+<li><a class="reference external" href="http://gcc.gnu.org/ml/gcc-patches/2003-09/msg00997.html">PowerPC64 alignment of long doubles (from GCC)</a></li>
+<li><a class="reference external" href="http://sources.redhat.com/ml/binutils/2002-04/msg00573.html">Long branch stubs for powerpc64-linux (from binutils)</a></li>
+</ul>
+</div>
+</div>
+<div class="section" id="amdgpu">
+<h3><a class="toc-backref" href="#id9">AMDGPU</a><a class="headerlink" href="#amdgpu" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="http://developer.amd.com/wordpress/media/2012/10/R600_Instruction_Set_Architecture.pdf">AMD R6xx shader ISA</a></li>
+<li><a class="reference external" href="http://developer.amd.com/wordpress/media/2012/10/R700-Family_Instruction_Set_Architecture.pdf">AMD R7xx shader ISA</a></li>
+<li><a class="reference external" href="http://developer.amd.com/wordpress/media/2012/10/AMD_Evergreen-Family_Instruction_Set_Architecture.pdf">AMD Evergreen shader ISA</a></li>
+<li><a class="reference external" href="http://developer.amd.com/wordpress/media/2012/10/AMD_HD_6900_Series_Instruction_Set_Architecture.pdf">AMD Cayman/Trinity shader ISA</a></li>
+<li><a class="reference external" href="http://developer.amd.com/wordpress/media/2012/12/AMD_Southern_Islands_Instruction_Set_Architecture.pdf">AMD Southern Islands Series ISA</a></li>
+<li><a class="reference external" href="http://developer.amd.com/wordpress/media/2013/07/AMD_Sea_Islands_Instruction_Set_Architecture.pdf">AMD Sea Islands Series ISA</a></li>
+<li><a class="reference external" href="http://amd-dev.wpengine.netdna-cdn.com/wordpress/media/2013/12/AMD_GCN3_Instruction_Set_Architecture_rev1.1.pdf">AMD GCN3 Instruction Set Architecture</a></li>
+<li><a class="reference external" href="http://developer.amd.com/download/AMD_Accelerated_Parallel_Processing_OpenCL_Programming_Guide.pdf">AMD GPU Programming Guide</a></li>
+<li><a class="reference external" href="http://developer.amd.com/tools/heterogeneous-computing/amd-accelerated-parallel-processing-app-sdk/documentation/">AMD Compute Resources</a></li>
+<li><a class="reference external" href="https://github.com/RadeonOpenCompute/ROCm-ComputeABI-Doc/blob/master/AMDGPU-ABI.md">AMDGPU Compute Application Binary Interface</a></li>
+</ul>
+</div>
+<div class="section" id="risc-v">
+<h3><a class="toc-backref" href="#id10">RISC-V</a><a class="headerlink" href="#risc-v" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="https://riscv.org/specifications/">RISC-V User-Level ISA Specification</a></li>
+</ul>
+</div>
+<div class="section" id="sparc">
+<h3><a class="toc-backref" href="#id11">SPARC</a><a class="headerlink" href="#sparc" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="http://sparc.org/standards">SPARC standards</a></li>
+<li><a class="reference external" href="http://sparc.org/standards/64.psabi.1.35.ps.Z">SPARC V9 ABI</a></li>
+<li><a class="reference external" href="http://sparc.org/standards/psABI3rd.pdf">SPARC V8 ABI</a></li>
+</ul>
+</div>
+<div class="section" id="systemz">
+<h3><a class="toc-backref" href="#id12">SystemZ</a><a class="headerlink" href="#systemz" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="http://www-01.ibm.com/support/docview.wss?uid=isg2b9de5f05a9d57819852571c500428f9a">z/Architecture Principles of Operation (registration required, free sign-up)</a></li>
+</ul>
+</div>
+<div class="section" id="x86">
+<h3><a class="toc-backref" href="#id13">X86</a><a class="headerlink" href="#x86" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="http://developer.amd.com/resources/developer-guides-manuals/">AMD processor manuals</a></li>
+<li><a class="reference external" href="http://www.intel.com/content/www/us/en/processors/architectures-software-developer-manuals.html">Intel 64 and IA-32 manuals</a></li>
+<li><a class="reference external" href="http://www.intel.com/design/itanium/documentation.htm?iid=ipp_srvr_proc_itanium2+techdocs">Intel Itanium documentation</a></li>
+<li><a class="reference external" href="https://github.com/hjl-tools/x86-psABI/wiki/X86-psABI">X86 and X86-64 SysV psABI</a></li>
+<li><a class="reference external" href="http://www.agner.org/optimize/calling_conventions.pdf">Calling conventions for different C++ compilers and operating systems</a></li>
+</ul>
+</div>
+<div class="section" id="xcore">
+<h3><a class="toc-backref" href="#id14">XCore</a><a class="headerlink" href="#xcore" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="https://www.xmos.com/en/download/public/The-XMOS-XS1-Architecture%28X7879A%29.pdf">The XMOS XS1 Architecture (ISA)</a></li>
+<li><a class="reference external" href="https://www.xmos.com/download/public/Tools-Development-Guide%28X9114A%29.pdf">Tools Development Guide (includes ABI)</a></li>
+</ul>
+</div>
+<div class="section" id="hexagon">
+<h3><a class="toc-backref" href="#id15">Hexagon</a><a class="headerlink" href="#hexagon" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="https://developer.qualcomm.com/software/hexagon-dsp-sdk/tools">Hexagon Programmer’s Reference Manuals and Hexagon ABI Specification (registration required, free sign-up)</a></li>
+</ul>
+</div>
+<div class="section" id="other-relevant-lists">
+<h3><a class="toc-backref" href="#id16">Other relevant lists</a><a class="headerlink" href="#other-relevant-lists" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="http://gcc.gnu.org/readings.html">GCC reading list</a></li>
+</ul>
+</div>
+</div>
+<div class="section" id="abi">
+<h2><a class="toc-backref" href="#id17">ABI</a><a class="headerlink" href="#abi" title="Permalink to this headline">¶</a></h2>
+<ul class="simple">
+<li><a class="reference external" href="http://www.sco.com/developers/gabi/latest/contents.html">System V Application Binary Interface</a></li>
+<li><a class="reference external" href="http://mentorembedded.github.io/cxx-abi/">Itanium C++ ABI</a></li>
+</ul>
+<div class="section" id="linux">
+<h3><a class="toc-backref" href="#id18">Linux</a><a class="headerlink" href="#linux" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="https://github.com/hjl-tools/linux-abi/wiki/Linux-Extensions-to-gABI">Linux extensions to gabi</a></li>
+<li><a class="reference external" href="http://www.linuxbase.org/spec/ELF/ppc64/">PowerPC 64-bit ELF ABI Supplement</a></li>
+<li><a class="reference external" href="http://infocenter.arm.com/help/topic/com.arm.doc.ihi0055a/IHI0055A_aapcs64.pdf">Procedure Call Standard for the AArch64 Architecture</a></li>
+<li><a class="reference external" href="http://infocenter.arm.com/help/topic/com.arm.doc.ihi0044e/IHI0044E_aaelf.pdf">ELF for the ARM Architecture</a></li>
+<li><a class="reference external" href="http://infocenter.arm.com/help/topic/com.arm.doc.ihi0056a/IHI0056A_aaelf64.pdf">ELF for the ARM 64-bit Architecture (AArch64)</a></li>
+<li><a class="reference external" href="http://legacy.redhat.com/pub/redhat/linux/7.1/es/os/s390x/doc/lzsabi0.pdf">System z ELF ABI Supplement</a></li>
+</ul>
+</div>
+<div class="section" id="os-x">
+<h3><a class="toc-backref" href="#id19">OS X</a><a class="headerlink" href="#os-x" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="http://developer.apple.com/documentation/Darwin/RuntimeArchitecture-date.html">Mach-O Runtime Architecture</a></li>
+<li><a class="reference external" href="http://www.unsanity.org/archives/000044.php">Notes on Mach-O ABI</a></li>
+</ul>
+</div>
+<div class="section" id="windows">
+<h3><a class="toc-backref" href="#id20">Windows</a><a class="headerlink" href="#windows" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li><a class="reference external" href="http://www.microsoft.com/whdc/system/platform/firmware/pecoff.mspx">Microsoft PE/COFF Specification</a></li>
+</ul>
+</div>
+</div>
+<div class="section" id="nvptx">
+<h2><a class="toc-backref" href="#id21">NVPTX</a><a class="headerlink" href="#nvptx" title="Permalink to this headline">¶</a></h2>
+<ul class="simple">
+<li><a class="reference external" href="http://docs.nvidia.com/cuda/index.html">CUDA Documentation</a> includes the PTX
+ISA and Driver API documentation</li>
+</ul>
+</div>
+<div class="section" id="miscellaneous-resources">
+<h2><a class="toc-backref" href="#id22">Miscellaneous Resources</a><a class="headerlink" href="#miscellaneous-resources" title="Permalink to this headline">¶</a></h2>
+<ul class="simple">
+<li><a class="reference external" href="http://www.nondot.org/sabre/os/articles/ExecutableFileFormats/">Executable File Format library</a></li>
+<li><a class="reference external" href="http://gcc.gnu.org/projects/prefetch.html">GCC prefetch project</a> page has a
+good survey of the prefetching capabilities of a variety of modern
+processors.</li>
+</ul>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="ExtendingLLVM.html" title="Extending LLVM: Adding instructions, intrinsics, types, etc."
+ >next</a> |</li>
+ <li class="right" >
+ <a href="CommandLine.html" title="CommandLine 2.0 Library Manual"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/Coroutines.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/Coroutines.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/Coroutines.html (added)
+++ www-releases/trunk/4.0.0/docs/Coroutines.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,1198 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>Coroutines in LLVM — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="Global Instruction Selection" href="GlobalISel.html" />
+ <link rel="prev" title="Machine IR (MIR) Format Reference Manual" href="MIRLangRef.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="GlobalISel.html" title="Global Instruction Selection"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="MIRLangRef.html" title="Machine IR (MIR) Format Reference Manual"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="coroutines-in-llvm">
+<h1>Coroutines in LLVM<a class="headerlink" href="#coroutines-in-llvm" title="Permalink to this headline">¶</a></h1>
+<div class="contents local topic" id="contents">
+<ul class="simple">
+<li><a class="reference internal" href="#introduction" id="id45">Introduction</a></li>
+<li><a class="reference internal" href="#coroutines-by-example" id="id46">Coroutines by Example</a><ul>
+<li><a class="reference internal" href="#coroutine-representation" id="id47">Coroutine Representation</a></li>
+<li><a class="reference internal" href="#coroutine-transformation" id="id48">Coroutine Transformation</a></li>
+<li><a class="reference internal" href="#avoiding-heap-allocations" id="id49">Avoiding Heap Allocations</a></li>
+<li><a class="reference internal" href="#multiple-suspend-points" id="id50">Multiple Suspend Points</a></li>
+<li><a class="reference internal" href="#distinct-save-and-suspend" id="id51">Distinct Save and Suspend</a></li>
+<li><a class="reference internal" href="#coroutine-promise" id="id52">Coroutine Promise</a></li>
+<li><a class="reference internal" href="#final-suspend" id="id53">Final Suspend</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#intrinsics" id="id54">Intrinsics</a><ul>
+<li><a class="reference internal" href="#coroutine-manipulation-intrinsics" id="id55">Coroutine Manipulation Intrinsics</a><ul>
+<li><a class="reference internal" href="#llvm-coro-destroy-intrinsic" id="id56">‘llvm.coro.destroy’ Intrinsic</a></li>
+<li><a class="reference internal" href="#llvm-coro-resume-intrinsic" id="id57">‘llvm.coro.resume’ Intrinsic</a></li>
+<li><a class="reference internal" href="#llvm-coro-done-intrinsic" id="id58">‘llvm.coro.done’ Intrinsic</a></li>
+<li><a class="reference internal" href="#llvm-coro-promise-intrinsic" id="id59">‘llvm.coro.promise’ Intrinsic</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#coroutine-structure-intrinsics" id="id60">Coroutine Structure Intrinsics</a><ul>
+<li><a class="reference internal" href="#llvm-coro-size-intrinsic" id="id61">‘llvm.coro.size’ Intrinsic</a></li>
+<li><a class="reference internal" href="#llvm-coro-begin-intrinsic" id="id62">‘llvm.coro.begin’ Intrinsic</a></li>
+<li><a class="reference internal" href="#llvm-coro-free-intrinsic" id="id63">‘llvm.coro.free’ Intrinsic</a></li>
+<li><a class="reference internal" href="#llvm-coro-alloc-intrinsic" id="id64">‘llvm.coro.alloc’ Intrinsic</a></li>
+<li><a class="reference internal" href="#llvm-coro-frame-intrinsic" id="id65">‘llvm.coro.frame’ Intrinsic</a></li>
+<li><a class="reference internal" href="#llvm-coro-id-intrinsic" id="id66">‘llvm.coro.id’ Intrinsic</a></li>
+<li><a class="reference internal" href="#llvm-coro-end-intrinsic" id="id67">‘llvm.coro.end’ Intrinsic</a></li>
+<li><a class="reference internal" href="#llvm-coro-suspend-intrinsic" id="id68">‘llvm.coro.suspend’ Intrinsic</a></li>
+<li><a class="reference internal" href="#llvm-coro-save-intrinsic" id="id69">‘llvm.coro.save’ Intrinsic</a></li>
+<li><a class="reference internal" href="#llvm-coro-param-intrinsic" id="id70">‘llvm.coro.param’ Intrinsic</a></li>
+</ul>
+</li>
+</ul>
+</li>
+<li><a class="reference internal" href="#coroutine-transformation-passes" id="id71">Coroutine Transformation Passes</a><ul>
+<li><a class="reference internal" href="#coroearly" id="id72">CoroEarly</a></li>
+<li><a class="reference internal" href="#corosplit" id="id73">CoroSplit</a></li>
+<li><a class="reference internal" href="#coroelide" id="id74">CoroElide</a></li>
+<li><a class="reference internal" href="#corocleanup" id="id75">CoroCleanup</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#areas-requiring-attention" id="id76">Areas Requiring Attention</a></li>
+</ul>
+</div>
+<div class="admonition warning">
+<p class="first admonition-title">Warning</p>
+<p class="last">This is a work in progress. Compatibility across LLVM releases is not
+guaranteed.</p>
+</div>
+<div class="section" id="introduction">
+<h2><a class="toc-backref" href="#id45">Introduction</a><a class="headerlink" href="#introduction" title="Permalink to this headline">¶</a></h2>
+<p id="coroutine-handle">LLVM coroutines are functions that have one or more <a class="reference internal" href="#suspend-points">suspend points</a>.
+When a suspend point is reached, the execution of a coroutine is suspended and
+control is returned back to its caller. A suspended coroutine can be resumed
+to continue execution from the last suspend point or it can be destroyed.</p>
+<p>In the following example, we call function <cite>f</cite> (which may or may not be a
+coroutine itself) that returns a handle to a suspended coroutine
+(<strong>coroutine handle</strong>) that is used by <cite>main</cite> to resume the coroutine twice and
+then destroy it:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">i32</span> <span class="vg">@main</span><span class="p">()</span> <span class="p">{</span>
+<span class="nl">entry:</span>
+ <span class="nv">%hdl</span> <span class="p">=</span> <span class="k">call</span> <span class="k">i8</span><span class="p">*</span> <span class="vg">@f</span><span class="p">(</span><span class="k">i32</span> <span class="m">4</span><span class="p">)</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@llvm.coro.resume</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%hdl</span><span class="p">)</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@llvm.coro.resume</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%hdl</span><span class="p">)</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@llvm.coro.destroy</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%hdl</span><span class="p">)</span>
+ <span class="k">ret</span> <span class="k">i32</span> <span class="m">0</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p id="coroutine-frame">In addition to the function stack frame which exists when a coroutine is
+executing, there is an additional region of storage that contains objects that
+keep the coroutine state when a coroutine is suspended. This region of storage
+is called <strong>coroutine frame</strong>. It is created when a coroutine is called and
+destroyed when a coroutine runs to completion or destroyed by a call to
+the <a class="reference internal" href="#coro-destroy">coro.destroy</a> intrinsic.</p>
+<p>An LLVM coroutine is represented as an LLVM function that has calls to
+<a class="reference internal" href="#coroutine-intrinsics">coroutine intrinsics</a> defining the structure of the coroutine.
+After lowering, a coroutine is split into several
+functions that represent three different ways of how control can enter the
+coroutine:</p>
+<ol class="arabic simple">
+<li>a ramp function, which represents an initial invocation of the coroutine that
+creates the coroutine frame and executes the coroutine code until it
+encounters a suspend point or reaches the end of the function;</li>
+<li>a coroutine resume function that is invoked when the coroutine is resumed;</li>
+<li>a coroutine destroy function that is invoked when the coroutine is destroyed.</li>
+</ol>
+<div class="admonition note">
+<p class="first admonition-title">Note</p>
+<p class="last">Splitting out resume and destroy functions are just one of the
+possible ways of lowering the coroutine. We chose it for initial
+implementation as it matches closely the mental model and results in
+reasonably nice code.</p>
+</div>
+</div>
+<div class="section" id="coroutines-by-example">
+<h2><a class="toc-backref" href="#id46">Coroutines by Example</a><a class="headerlink" href="#coroutines-by-example" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="coroutine-representation">
+<h3><a class="toc-backref" href="#id47">Coroutine Representation</a><a class="headerlink" href="#coroutine-representation" title="Permalink to this headline">¶</a></h3>
+<p>Let’s look at an example of an LLVM coroutine with the behavior sketched
+by the following pseudo-code.</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="kt">void</span> <span class="o">*</span><span class="nf">f</span><span class="p">(</span><span class="kt">int</span> <span class="n">n</span><span class="p">)</span> <span class="p">{</span>
+ <span class="k">for</span><span class="p">(;;)</span> <span class="p">{</span>
+ <span class="n">print</span><span class="p">(</span><span class="n">n</span><span class="o">++</span><span class="p">);</span>
+ <span class="o"><</span><span class="n">suspend</span><span class="o">></span> <span class="c1">// returns a coroutine handle on first suspend</span>
+ <span class="p">}</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>This coroutine calls some function <cite>print</cite> with value <cite>n</cite> as an argument and
+suspends execution. Every time this coroutine resumes, it calls <cite>print</cite> again with an argument one bigger than the last time. This coroutine never completes by itself and must be destroyed explicitly. If we use this coroutine with
+a <cite>main</cite> shown in the previous section. It will call <cite>print</cite> with values 4, 5
+and 6 after which the coroutine will be destroyed.</p>
+<p>The LLVM IR for this coroutine looks like this:</p>
+<div class="highlight-none"><div class="highlight"><pre>define i8* @f(i32 %n) {
+entry:
+ %id = call token @llvm.coro.id(i32 0, i8* null, i8* null, i8* null)
+ %size = call i32 @llvm.coro.size.i32()
+ %alloc = call i8* @malloc(i32 %size)
+ %hdl = call noalias i8* @llvm.coro.begin(token %id, i8* %alloc)
+ br label %loop
+loop:
+ %n.val = phi i32 [ %n, %entry ], [ %inc, %loop ]
+ %inc = add nsw i32 %n.val, 1
+ call void @print(i32 %n.val)
+ %0 = call i8 @llvm.coro.suspend(token none, i1 false)
+ switch i8 %0, label %suspend [i8 0, label %loop
+ i8 1, label %cleanup]
+cleanup:
+ %mem = call i8* @llvm.coro.free(token %id, i8* %hdl)
+ call void @free(i8* %mem)
+ br label %suspend
+suspend:
+ call void @llvm.coro.end(i8* %hdl, i1 false)
+ ret i8* %hdl
+}
+</pre></div>
+</div>
+<p>The <cite>entry</cite> block establishes the coroutine frame. The <a class="reference internal" href="#coro-size">coro.size</a> intrinsic is
+lowered to a constant representing the size required for the coroutine frame.
+The <a class="reference internal" href="#coro-begin">coro.begin</a> intrinsic initializes the coroutine frame and returns the
+coroutine handle. The second parameter of <cite>coro.begin</cite> is given a block of memory
+to be used if the coroutine frame needs to be allocated dynamically.
+The <a class="reference internal" href="#coro-id">coro.id</a> intrinsic serves as coroutine identity useful in cases when the
+<a class="reference internal" href="#coro-begin">coro.begin</a> intrinsic get duplicated by optimization passes such as
+jump-threading.</p>
+<p>The <cite>cleanup</cite> block destroys the coroutine frame. The <a class="reference internal" href="#coro-free">coro.free</a> intrinsic,
+given the coroutine handle, returns a pointer of the memory block to be freed or
+<cite>null</cite> if the coroutine frame was not allocated dynamically. The <cite>cleanup</cite>
+block is entered when coroutine runs to completion by itself or destroyed via
+call to the <a class="reference internal" href="#coro-destroy">coro.destroy</a> intrinsic.</p>
+<p>The <cite>suspend</cite> block contains code to be executed when coroutine runs to
+completion or suspended. The <a class="reference internal" href="#coro-end">coro.end</a> intrinsic marks the point where
+a coroutine needs to return control back to the caller if it is not an initial
+invocation of the coroutine.</p>
+<p>The <cite>loop</cite> blocks represents the body of the coroutine. The <a class="reference internal" href="#coro-suspend">coro.suspend</a>
+intrinsic in combination with the following switch indicates what happens to
+control flow when a coroutine is suspended (default case), resumed (case 0) or
+destroyed (case 1).</p>
+</div>
+<div class="section" id="coroutine-transformation">
+<h3><a class="toc-backref" href="#id48">Coroutine Transformation</a><a class="headerlink" href="#coroutine-transformation" title="Permalink to this headline">¶</a></h3>
+<p>One of the steps of coroutine lowering is building the coroutine frame. The
+def-use chains are analyzed to determine which objects need be kept alive across
+suspend points. In the coroutine shown in the previous section, use of virtual register
+<cite>%n.val</cite> is separated from the definition by a suspend point, therefore, it
+cannot reside on the stack frame since the latter goes away once the coroutine
+is suspended and control is returned back to the caller. An i32 slot is
+allocated in the coroutine frame and <cite>%n.val</cite> is spilled and reloaded from that
+slot as needed.</p>
+<p>We also store addresses of the resume and destroy functions so that the
+<cite>coro.resume</cite> and <cite>coro.destroy</cite> intrinsics can resume and destroy the coroutine
+when its identity cannot be determined statically at compile time. For our
+example, the coroutine frame will be:</p>
+<div class="highlight-text"><div class="highlight"><pre>%f.frame = type { void (%f.frame*)*, void (%f.frame*)*, i32 }
+</pre></div>
+</div>
+<p>After resume and destroy parts are outlined, function <cite>f</cite> will contain only the
+code responsible for creation and initialization of the coroutine frame and
+execution of the coroutine until a suspend point is reached:</p>
+<div class="highlight-none"><div class="highlight"><pre>define i8* @f(i32 %n) {
+entry:
+ %id = call token @llvm.coro.id(i32 0, i8* null, i8* null, i8* null)
+ %alloc = call noalias i8* @malloc(i32 24)
+ %0 = call noalias i8* @llvm.coro.begin(token %id, i8* %alloc)
+ %frame = bitcast i8* %0 to %f.frame*
+ %1 = getelementptr %f.frame, %f.frame* %frame, i32 0, i32 0
+ store void (%f.frame*)* @f.resume, void (%f.frame*)** %1
+ %2 = getelementptr %f.frame, %f.frame* %frame, i32 0, i32 1
+ store void (%f.frame*)* @f.destroy, void (%f.frame*)** %2
+
+ %inc = add nsw i32 %n, 1
+ %inc.spill.addr = getelementptr inbounds %f.Frame, %f.Frame* %FramePtr, i32 0, i32 2
+ store i32 %inc, i32* %inc.spill.addr
+ call void @print(i32 %n)
+
+ ret i8* %frame
+}
+</pre></div>
+</div>
+<p>Outlined resume part of the coroutine will reside in function <cite>f.resume</cite>:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">internal</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="vg">@f.resume</span><span class="p">(</span><span class="nv">%f.frame</span><span class="p">*</span> <span class="nv">%frame.ptr.resume</span><span class="p">)</span> <span class="p">{</span>
+<span class="nl">entry:</span>
+ <span class="nv">%inc.spill.addr</span> <span class="p">=</span> <span class="k">getelementptr</span> <span class="nv">%f.frame</span><span class="p">,</span> <span class="nv">%f.frame</span><span class="p">*</span> <span class="nv">%frame.ptr.resume</span><span class="p">,</span> <span class="k">i64</span> <span class="m">0</span><span class="p">,</span> <span class="k">i32</span> <span class="m">2</span>
+ <span class="nv">%inc.spill</span> <span class="p">=</span> <span class="k">load</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span><span class="p">*</span> <span class="nv">%inc.spill.addr</span><span class="p">,</span> <span class="k">align</span> <span class="m">4</span>
+ <span class="nv">%inc</span> <span class="p">=</span> <span class="k">add</span> <span class="k">i32</span> <span class="nv">%n.val</span><span class="p">,</span> <span class="m">1</span>
+ <span class="k">store</span> <span class="k">i32</span> <span class="nv">%inc</span><span class="p">,</span> <span class="k">i32</span><span class="p">*</span> <span class="nv">%inc.spill.addr</span><span class="p">,</span> <span class="k">align</span> <span class="m">4</span>
+ <span class="k">tail</span> <span class="k">call</span> <span class="kt">void</span> <span class="vg">@print</span><span class="p">(</span><span class="k">i32</span> <span class="nv">%inc</span><span class="p">)</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>Whereas function <cite>f.destroy</cite> will contain the cleanup code for the coroutine:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">internal</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="vg">@f.destroy</span><span class="p">(</span><span class="nv">%f.frame</span><span class="p">*</span> <span class="nv">%frame.ptr.destroy</span><span class="p">)</span> <span class="p">{</span>
+<span class="nl">entry:</span>
+ <span class="nv-Anonymous">%0</span> <span class="p">=</span> <span class="k">bitcast</span> <span class="nv">%f.frame</span><span class="p">*</span> <span class="nv">%frame.ptr.destroy</span> <span class="k">to</span> <span class="k">i8</span><span class="p">*</span>
+ <span class="k">tail</span> <span class="k">call</span> <span class="kt">void</span> <span class="vg">@free</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv-Anonymous">%0</span><span class="p">)</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+</div>
+<div class="section" id="avoiding-heap-allocations">
+<h3><a class="toc-backref" href="#id49">Avoiding Heap Allocations</a><a class="headerlink" href="#avoiding-heap-allocations" title="Permalink to this headline">¶</a></h3>
+<p>A particular coroutine usage pattern, which is illustrated by the <cite>main</cite>
+function in the overview section, where a coroutine is created, manipulated and
+destroyed by the same calling function, is common for coroutines implementing
+RAII idiom and is suitable for allocation elision optimization which avoid
+dynamic allocation by storing the coroutine frame as a static <cite>alloca</cite> in its
+caller.</p>
+<p>In the entry block, we will call <a class="reference internal" href="#coro-alloc">coro.alloc</a> intrinsic that will return <cite>true</cite>
+when dynamic allocation is required, and <cite>false</cite> if dynamic allocation is
+elided.</p>
+<div class="highlight-none"><div class="highlight"><pre>entry:
+ %id = call token @llvm.coro.id(i32 0, i8* null, i8* null, i8* null)
+ %need.dyn.alloc = call i1 @llvm.coro.alloc(token %id)
+ br i1 %need.dyn.alloc, label %dyn.alloc, label %coro.begin
+dyn.alloc:
+ %size = call i32 @llvm.coro.size.i32()
+ %alloc = call i8* @CustomAlloc(i32 %size)
+ br label %coro.begin
+coro.begin:
+ %phi = phi i8* [ null, %entry ], [ %alloc, %dyn.alloc ]
+ %hdl = call noalias i8* @llvm.coro.begin(token %id, i8* %phi)
+</pre></div>
+</div>
+<p>In the cleanup block, we will make freeing the coroutine frame conditional on
+<a class="reference internal" href="#coro-free">coro.free</a> intrinsic. If allocation is elided, <a class="reference internal" href="#coro-free">coro.free</a> returns <cite>null</cite>
+thus skipping the deallocation code:</p>
+<div class="highlight-text"><div class="highlight"><pre>cleanup:
+ %mem = call i8* @llvm.coro.free(token %id, i8* %hdl)
+ %need.dyn.free = icmp ne i8* %mem, null
+ br i1 %need.dyn.free, label %dyn.free, label %if.end
+dyn.free:
+ call void @CustomFree(i8* %mem)
+ br label %if.end
+if.end:
+ ...
+</pre></div>
+</div>
+<p>With allocations and deallocations represented as described as above, after
+coroutine heap allocation elision optimization, the resulting main will be:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">i32</span> <span class="vg">@main</span><span class="p">()</span> <span class="p">{</span>
+<span class="nl">entry:</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@print</span><span class="p">(</span><span class="k">i32</span> <span class="m">4</span><span class="p">)</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@print</span><span class="p">(</span><span class="k">i32</span> <span class="m">5</span><span class="p">)</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@print</span><span class="p">(</span><span class="k">i32</span> <span class="m">6</span><span class="p">)</span>
+ <span class="k">ret</span> <span class="k">i32</span> <span class="m">0</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+</div>
+<div class="section" id="multiple-suspend-points">
+<h3><a class="toc-backref" href="#id50">Multiple Suspend Points</a><a class="headerlink" href="#multiple-suspend-points" title="Permalink to this headline">¶</a></h3>
+<p>Let’s consider the coroutine that has more than one suspend point:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="kt">void</span> <span class="o">*</span><span class="nf">f</span><span class="p">(</span><span class="kt">int</span> <span class="n">n</span><span class="p">)</span> <span class="p">{</span>
+ <span class="k">for</span><span class="p">(;;)</span> <span class="p">{</span>
+ <span class="n">print</span><span class="p">(</span><span class="n">n</span><span class="o">++</span><span class="p">);</span>
+ <span class="o"><</span><span class="n">suspend</span><span class="o">></span>
+ <span class="n">print</span><span class="p">(</span><span class="o">-</span><span class="n">n</span><span class="p">);</span>
+ <span class="o"><</span><span class="n">suspend</span><span class="o">></span>
+ <span class="p">}</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>Matching LLVM code would look like (with the rest of the code remaining the same
+as the code in the previous section):</p>
+<div class="highlight-text"><div class="highlight"><pre>loop:
+ %n.addr = phi i32 [ %n, %entry ], [ %inc, %loop.resume ]
+ call void @print(i32 %n.addr) #4
+ %2 = call i8 @llvm.coro.suspend(token none, i1 false)
+ switch i8 %2, label %suspend [i8 0, label %loop.resume
+ i8 1, label %cleanup]
+loop.resume:
+ %inc = add nsw i32 %n.addr, 1
+ %sub = xor i32 %n.addr, -1
+ call void @print(i32 %sub)
+ %3 = call i8 @llvm.coro.suspend(token none, i1 false)
+ switch i8 %3, label %suspend [i8 0, label %loop
+ i8 1, label %cleanup]
+</pre></div>
+</div>
+<p>In this case, the coroutine frame would include a suspend index that will
+indicate at which suspend point the coroutine needs to resume. The resume
+function will use an index to jump to an appropriate basic block and will look
+as follows:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">internal</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="vg">@f.Resume</span><span class="p">(</span><span class="nv">%f.Frame</span><span class="p">*</span> <span class="nv">%FramePtr</span><span class="p">)</span> <span class="p">{</span>
+<span class="nl">entry.Resume:</span>
+ <span class="nv">%index.addr</span> <span class="p">=</span> <span class="k">getelementptr</span> <span class="k">inbounds</span> <span class="nv">%f.Frame</span><span class="p">,</span> <span class="nv">%f.Frame</span><span class="p">*</span> <span class="nv">%FramePtr</span><span class="p">,</span> <span class="k">i64</span> <span class="m">0</span><span class="p">,</span> <span class="k">i32</span> <span class="m">2</span>
+ <span class="nv">%index</span> <span class="p">=</span> <span class="k">load</span> <span class="k">i8</span><span class="p">,</span> <span class="k">i8</span><span class="p">*</span> <span class="nv">%index.addr</span><span class="p">,</span> <span class="k">align</span> <span class="m">1</span>
+ <span class="nv">%switch</span> <span class="p">=</span> <span class="k">icmp</span> <span class="k">eq</span> <span class="k">i8</span> <span class="nv">%index</span><span class="p">,</span> <span class="m">0</span>
+ <span class="nv">%n.addr</span> <span class="p">=</span> <span class="k">getelementptr</span> <span class="k">inbounds</span> <span class="nv">%f.Frame</span><span class="p">,</span> <span class="nv">%f.Frame</span><span class="p">*</span> <span class="nv">%FramePtr</span><span class="p">,</span> <span class="k">i64</span> <span class="m">0</span><span class="p">,</span> <span class="k">i32</span> <span class="m">3</span>
+ <span class="nv">%n</span> <span class="p">=</span> <span class="k">load</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span><span class="p">*</span> <span class="nv">%n.addr</span><span class="p">,</span> <span class="k">align</span> <span class="m">4</span>
+ <span class="k">br</span> <span class="k">i1</span> <span class="nv">%switch</span><span class="p">,</span> <span class="kt">label</span> <span class="nv">%loop.resume</span><span class="p">,</span> <span class="kt">label</span> <span class="nv">%loop</span>
+
+<span class="nl">loop.resume:</span>
+ <span class="nv">%sub</span> <span class="p">=</span> <span class="k">xor</span> <span class="k">i32</span> <span class="nv">%n</span><span class="p">,</span> <span class="m">-1</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@print</span><span class="p">(</span><span class="k">i32</span> <span class="nv">%sub</span><span class="p">)</span>
+ <span class="k">br</span> <span class="kt">label</span> <span class="nv">%suspend</span>
+<span class="nl">loop:</span>
+ <span class="nv">%inc</span> <span class="p">=</span> <span class="k">add</span> <span class="k">nsw</span> <span class="k">i32</span> <span class="nv">%n</span><span class="p">,</span> <span class="m">1</span>
+ <span class="k">store</span> <span class="k">i32</span> <span class="nv">%inc</span><span class="p">,</span> <span class="k">i32</span><span class="p">*</span> <span class="nv">%n.addr</span><span class="p">,</span> <span class="k">align</span> <span class="m">4</span>
+ <span class="k">tail</span> <span class="k">call</span> <span class="kt">void</span> <span class="vg">@print</span><span class="p">(</span><span class="k">i32</span> <span class="nv">%inc</span><span class="p">)</span>
+ <span class="k">br</span> <span class="kt">label</span> <span class="nv">%suspend</span>
+
+<span class="nl">suspend:</span>
+ <span class="nv">%storemerge</span> <span class="p">=</span> <span class="k">phi</span> <span class="k">i8</span> <span class="p">[</span> <span class="m">0</span><span class="p">,</span> <span class="nv">%loop</span> <span class="p">],</span> <span class="p">[</span> <span class="m">1</span><span class="p">,</span> <span class="nv">%loop.resume</span> <span class="p">]</span>
+ <span class="k">store</span> <span class="k">i8</span> <span class="nv">%storemerge</span><span class="p">,</span> <span class="k">i8</span><span class="p">*</span> <span class="nv">%index.addr</span><span class="p">,</span> <span class="k">align</span> <span class="m">1</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>If different cleanup code needs to get executed for different suspend points,
+a similar switch will be in the <cite>f.destroy</cite> function.</p>
+<div class="admonition note">
+<p class="first admonition-title">Note</p>
+<p class="last">Using suspend index in a coroutine state and having a switch in <cite>f.resume</cite> and
+<cite>f.destroy</cite> is one of the possible implementation strategies. We explored
+another option where a distinct <cite>f.resume1</cite>, <cite>f.resume2</cite>, etc. are created for
+every suspend point, and instead of storing an index, the resume and destroy
+function pointers are updated at every suspend. Early testing showed that the
+current approach is easier on the optimizer than the latter so it is a
+lowering strategy implemented at the moment.</p>
+</div>
+</div>
+<div class="section" id="distinct-save-and-suspend">
+<h3><a class="toc-backref" href="#id51">Distinct Save and Suspend</a><a class="headerlink" href="#distinct-save-and-suspend" title="Permalink to this headline">¶</a></h3>
+<p>In the previous example, setting a resume index (or some other state change that
+needs to happen to prepare a coroutine for resumption) happens at the same time as
+a suspension of a coroutine. However, in certain cases, it is necessary to control
+when coroutine is prepared for resumption and when it is suspended.</p>
+<p>In the following example, a coroutine represents some activity that is driven
+by completions of asynchronous operations <cite>async_op1</cite> and <cite>async_op2</cite> which get
+a coroutine handle as a parameter and resume the coroutine once async
+operation is finished.</p>
+<div class="highlight-text"><div class="highlight"><pre>void g() {
+ for (;;)
+ if (cond()) {
+ async_op1(<coroutine-handle>); // will resume once async_op1 completes
+ <suspend>
+ do_one();
+ }
+ else {
+ async_op2(<coroutine-handle>); // will resume once async_op2 completes
+ <suspend>
+ do_two();
+ }
+ }
+}
+</pre></div>
+</div>
+<p>In this case, coroutine should be ready for resumption prior to a call to
+<cite>async_op1</cite> and <cite>async_op2</cite>. The <a class="reference internal" href="#coro-save">coro.save</a> intrinsic is used to indicate a
+point when coroutine should be ready for resumption (namely, when a resume index
+should be stored in the coroutine frame, so that it can be resumed at the
+correct resume point):</p>
+<div class="highlight-text"><div class="highlight"><pre>if.true:
+ %save1 = call token @llvm.coro.save(i8* %hdl)
+ call void async_op1(i8* %hdl)
+ %suspend1 = call i1 @llvm.coro.suspend(token %save1, i1 false)
+ switch i8 %suspend1, label %suspend [i8 0, label %resume1
+ i8 1, label %cleanup]
+if.false:
+ %save2 = call token @llvm.coro.save(i8* %hdl)
+ call void async_op2(i8* %hdl)
+ %suspend2 = call i1 @llvm.coro.suspend(token %save2, i1 false)
+ switch i8 %suspend1, label %suspend [i8 0, label %resume2
+ i8 1, label %cleanup]
+</pre></div>
+</div>
+</div>
+<div class="section" id="coroutine-promise">
+<span id="id1"></span><h3><a class="toc-backref" href="#id52">Coroutine Promise</a><a class="headerlink" href="#coroutine-promise" title="Permalink to this headline">¶</a></h3>
+<p>A coroutine author or a frontend may designate a distinguished <cite>alloca</cite> that can
+be used to communicate with the coroutine. This distinguished alloca is called
+<strong>coroutine promise</strong> and is provided as the second parameter to the
+<a class="reference internal" href="#coro-id">coro.id</a> intrinsic.</p>
+<p>The following coroutine designates a 32 bit integer <cite>promise</cite> and uses it to
+store the current value produced by a coroutine.</p>
+<div class="highlight-text"><div class="highlight"><pre>define i8* @f(i32 %n) {
+entry:
+ %promise = alloca i32
+ %pv = bitcast i32* %promise to i8*
+ %id = call token @llvm.coro.id(i32 0, i8* %pv, i8* null, i8* null)
+ %need.dyn.alloc = call i1 @llvm.coro.alloc(token %id)
+ br i1 %need.dyn.alloc, label %dyn.alloc, label %coro.begin
+dyn.alloc:
+ %size = call i32 @llvm.coro.size.i32()
+ %alloc = call i8* @malloc(i32 %size)
+ br label %coro.begin
+coro.begin:
+ %phi = phi i8* [ null, %entry ], [ %alloc, %dyn.alloc ]
+ %hdl = call noalias i8* @llvm.coro.begin(token %id, i8* %phi)
+ br label %loop
+loop:
+ %n.val = phi i32 [ %n, %coro.begin ], [ %inc, %loop ]
+ %inc = add nsw i32 %n.val, 1
+ store i32 %n.val, i32* %promise
+ %0 = call i8 @llvm.coro.suspend(token none, i1 false)
+ switch i8 %0, label %suspend [i8 0, label %loop
+ i8 1, label %cleanup]
+cleanup:
+ %mem = call i8* @llvm.coro.free(token %id, i8* %hdl)
+ call void @free(i8* %mem)
+ br label %suspend
+suspend:
+ call void @llvm.coro.end(i8* %hdl, i1 false)
+ ret i8* %hdl
+}
+</pre></div>
+</div>
+<p>A coroutine consumer can rely on the <a class="reference internal" href="#coro-promise">coro.promise</a> intrinsic to access the
+coroutine promise.</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">i32</span> <span class="vg">@main</span><span class="p">()</span> <span class="p">{</span>
+<span class="nl">entry:</span>
+ <span class="nv">%hdl</span> <span class="p">=</span> <span class="k">call</span> <span class="k">i8</span><span class="p">*</span> <span class="vg">@f</span><span class="p">(</span><span class="k">i32</span> <span class="m">4</span><span class="p">)</span>
+ <span class="nv">%promise.addr.raw</span> <span class="p">=</span> <span class="k">call</span> <span class="k">i8</span><span class="p">*</span> <span class="vg">@llvm.coro.promise</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%hdl</span><span class="p">,</span> <span class="k">i32</span> <span class="m">4</span><span class="p">,</span> <span class="k">i1</span> <span class="k">false</span><span class="p">)</span>
+ <span class="nv">%promise.addr</span> <span class="p">=</span> <span class="k">bitcast</span> <span class="k">i8</span><span class="p">*</span> <span class="nv">%promise.addr.raw</span> <span class="k">to</span> <span class="k">i32</span><span class="p">*</span>
+ <span class="nv">%val0</span> <span class="p">=</span> <span class="k">load</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span><span class="p">*</span> <span class="nv">%promise.addr</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@print</span><span class="p">(</span><span class="k">i32</span> <span class="nv">%val0</span><span class="p">)</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@llvm.coro.resume</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%hdl</span><span class="p">)</span>
+ <span class="nv">%val1</span> <span class="p">=</span> <span class="k">load</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span><span class="p">*</span> <span class="nv">%promise.addr</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@print</span><span class="p">(</span><span class="k">i32</span> <span class="nv">%val1</span><span class="p">)</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@llvm.coro.resume</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%hdl</span><span class="p">)</span>
+ <span class="nv">%val2</span> <span class="p">=</span> <span class="k">load</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span><span class="p">*</span> <span class="nv">%promise.addr</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@print</span><span class="p">(</span><span class="k">i32</span> <span class="nv">%val2</span><span class="p">)</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@llvm.coro.destroy</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%hdl</span><span class="p">)</span>
+ <span class="k">ret</span> <span class="k">i32</span> <span class="m">0</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>After example in this section is compiled, result of the compilation will be:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">i32</span> <span class="vg">@main</span><span class="p">()</span> <span class="p">{</span>
+<span class="nl">entry:</span>
+ <span class="k">tail</span> <span class="k">call</span> <span class="kt">void</span> <span class="vg">@print</span><span class="p">(</span><span class="k">i32</span> <span class="m">4</span><span class="p">)</span>
+ <span class="k">tail</span> <span class="k">call</span> <span class="kt">void</span> <span class="vg">@print</span><span class="p">(</span><span class="k">i32</span> <span class="m">5</span><span class="p">)</span>
+ <span class="k">tail</span> <span class="k">call</span> <span class="kt">void</span> <span class="vg">@print</span><span class="p">(</span><span class="k">i32</span> <span class="m">6</span><span class="p">)</span>
+ <span class="k">ret</span> <span class="k">i32</span> <span class="m">0</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+</div>
+<div class="section" id="final-suspend">
+<span id="final"></span><span id="id2"></span><h3><a class="toc-backref" href="#id53">Final Suspend</a><a class="headerlink" href="#final-suspend" title="Permalink to this headline">¶</a></h3>
+<p>A coroutine author or a frontend may designate a particular suspend to be final,
+by setting the second argument of the <a class="reference internal" href="#coro-suspend">coro.suspend</a> intrinsic to <cite>true</cite>.
+Such a suspend point has two properties:</p>
+<ul class="simple">
+<li>it is possible to check whether a suspended coroutine is at the final suspend
+point via <a class="reference internal" href="#coro-done">coro.done</a> intrinsic;</li>
+<li>a resumption of a coroutine stopped at the final suspend point leads to
+undefined behavior. The only possible action for a coroutine at a final
+suspend point is destroying it via <a class="reference internal" href="#coro-destroy">coro.destroy</a> intrinsic.</li>
+</ul>
+<p>From the user perspective, the final suspend point represents an idea of a
+coroutine reaching the end. From the compiler perspective, it is an optimization
+opportunity for reducing number of resume points (and therefore switch cases) in
+the resume function.</p>
+<p>The following is an example of a function that keeps resuming the coroutine
+until the final suspend point is reached after which point the coroutine is
+destroyed:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">i32</span> <span class="vg">@main</span><span class="p">()</span> <span class="p">{</span>
+<span class="nl">entry:</span>
+ <span class="nv">%hdl</span> <span class="p">=</span> <span class="k">call</span> <span class="k">i8</span><span class="p">*</span> <span class="vg">@f</span><span class="p">(</span><span class="k">i32</span> <span class="m">4</span><span class="p">)</span>
+ <span class="k">br</span> <span class="kt">label</span> <span class="nv">%while</span>
+<span class="nl">while:</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@llvm.coro.resume</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%hdl</span><span class="p">)</span>
+ <span class="nv">%done</span> <span class="p">=</span> <span class="k">call</span> <span class="k">i1</span> <span class="vg">@llvm.coro.done</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%hdl</span><span class="p">)</span>
+ <span class="k">br</span> <span class="k">i1</span> <span class="nv">%done</span><span class="p">,</span> <span class="kt">label</span> <span class="nv">%end</span><span class="p">,</span> <span class="kt">label</span> <span class="nv">%while</span>
+<span class="nl">end:</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@llvm.coro.destroy</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%hdl</span><span class="p">)</span>
+ <span class="k">ret</span> <span class="k">i32</span> <span class="m">0</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>Usually, final suspend point is a frontend injected suspend point that does not
+correspond to any explicitly authored suspend point of the high level language.
+For example, for a Python generator that has only one suspend point:</p>
+<div class="highlight-python"><div class="highlight"><pre><span class="k">def</span> <span class="nf">coroutine</span><span class="p">(</span><span class="n">n</span><span class="p">):</span>
+ <span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="n">n</span><span class="p">):</span>
+ <span class="k">yield</span> <span class="n">i</span>
+</pre></div>
+</div>
+<p>Python frontend would inject two more suspend points, so that the actual code
+looks like this:</p>
+<div class="highlight-c"><div class="highlight"><pre><span class="kt">void</span><span class="o">*</span> <span class="nf">coroutine</span><span class="p">(</span><span class="kt">int</span> <span class="n">n</span><span class="p">)</span> <span class="p">{</span>
+ <span class="kt">int</span> <span class="n">current_value</span><span class="p">;</span>
+ <span class="o"><</span><span class="n">designate</span> <span class="n">current_value</span> <span class="n">to</span> <span class="n">be</span> <span class="n">coroutine</span> <span class="n">promise</span><span class="o">></span>
+ <span class="o"><</span><span class="n">SUSPEND</span><span class="o">></span> <span class="c1">// injected suspend point, so that the coroutine starts suspended</span>
+ <span class="k">for</span> <span class="p">(</span><span class="kt">int</span> <span class="n">i</span> <span class="o">=</span> <span class="mi">0</span><span class="p">;</span> <span class="n">i</span> <span class="o"><</span> <span class="n">n</span><span class="p">;</span> <span class="o">++</span><span class="n">i</span><span class="p">)</span> <span class="p">{</span>
+ <span class="n">current_value</span> <span class="o">=</span> <span class="n">i</span><span class="p">;</span> <span class="o"><</span><span class="n">SUSPEND</span><span class="o">></span><span class="p">;</span> <span class="c1">// corresponds to "yield i"</span>
+ <span class="p">}</span>
+ <span class="o"><</span><span class="n">SUSPEND</span> <span class="n">final</span><span class="o">=</span><span class="nb">true</span><span class="o">></span> <span class="c1">// injected final suspend point</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>and python iterator <cite>__next__</cite> would look like:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="kt">int</span> <span class="nf">__next__</span><span class="p">(</span><span class="kt">void</span><span class="o">*</span> <span class="n">hdl</span><span class="p">)</span> <span class="p">{</span>
+ <span class="n">coro</span><span class="p">.</span><span class="n">resume</span><span class="p">(</span><span class="n">hdl</span><span class="p">);</span>
+ <span class="k">if</span> <span class="p">(</span><span class="n">coro</span><span class="p">.</span><span class="n">done</span><span class="p">(</span><span class="n">hdl</span><span class="p">))</span> <span class="k">throw</span> <span class="n">StopIteration</span><span class="p">();</span>
+ <span class="k">return</span> <span class="o">*</span><span class="p">(</span><span class="kt">int</span><span class="o">*</span><span class="p">)</span><span class="n">coro</span><span class="p">.</span><span class="n">promise</span><span class="p">(</span><span class="n">hdl</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="nb">false</span><span class="p">);</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+</div>
+</div>
+<div class="section" id="intrinsics">
+<h2><a class="toc-backref" href="#id54">Intrinsics</a><a class="headerlink" href="#intrinsics" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="coroutine-manipulation-intrinsics">
+<h3><a class="toc-backref" href="#id55">Coroutine Manipulation Intrinsics</a><a class="headerlink" href="#coroutine-manipulation-intrinsics" title="Permalink to this headline">¶</a></h3>
+<p>Intrinsics described in this section are used to manipulate an existing
+coroutine. They can be used in any function which happen to have a pointer
+to a <a class="reference internal" href="#coroutine-frame">coroutine frame</a> or a pointer to a <a class="reference internal" href="#coroutine-promise">coroutine promise</a>.</p>
+<div class="section" id="llvm-coro-destroy-intrinsic">
+<span id="coro-destroy"></span><h4><a class="toc-backref" href="#id56">‘llvm.coro.destroy’ Intrinsic</a><a class="headerlink" href="#llvm-coro-destroy-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="section" id="syntax">
+<h5>Syntax:<a class="headerlink" href="#syntax" title="Permalink to this headline">¶</a></h5>
+<div class="highlight-python"><div class="highlight"><pre>declare void @llvm.coro.destroy(i8* <handle>)
+</pre></div>
+</div>
+</div>
+<div class="section" id="overview">
+<h5>Overview:<a class="headerlink" href="#overview" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.destroy</span></tt>‘ intrinsic destroys a suspended
+coroutine.</p>
+</div>
+<div class="section" id="arguments">
+<h5>Arguments:<a class="headerlink" href="#arguments" title="Permalink to this headline">¶</a></h5>
+<p>The argument is a coroutine handle to a suspended coroutine.</p>
+</div>
+<div class="section" id="semantics">
+<h5>Semantics:<a class="headerlink" href="#semantics" title="Permalink to this headline">¶</a></h5>
+<p>When possible, the <cite>coro.destroy</cite> intrinsic is replaced with a direct call to
+the coroutine destroy function. Otherwise it is replaced with an indirect call
+based on the function pointer for the destroy function stored in the coroutine
+frame. Destroying a coroutine that is not suspended leads to undefined behavior.</p>
+</div>
+</div>
+<div class="section" id="llvm-coro-resume-intrinsic">
+<span id="coro-resume"></span><h4><a class="toc-backref" href="#id57">‘llvm.coro.resume’ Intrinsic</a><a class="headerlink" href="#llvm-coro-resume-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare void @llvm.coro.resume(i8* <handle>)
+</pre></div>
+</div>
+<div class="section" id="id3">
+<h5>Overview:<a class="headerlink" href="#id3" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.resume</span></tt>‘ intrinsic resumes a suspended coroutine.</p>
+</div>
+<div class="section" id="id4">
+<h5>Arguments:<a class="headerlink" href="#id4" title="Permalink to this headline">¶</a></h5>
+<p>The argument is a handle to a suspended coroutine.</p>
+</div>
+<div class="section" id="id5">
+<h5>Semantics:<a class="headerlink" href="#id5" title="Permalink to this headline">¶</a></h5>
+<p>When possible, the <cite>coro.resume</cite> intrinsic is replaced with a direct call to the
+coroutine resume function. Otherwise it is replaced with an indirect call based
+on the function pointer for the resume function stored in the coroutine frame.
+Resuming a coroutine that is not suspended leads to undefined behavior.</p>
+</div>
+</div>
+<div class="section" id="llvm-coro-done-intrinsic">
+<span id="coro-done"></span><h4><a class="toc-backref" href="#id58">‘llvm.coro.done’ Intrinsic</a><a class="headerlink" href="#llvm-coro-done-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare i1 @llvm.coro.done(i8* <handle>)
+</pre></div>
+</div>
+<div class="section" id="id6">
+<h5>Overview:<a class="headerlink" href="#id6" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.done</span></tt>‘ intrinsic checks whether a suspended coroutine is at
+the final suspend point or not.</p>
+</div>
+<div class="section" id="id7">
+<h5>Arguments:<a class="headerlink" href="#id7" title="Permalink to this headline">¶</a></h5>
+<p>The argument is a handle to a suspended coroutine.</p>
+</div>
+<div class="section" id="id8">
+<h5>Semantics:<a class="headerlink" href="#id8" title="Permalink to this headline">¶</a></h5>
+<p>Using this intrinsic on a coroutine that does not have a <a class="reference internal" href="#final-suspend">final suspend</a> point
+or on a coroutine that is not suspended leads to undefined behavior.</p>
+</div>
+</div>
+<div class="section" id="llvm-coro-promise-intrinsic">
+<span id="coro-promise"></span><h4><a class="toc-backref" href="#id59">‘llvm.coro.promise’ Intrinsic</a><a class="headerlink" href="#llvm-coro-promise-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare i8* @llvm.coro.promise(i8* <ptr>, i32 <alignment>, i1 <from>)
+</pre></div>
+</div>
+<div class="section" id="id9">
+<h5>Overview:<a class="headerlink" href="#id9" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.promise</span></tt>‘ intrinsic obtains a pointer to a
+<a class="reference internal" href="#coroutine-promise">coroutine promise</a> given a coroutine handle and vice versa.</p>
+</div>
+<div class="section" id="id10">
+<h5>Arguments:<a class="headerlink" href="#id10" title="Permalink to this headline">¶</a></h5>
+<p>The first argument is a handle to a coroutine if <cite>from</cite> is false. Otherwise,
+it is a pointer to a coroutine promise.</p>
+<p>The second argument is an alignment requirements of the promise.
+If a frontend designated <cite>%promise = alloca i32</cite> as a promise, the alignment
+argument to <cite>coro.promise</cite> should be the alignment of <cite>i32</cite> on the target
+platform. If a frontend designated <cite>%promise = alloca i32, align 16</cite> as a
+promise, the alignment argument should be 16.
+This argument only accepts constants.</p>
+<p>The third argument is a boolean indicating a direction of the transformation.
+If <cite>from</cite> is true, the intrinsic returns a coroutine handle given a pointer
+to a promise. If <cite>from</cite> is false, the intrinsics return a pointer to a promise
+from a coroutine handle. This argument only accepts constants.</p>
+</div>
+<div class="section" id="id11">
+<h5>Semantics:<a class="headerlink" href="#id11" title="Permalink to this headline">¶</a></h5>
+<p>Using this intrinsic on a coroutine that does not have a coroutine promise
+leads to undefined behavior. It is possible to read and modify coroutine
+promise of the coroutine which is currently executing. The coroutine author and
+a coroutine user are responsible to makes sure there is no data races.</p>
+</div>
+<div class="section" id="example">
+<h5>Example:<a class="headerlink" href="#example" title="Permalink to this headline">¶</a></h5>
+<div class="highlight-text"><div class="highlight"><pre>define i8* @f(i32 %n) {
+entry:
+ %promise = alloca i32
+ %pv = bitcast i32* %promise to i8*
+ ; the second argument to coro.id points to the coroutine promise.
+ %id = call token @llvm.coro.id(i32 0, i8* %pv, i8* null, i8* null)
+ ...
+ %hdl = call noalias i8* @llvm.coro.begin(token %id, i8* %alloc)
+ ...
+ store i32 42, i32* %promise ; store something into the promise
+ ...
+ ret i8* %hdl
+}
+
+define i32 @main() {
+entry:
+ %hdl = call i8* @f(i32 4) ; starts the coroutine and returns its handle
+ %promise.addr.raw = call i8* @llvm.coro.promise(i8* %hdl, i32 4, i1 false)
+ %promise.addr = bitcast i8* %promise.addr.raw to i32*
+ %val = load i32, i32* %promise.addr ; load a value from the promise
+ call void @print(i32 %val)
+ call void @llvm.coro.destroy(i8* %hdl)
+ ret i32 0
+}
+</pre></div>
+</div>
+</div>
+</div>
+</div>
+<div class="section" id="coroutine-structure-intrinsics">
+<span id="coroutine-intrinsics"></span><h3><a class="toc-backref" href="#id60">Coroutine Structure Intrinsics</a><a class="headerlink" href="#coroutine-structure-intrinsics" title="Permalink to this headline">¶</a></h3>
+<p>Intrinsics described in this section are used within a coroutine to describe
+the coroutine structure. They should not be used outside of a coroutine.</p>
+<div class="section" id="llvm-coro-size-intrinsic">
+<span id="coro-size"></span><h4><a class="toc-backref" href="#id61">‘llvm.coro.size’ Intrinsic</a><a class="headerlink" href="#llvm-coro-size-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare i32 @llvm.coro.size.i32()
+declare i64 @llvm.coro.size.i64()
+</pre></div>
+</div>
+<div class="section" id="id12">
+<h5>Overview:<a class="headerlink" href="#id12" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.size</span></tt>‘ intrinsic returns the number of bytes
+required to store a <a class="reference internal" href="#coroutine-frame">coroutine frame</a>.</p>
+</div>
+<div class="section" id="id13">
+<h5>Arguments:<a class="headerlink" href="#id13" title="Permalink to this headline">¶</a></h5>
+<p>None</p>
+</div>
+<div class="section" id="id14">
+<h5>Semantics:<a class="headerlink" href="#id14" title="Permalink to this headline">¶</a></h5>
+<p>The <cite>coro.size</cite> intrinsic is lowered to a constant representing the size of
+the coroutine frame.</p>
+</div>
+</div>
+<div class="section" id="llvm-coro-begin-intrinsic">
+<span id="coro-begin"></span><h4><a class="toc-backref" href="#id62">‘llvm.coro.begin’ Intrinsic</a><a class="headerlink" href="#llvm-coro-begin-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare i8* @llvm.coro.begin(token <id>, i8* <mem>)
+</pre></div>
+</div>
+<div class="section" id="id15">
+<h5>Overview:<a class="headerlink" href="#id15" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.begin</span></tt>‘ intrinsic returns an address of the coroutine frame.</p>
+</div>
+<div class="section" id="id16">
+<h5>Arguments:<a class="headerlink" href="#id16" title="Permalink to this headline">¶</a></h5>
+<p>The first argument is a token returned by a call to ‘<tt class="docutils literal"><span class="pre">llvm.coro.id</span></tt>‘
+identifying the coroutine.</p>
+<p>The second argument is a pointer to a block of memory where coroutine frame
+will be stored if it is allocated dynamically.</p>
+</div>
+<div class="section" id="id17">
+<h5>Semantics:<a class="headerlink" href="#id17" title="Permalink to this headline">¶</a></h5>
+<p>Depending on the alignment requirements of the objects in the coroutine frame
+and/or on the codegen compactness reasons the pointer returned from <cite>coro.begin</cite>
+may be at offset to the <cite>%mem</cite> argument. (This could be beneficial if
+instructions that express relative access to data can be more compactly encoded
+with small positive and negative offsets).</p>
+<p>A frontend should emit exactly one <cite>coro.begin</cite> intrinsic per coroutine.</p>
+</div>
+</div>
+<div class="section" id="llvm-coro-free-intrinsic">
+<span id="coro-free"></span><h4><a class="toc-backref" href="#id63">‘llvm.coro.free’ Intrinsic</a><a class="headerlink" href="#llvm-coro-free-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare i8* @llvm.coro.free(token %id, i8* <frame>)
+</pre></div>
+</div>
+<div class="section" id="id18">
+<h5>Overview:<a class="headerlink" href="#id18" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.free</span></tt>‘ intrinsic returns a pointer to a block of memory where
+coroutine frame is stored or <cite>null</cite> if this instance of a coroutine did not use
+dynamically allocated memory for its coroutine frame.</p>
+</div>
+<div class="section" id="id19">
+<h5>Arguments:<a class="headerlink" href="#id19" title="Permalink to this headline">¶</a></h5>
+<p>The first argument is a token returned by a call to ‘<tt class="docutils literal"><span class="pre">llvm.coro.id</span></tt>‘
+identifying the coroutine.</p>
+<p>The second argument is a pointer to the coroutine frame. This should be the same
+pointer that was returned by prior <cite>coro.begin</cite> call.</p>
+</div>
+<div class="section" id="example-custom-deallocation-function">
+<h5>Example (custom deallocation function):<a class="headerlink" href="#example-custom-deallocation-function" title="Permalink to this headline">¶</a></h5>
+<div class="highlight-text"><div class="highlight"><pre>cleanup:
+ %mem = call i8* @llvm.coro.free(token %id, i8* %frame)
+ %mem_not_null = icmp ne i8* %mem, null
+ br i1 %mem_not_null, label %if.then, label %if.end
+if.then:
+ call void @CustomFree(i8* %mem)
+ br label %if.end
+if.end:
+ ret void
+</pre></div>
+</div>
+</div>
+<div class="section" id="example-standard-deallocation-functions">
+<h5>Example (standard deallocation functions):<a class="headerlink" href="#example-standard-deallocation-functions" title="Permalink to this headline">¶</a></h5>
+<div class="highlight-text"><div class="highlight"><pre>cleanup:
+ %mem = call i8* @llvm.coro.free(token %id, i8* %frame)
+ call void @free(i8* %mem)
+ ret void
+</pre></div>
+</div>
+</div>
+</div>
+<div class="section" id="llvm-coro-alloc-intrinsic">
+<span id="coro-alloc"></span><h4><a class="toc-backref" href="#id64">‘llvm.coro.alloc’ Intrinsic</a><a class="headerlink" href="#llvm-coro-alloc-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare i1 @llvm.coro.alloc(token <id>)
+</pre></div>
+</div>
+<div class="section" id="id20">
+<h5>Overview:<a class="headerlink" href="#id20" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.alloc</span></tt>‘ intrinsic returns <cite>true</cite> if dynamic allocation is
+required to obtain a memory for the corutine frame and <cite>false</cite> otherwise.</p>
+</div>
+<div class="section" id="id21">
+<h5>Arguments:<a class="headerlink" href="#id21" title="Permalink to this headline">¶</a></h5>
+<p>The first argument is a token returned by a call to ‘<tt class="docutils literal"><span class="pre">llvm.coro.id</span></tt>‘
+identifying the coroutine.</p>
+</div>
+<div class="section" id="id22">
+<h5>Semantics:<a class="headerlink" href="#id22" title="Permalink to this headline">¶</a></h5>
+<p>A frontend should emit at most one <cite>coro.alloc</cite> intrinsic per coroutine.
+The intrinsic is used to suppress dynamic allocation of the coroutine frame
+when possible.</p>
+</div>
+<div class="section" id="id23">
+<h5>Example:<a class="headerlink" href="#id23" title="Permalink to this headline">¶</a></h5>
+<div class="highlight-text"><div class="highlight"><pre>entry:
+ %id = call token @llvm.coro.id(i32 0, i8* null, i8* null, i8* null)
+ %dyn.alloc.required = call i1 @llvm.coro.alloc(token %id)
+ br i1 %dyn.alloc.required, label %coro.alloc, label %coro.begin
+
+coro.alloc:
+ %frame.size = call i32 @llvm.coro.size()
+ %alloc = call i8* @MyAlloc(i32 %frame.size)
+ br label %coro.begin
+
+coro.begin:
+ %phi = phi i8* [ null, %entry ], [ %alloc, %coro.alloc ]
+ %frame = call i8* @llvm.coro.begin(token %id, i8* %phi)
+</pre></div>
+</div>
+</div>
+</div>
+<div class="section" id="llvm-coro-frame-intrinsic">
+<span id="coro-frame"></span><h4><a class="toc-backref" href="#id65">‘llvm.coro.frame’ Intrinsic</a><a class="headerlink" href="#llvm-coro-frame-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare i8* @llvm.coro.frame()
+</pre></div>
+</div>
+<div class="section" id="id24">
+<h5>Overview:<a class="headerlink" href="#id24" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.frame</span></tt>‘ intrinsic returns an address of the coroutine frame of
+the enclosing coroutine.</p>
+</div>
+<div class="section" id="id25">
+<h5>Arguments:<a class="headerlink" href="#id25" title="Permalink to this headline">¶</a></h5>
+<p>None</p>
+</div>
+<div class="section" id="id26">
+<h5>Semantics:<a class="headerlink" href="#id26" title="Permalink to this headline">¶</a></h5>
+<p>This intrinsic is lowered to refer to the <a class="reference internal" href="#coro-begin">coro.begin</a> instruction. This is
+a frontend convenience intrinsic that makes it easier to refer to the
+coroutine frame.</p>
+</div>
+</div>
+<div class="section" id="llvm-coro-id-intrinsic">
+<span id="coro-id"></span><h4><a class="toc-backref" href="#id66">‘llvm.coro.id’ Intrinsic</a><a class="headerlink" href="#llvm-coro-id-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare token @llvm.coro.id(i32 <align>, i8* <promise>, i8* <coroaddr>,
+ i8* <fnaddrs>)
+</pre></div>
+</div>
+<div class="section" id="id27">
+<h5>Overview:<a class="headerlink" href="#id27" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.id</span></tt>‘ intrinsic returns a token identifying a coroutine.</p>
+</div>
+<div class="section" id="id28">
+<h5>Arguments:<a class="headerlink" href="#id28" title="Permalink to this headline">¶</a></h5>
+<p>The first argument provides information on the alignment of the memory returned
+by the allocation function and given to <cite>coro.begin</cite> by the first argument. If
+this argument is 0, the memory is assumed to be aligned to 2 * sizeof(i8*).
+This argument only accepts constants.</p>
+<p>The second argument, if not <cite>null</cite>, designates a particular alloca instruction
+to be a <a class="reference internal" href="#coroutine-promise">coroutine promise</a>.</p>
+<p>The third argument is <cite>null</cite> coming out of the frontend. The CoroEarly pass sets
+this argument to point to the function this coro.id belongs to.</p>
+<p>The fourth argument is <cite>null</cite> before coroutine is split, and later is replaced
+to point to a private global constant array containing function pointers to
+outlined resume and destroy parts of the coroutine.</p>
+</div>
+<div class="section" id="id29">
+<h5>Semantics:<a class="headerlink" href="#id29" title="Permalink to this headline">¶</a></h5>
+<p>The purpose of this intrinsic is to tie together <cite>coro.id</cite>, <cite>coro.alloc</cite> and
+<cite>coro.begin</cite> belonging to the same coroutine to prevent optimization passes from
+duplicating any of these instructions unless entire body of the coroutine is
+duplicated.</p>
+<p>A frontend should emit exactly one <cite>coro.id</cite> intrinsic per coroutine.</p>
+</div>
+</div>
+<div class="section" id="llvm-coro-end-intrinsic">
+<span id="coro-end"></span><h4><a class="toc-backref" href="#id67">‘llvm.coro.end’ Intrinsic</a><a class="headerlink" href="#llvm-coro-end-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare void @llvm.coro.end(i8* <handle>, i1 <unwind>)
+</pre></div>
+</div>
+<div class="section" id="id30">
+<h5>Overview:<a class="headerlink" href="#id30" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.end</span></tt>‘ marks the point where execution of the resume part of
+the coroutine should end and control returns back to the caller.</p>
+</div>
+<div class="section" id="id31">
+<h5>Arguments:<a class="headerlink" href="#id31" title="Permalink to this headline">¶</a></h5>
+<p>The first argument should refer to the coroutine handle of the enclosing coroutine.</p>
+<p>The second argument should be <cite>true</cite> if this coro.end is in the block that is
+part of the unwind sequence leaving the coroutine body due to exception prior to
+the first reaching any suspend points, and <cite>false</cite> otherwise.</p>
+</div>
+<div class="section" id="id32">
+<h5>Semantics:<a class="headerlink" href="#id32" title="Permalink to this headline">¶</a></h5>
+<p>The <a class="reference internal" href="#coro-end">coro.end</a> intrinsic is a no-op during an initial invocation of the
+coroutine. When the coroutine resumes, the intrinsic marks the point when
+coroutine need to return control back to the caller.</p>
+<p>This intrinsic is removed by the CoroSplit pass when a coroutine is split into
+the start, resume and destroy parts. In start part, the intrinsic is removed,
+in resume and destroy parts, it is replaced with <cite>ret void</cite> instructions and
+the rest of the block containing <cite>coro.end</cite> instruction is discarded.</p>
+<p>In landing pads it is replaced with an appropriate instruction to unwind to
+caller.</p>
+<p>A frontend is allowed to supply null as the first parameter, in this case
+<cite>coro-early</cite> pass will replace the null with an appropriate coroutine handle
+value.</p>
+</div>
+</div>
+<div class="section" id="llvm-coro-suspend-intrinsic">
+<span id="suspend-points"></span><span id="coro-suspend"></span><h4><a class="toc-backref" href="#id68">‘llvm.coro.suspend’ Intrinsic</a><a class="headerlink" href="#llvm-coro-suspend-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare i8 @llvm.coro.suspend(token <save>, i1 <final>)
+</pre></div>
+</div>
+<div class="section" id="id33">
+<h5>Overview:<a class="headerlink" href="#id33" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.suspend</span></tt>‘ marks the point where execution of the coroutine
+need to get suspended and control returned back to the caller.
+Conditional branches consuming the result of this intrinsic lead to basic blocks
+where coroutine should proceed when suspended (-1), resumed (0) or destroyed
+(1).</p>
+</div>
+<div class="section" id="id34">
+<h5>Arguments:<a class="headerlink" href="#id34" title="Permalink to this headline">¶</a></h5>
+<p>The first argument refers to a token of <cite>coro.save</cite> intrinsic that marks the
+point when coroutine state is prepared for suspension. If <cite>none</cite> token is passed,
+the intrinsic behaves as if there were a <cite>coro.save</cite> immediately preceding
+the <cite>coro.suspend</cite> intrinsic.</p>
+<p>The second argument indicates whether this suspension point is <a class="reference internal" href="#final">final</a>.
+The second argument only accepts constants. If more than one suspend point is
+designated as final, the resume and destroy branches should lead to the same
+basic blocks.</p>
+</div>
+<div class="section" id="example-normal-suspend-point">
+<h5>Example (normal suspend point):<a class="headerlink" href="#example-normal-suspend-point" title="Permalink to this headline">¶</a></h5>
+<div class="highlight-text"><div class="highlight"><pre>%0 = call i8 @llvm.coro.suspend(token none, i1 false)
+switch i8 %0, label %suspend [i8 0, label %resume
+ i8 1, label %cleanup]
+</pre></div>
+</div>
+</div>
+<div class="section" id="example-final-suspend-point">
+<h5>Example (final suspend point):<a class="headerlink" href="#example-final-suspend-point" title="Permalink to this headline">¶</a></h5>
+<div class="highlight-text"><div class="highlight"><pre>while.end:
+ %s.final = call i8 @llvm.coro.suspend(token none, i1 true)
+ switch i8 %s.final, label %suspend [i8 0, label %trap
+ i8 1, label %cleanup]
+trap:
+ call void @llvm.trap()
+ unreachable
+</pre></div>
+</div>
+</div>
+<div class="section" id="id35">
+<h5>Semantics:<a class="headerlink" href="#id35" title="Permalink to this headline">¶</a></h5>
+<p>If a coroutine that was suspended at the suspend point marked by this intrinsic
+is resumed via <a class="reference internal" href="#coro-resume">coro.resume</a> the control will transfer to the basic block
+of the 0-case. If it is resumed via <a class="reference internal" href="#coro-destroy">coro.destroy</a>, it will proceed to the
+basic block indicated by the 1-case. To suspend, coroutine proceed to the
+default label.</p>
+<p>If suspend intrinsic is marked as final, it can consider the <cite>true</cite> branch
+unreachable and can perform optimizations that can take advantage of that fact.</p>
+</div>
+</div>
+<div class="section" id="llvm-coro-save-intrinsic">
+<span id="coro-save"></span><h4><a class="toc-backref" href="#id69">‘llvm.coro.save’ Intrinsic</a><a class="headerlink" href="#llvm-coro-save-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare token @llvm.coro.save(i8* <handle>)
+</pre></div>
+</div>
+<div class="section" id="id36">
+<h5>Overview:<a class="headerlink" href="#id36" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.save</span></tt>‘ marks the point where a coroutine need to update its
+state to prepare for resumption to be considered suspended (and thus eligible
+for resumption).</p>
+</div>
+<div class="section" id="id37">
+<h5>Arguments:<a class="headerlink" href="#id37" title="Permalink to this headline">¶</a></h5>
+<p>The first argument points to a coroutine handle of the enclosing coroutine.</p>
+</div>
+<div class="section" id="id38">
+<h5>Semantics:<a class="headerlink" href="#id38" title="Permalink to this headline">¶</a></h5>
+<p>Whatever coroutine state changes are required to enable resumption of
+the coroutine from the corresponding suspend point should be done at the point
+of <cite>coro.save</cite> intrinsic.</p>
+</div>
+<div class="section" id="id39">
+<h5>Example:<a class="headerlink" href="#id39" title="Permalink to this headline">¶</a></h5>
+<p>Separate save and suspend points are necessary when a coroutine is used to
+represent an asynchronous control flow driven by callbacks representing
+completions of asynchronous operations.</p>
+<p>In such a case, a coroutine should be ready for resumption prior to a call to
+<cite>async_op</cite> function that may trigger resumption of a coroutine from the same or
+a different thread possibly prior to <cite>async_op</cite> call returning control back
+to the coroutine:</p>
+<div class="highlight-text"><div class="highlight"><pre>%save1 = call token @llvm.coro.save(i8* %hdl)
+call void async_op1(i8* %hdl)
+%suspend1 = call i1 @llvm.coro.suspend(token %save1, i1 false)
+switch i8 %suspend1, label %suspend [i8 0, label %resume1
+ i8 1, label %cleanup]
+</pre></div>
+</div>
+</div>
+</div>
+<div class="section" id="llvm-coro-param-intrinsic">
+<span id="coro-param"></span><h4><a class="toc-backref" href="#id70">‘llvm.coro.param’ Intrinsic</a><a class="headerlink" href="#llvm-coro-param-intrinsic" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-python"><div class="highlight"><pre>declare i1 @llvm.coro.param(i8* <original>, i8* <copy>)
+</pre></div>
+</div>
+<div class="section" id="id40">
+<h5>Overview:<a class="headerlink" href="#id40" title="Permalink to this headline">¶</a></h5>
+<p>The ‘<tt class="docutils literal"><span class="pre">llvm.coro.param</span></tt>‘ is used by a frontend to mark up the code used to
+construct and destruct copies of the parameters. If the optimizer discovers that
+a particular parameter copy is not used after any suspends, it can remove the
+construction and destruction of the copy by replacing corresponding coro.param
+with <cite>i1 false</cite> and replacing any use of the <cite>copy</cite> with the <cite>original</cite>.</p>
+</div>
+<div class="section" id="id41">
+<h5>Arguments:<a class="headerlink" href="#id41" title="Permalink to this headline">¶</a></h5>
+<p>The first argument points to an <cite>alloca</cite> storing the value of a parameter to a
+coroutine.</p>
+<p>The second argument points to an <cite>alloca</cite> storing the value of the copy of that
+parameter.</p>
+</div>
+<div class="section" id="id42">
+<h5>Semantics:<a class="headerlink" href="#id42" title="Permalink to this headline">¶</a></h5>
+<p>The optimizer is free to always replace this intrinsic with <cite>i1 true</cite>.</p>
+<p>The optimizer is also allowed to replace it with <cite>i1 false</cite> provided that the
+parameter copy is only used prior to control flow reaching any of the suspend
+points. The code that would be DCE’d if the <cite>coro.param</cite> is replaced with
+<cite>i1 false</cite> is not considered to be a use of the parameter copy.</p>
+<p>The frontend can emit this intrinsic if its language rules allow for this
+optimization.</p>
+</div>
+<div class="section" id="id43">
+<h5>Example:<a class="headerlink" href="#id43" title="Permalink to this headline">¶</a></h5>
+<p>Consider the following example. A coroutine takes two parameters <cite>a</cite> and <cite>b</cite>
+that has a destructor and a move constructor.</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">struct</span> <span class="n">A</span> <span class="p">{</span> <span class="o">~</span><span class="n">A</span><span class="p">();</span> <span class="n">A</span><span class="p">(</span><span class="n">A</span><span class="o">&&</span><span class="p">);</span> <span class="kt">bool</span> <span class="nf">foo</span><span class="p">();</span> <span class="kt">void</span> <span class="nf">bar</span><span class="p">();</span> <span class="p">};</span>
+
+<span class="n">task</span><span class="o"><</span><span class="kt">int</span><span class="o">></span> <span class="n">f</span><span class="p">(</span><span class="n">A</span> <span class="n">a</span><span class="p">,</span> <span class="n">A</span> <span class="n">b</span><span class="p">)</span> <span class="p">{</span>
+ <span class="k">if</span> <span class="p">(</span><span class="n">a</span><span class="p">.</span><span class="n">foo</span><span class="p">())</span>
+ <span class="k">return</span> <span class="mi">42</span><span class="p">;</span>
+
+ <span class="n">a</span><span class="p">.</span><span class="n">bar</span><span class="p">();</span>
+ <span class="n">co_await</span> <span class="nf">read_async</span><span class="p">();</span> <span class="c1">// introduces suspend point</span>
+ <span class="n">b</span><span class="p">.</span><span class="n">bar</span><span class="p">();</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>Note that, uses of <cite>b</cite> is used after a suspend point and thus must be copied
+into a coroutine frame, whereas <cite>a</cite> does not have to, since it never used
+after suspend.</p>
+<p>A frontend can create parameter copies for <cite>a</cite> and <cite>b</cite> as follows:</p>
+<div class="highlight-text"><div class="highlight"><pre>task<int> f(A a', A b') {
+ a = alloca A;
+ b = alloca A;
+ // move parameters to its copies
+ if (coro.param(a', a)) A::A(a, A&& a');
+ if (coro.param(b', b)) A::A(b, A&& b');
+ ...
+ // destroy parameters copies
+ if (coro.param(a', a)) A::~A(a);
+ if (coro.param(b', b)) A::~A(b);
+}
+</pre></div>
+</div>
+<p>The optimizer can replace coro.param(a’,a) with <cite>i1 false</cite> and replace all uses
+of <cite>a</cite> with <cite>a’</cite>, since it is not used after suspend.</p>
+<p>The optimizer must replace coro.param(b’, b) with <cite>i1 true</cite>, since <cite>b</cite> is used
+after suspend and therefore, it has to reside in the coroutine frame.</p>
+</div>
+</div>
+</div>
+</div>
+<div class="section" id="coroutine-transformation-passes">
+<h2><a class="toc-backref" href="#id71">Coroutine Transformation Passes</a><a class="headerlink" href="#coroutine-transformation-passes" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="coroearly">
+<h3><a class="toc-backref" href="#id72">CoroEarly</a><a class="headerlink" href="#coroearly" title="Permalink to this headline">¶</a></h3>
+<p>The pass CoroEarly lowers coroutine intrinsics that hide the details of the
+structure of the coroutine frame, but, otherwise not needed to be preserved to
+help later coroutine passes. This pass lowers <a class="reference internal" href="#coro-frame">coro.frame</a>, <a class="reference internal" href="#coro-done">coro.done</a>,
+and <a class="reference internal" href="#coro-promise">coro.promise</a> intrinsics.</p>
+</div>
+<div class="section" id="corosplit">
+<span id="id44"></span><h3><a class="toc-backref" href="#id73">CoroSplit</a><a class="headerlink" href="#corosplit" title="Permalink to this headline">¶</a></h3>
+<p>The pass CoroSplit buides coroutine frame and outlines resume and destroy parts
+into separate functions.</p>
+</div>
+<div class="section" id="coroelide">
+<h3><a class="toc-backref" href="#id74">CoroElide</a><a class="headerlink" href="#coroelide" title="Permalink to this headline">¶</a></h3>
+<p>The pass CoroElide examines if the inlined coroutine is eligible for heap
+allocation elision optimization. If so, it replaces
+<cite>coro.begin</cite> intrinsic with an address of a coroutine frame placed on its caller
+and replaces <cite>coro.alloc</cite> and <cite>coro.free</cite> intrinsics with <cite>false</cite> and <cite>null</cite>
+respectively to remove the deallocation code.
+This pass also replaces <cite>coro.resume</cite> and <cite>coro.destroy</cite> intrinsics with direct
+calls to resume and destroy functions for a particular coroutine where possible.</p>
+</div>
+<div class="section" id="corocleanup">
+<h3><a class="toc-backref" href="#id75">CoroCleanup</a><a class="headerlink" href="#corocleanup" title="Permalink to this headline">¶</a></h3>
+<p>This pass runs late to lower all coroutine related intrinsics not replaced by
+earlier passes.</p>
+</div>
+</div>
+<div class="section" id="areas-requiring-attention">
+<h2><a class="toc-backref" href="#id76">Areas Requiring Attention</a><a class="headerlink" href="#areas-requiring-attention" title="Permalink to this headline">¶</a></h2>
+<ol class="arabic simple">
+<li>A coroutine frame is bigger than it could be. Adding stack packing and stack
+coloring like optimization on the coroutine frame will result in tighter
+coroutine frames.</li>
+<li>Take advantage of the lifetime intrinsics for the data that goes into the
+coroutine frame. Leave lifetime intrinsics as is for the data that stays in
+allocas.</li>
+<li>The CoroElide optimization pass relies on coroutine ramp function to be
+inlined. It would be beneficial to split the ramp function further to
+increase the chance that it will get inlined into its caller.</li>
+<li>Design a convention that would make it possible to apply coroutine heap
+elision optimization across ABI boundaries.</li>
+<li>Cannot handle coroutines with <cite>inalloca</cite> parameters (used in x86 on Windows).</li>
+<li>Alignment is ignored by coro.begin and coro.free intrinsics.</li>
+<li>Make required changes to make sure that coroutine optimizations work with
+LTO.</li>
+<li>More tests, more tests, more tests</li>
+</ol>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="GlobalISel.html" title="Global Instruction Selection"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="MIRLangRef.html" title="Machine IR (MIR) Format Reference Manual"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/CoverageMappingFormat.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/CoverageMappingFormat.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/CoverageMappingFormat.html (added)
+++ www-releases/trunk/4.0.0/docs/CoverageMappingFormat.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,660 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>LLVM Code Coverage Mapping Format — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="Garbage Collection Safepoints in LLVM" href="Statepoints.html" />
+ <link rel="prev" title="Using ARM NEON instructions in big endian mode" href="BigEndianNEON.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="Statepoints.html" title="Garbage Collection Safepoints in LLVM"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="BigEndianNEON.html" title="Using ARM NEON instructions in big endian mode"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-code-coverage-mapping-format">
+<h1>LLVM Code Coverage Mapping Format<a class="headerlink" href="#llvm-code-coverage-mapping-format" title="Permalink to this headline">¶</a></h1>
+<div class="contents local topic" id="contents">
+<ul class="simple">
+<li><a class="reference internal" href="#introduction" id="id11">Introduction</a></li>
+<li><a class="reference internal" href="#quick-start" id="id12">Quick Start</a></li>
+<li><a class="reference internal" href="#high-level-overview" id="id13">High Level Overview</a></li>
+<li><a class="reference internal" href="#advanced-concepts" id="id14">Advanced Concepts</a><ul>
+<li><a class="reference internal" href="#mapping-region" id="id15">Mapping Region</a><ul>
+<li><a class="reference internal" href="#source-range" id="id16">Source Range:</a></li>
+<li><a class="reference internal" href="#file-id" id="id17">File ID:</a></li>
+<li><a class="reference internal" href="#counter" id="id18">Counter:</a></li>
+</ul>
+</li>
+</ul>
+</li>
+<li><a class="reference internal" href="#llvm-ir-representation" id="id19">LLVM IR Representation</a><ul>
+<li><a class="reference internal" href="#coverage-mapping-header" id="id20">Coverage Mapping Header:</a></li>
+<li><a class="reference internal" href="#function-record" id="id21">Function record:</a></li>
+<li><a class="reference internal" href="#encoded-data" id="id22">Encoded data:</a><ul>
+<li><a class="reference internal" href="#dissecting-the-sample" id="id23">Dissecting the sample:</a></li>
+</ul>
+</li>
+</ul>
+</li>
+<li><a class="reference internal" href="#encoding" id="id24">Encoding</a><ul>
+<li><a class="reference internal" href="#types" id="id25">Types</a><ul>
+<li><a class="reference internal" href="#leb128" id="id26">LEB128</a></li>
+<li><a class="reference internal" href="#strings" id="id27">Strings</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#file-id-mapping" id="id28">File ID Mapping</a></li>
+<li><a class="reference internal" href="#id3" id="id29">Counter</a><ul>
+<li><a class="reference internal" href="#tag" id="id30">Tag:</a></li>
+<li><a class="reference internal" href="#data" id="id31">Data:</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#counter-expressions" id="id32">Counter Expressions</a></li>
+<li><a class="reference internal" href="#mapping-regions" id="id33">Mapping Regions</a><ul>
+<li><a class="reference internal" href="#sub-array-of-regions" id="id34">Sub-Array of Regions</a></li>
+<li><a class="reference internal" href="#id6" id="id35">Mapping Region</a></li>
+<li><a class="reference internal" href="#header" id="id36">Header</a><ul>
+<li><a class="reference internal" href="#id8" id="id37">Counter:</a></li>
+<li><a class="reference internal" href="#pseudo-counter" id="id38">Pseudo-Counter:</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#id9" id="id39">Source Range</a></li>
+</ul>
+</li>
+</ul>
+</li>
+</ul>
+</div>
+<div class="section" id="introduction">
+<h2><a class="toc-backref" href="#id11">Introduction</a><a class="headerlink" href="#introduction" title="Permalink to this headline">¶</a></h2>
+<p>LLVM’s code coverage mapping format is used to provide code coverage
+analysis using LLVM’s and Clang’s instrumenation based profiling
+(Clang’s <tt class="docutils literal"><span class="pre">-fprofile-instr-generate</span></tt> option).</p>
+<p>This document is aimed at those who use LLVM’s code coverage mapping to provide
+code coverage analysis for their own programs, and for those who would like
+to know how it works under the hood. A prior knowledge of how Clang’s profile
+guided optimization works is useful, but not required.</p>
+<p>We start by showing how to use LLVM and Clang for code coverage analysis,
+then we briefly desribe LLVM’s code coverage mapping format and the
+way that Clang and LLVM’s code coverage tool work with this format. After
+the basics are down, more advanced features of the coverage mapping format
+are discussed - such as the data structures, LLVM IR representation and
+the binary encoding.</p>
+</div>
+<div class="section" id="quick-start">
+<h2><a class="toc-backref" href="#id12">Quick Start</a><a class="headerlink" href="#quick-start" title="Permalink to this headline">¶</a></h2>
+<p>Here’s a short story that describes how to generate code coverage overview
+for a sample source file called <em>test.c</em>.</p>
+<ul>
+<li><p class="first">First, compile an instrumented version of your program using Clang’s
+<tt class="docutils literal"><span class="pre">-fprofile-instr-generate</span></tt> option with the additional <tt class="docutils literal"><span class="pre">-fcoverage-mapping</span></tt>
+option:</p>
+<p><tt class="docutils literal"><span class="pre">clang</span> <span class="pre">-o</span> <span class="pre">test</span> <span class="pre">-fprofile-instr-generate</span> <span class="pre">-fcoverage-mapping</span> <span class="pre">test.c</span></tt></p>
+</li>
+<li><p class="first">Then, run the instrumented binary. The runtime will produce a file called
+<em>default.profraw</em> containing the raw profile instrumentation data:</p>
+<p><tt class="docutils literal"><span class="pre">./test</span></tt></p>
+</li>
+<li><p class="first">After that, merge the profile data using the <em>llvm-profdata</em> tool:</p>
+<p><tt class="docutils literal"><span class="pre">llvm-profdata</span> <span class="pre">merge</span> <span class="pre">-o</span> <span class="pre">test.profdata</span> <span class="pre">default.profraw</span></tt></p>
+</li>
+<li><p class="first">Finally, run LLVM’s code coverage tool (<em>llvm-cov</em>) to produce the code
+coverage overview for the sample source file:</p>
+<p><tt class="docutils literal"><span class="pre">llvm-cov</span> <span class="pre">show</span> <span class="pre">./test</span> <span class="pre">-instr-profile=test.profdata</span> <span class="pre">test.c</span></tt></p>
+</li>
+</ul>
+</div>
+<div class="section" id="high-level-overview">
+<h2><a class="toc-backref" href="#id13">High Level Overview</a><a class="headerlink" href="#high-level-overview" title="Permalink to this headline">¶</a></h2>
+<p>LLVM’s code coverage mapping format is designed to be a self contained
+data format, that can be embedded into the LLVM IR and object files.
+It’s described in this document as a <strong>mapping</strong> format because its goal is
+to store the data that is required for a code coverage tool to map between
+the specific source ranges in a file and the execution counts obtained
+after running the instrumented version of the program.</p>
+<p>The mapping data is used in two places in the code coverage process:</p>
+<ol class="arabic simple">
+<li>When clang compiles a source file with <tt class="docutils literal"><span class="pre">-fcoverage-mapping</span></tt>, it
+generates the mapping information that describes the mapping between the
+source ranges and the profiling instrumentation counters.
+This information gets embedded into the LLVM IR and conveniently
+ends up in the final executable file when the program is linked.</li>
+<li>It is also used by <em>llvm-cov</em> - the mapping information is extracted from an
+object file and is used to associate the execution counts (the values of the
+profile instrumentation counters), and the source ranges in a file.
+After that, the tool is able to generate various code coverage reports
+for the program.</li>
+</ol>
+<p>The coverage mapping format aims to be a “universal format” that would be
+suitable for usage by any frontend, and not just by Clang. It also aims to
+provide the frontend the possibility of generating the minimal coverage mapping
+data in order to reduce the size of the IR and object files - for example,
+instead of emitting mapping information for each statement in a function, the
+frontend is allowed to group the statements with the same execution count into
+regions of code, and emit the mapping information only for those regions.</p>
+</div>
+<div class="section" id="advanced-concepts">
+<h2><a class="toc-backref" href="#id14">Advanced Concepts</a><a class="headerlink" href="#advanced-concepts" title="Permalink to this headline">¶</a></h2>
+<p>The remainder of this guide is meant to give you insight into the way the
+coverage mapping format works.</p>
+<p>The coverage mapping format operates on a per-function level as the
+profile instrumentation counters are associated with a specific function.
+For each function that requires code coverage, the frontend has to create
+coverage mapping data that can map between the source code ranges and
+the profile instrumentation counters for that function.</p>
+<div class="section" id="mapping-region">
+<h3><a class="toc-backref" href="#id15">Mapping Region</a><a class="headerlink" href="#mapping-region" title="Permalink to this headline">¶</a></h3>
+<p>The function’s coverage mapping data contains an array of mapping regions.
+A mapping region stores the <a class="reference internal" href="#source-code-range">source code range</a> that is covered by this region,
+the <a class="reference internal" href="#coverage-file-id">file id</a>, the <a class="reference internal" href="#coverage-mapping-counter">coverage mapping counter</a> and
+the region’s kind.
+There are several kinds of mapping regions:</p>
+<ul>
+<li><p class="first">Code regions associate portions of source code and <a class="reference internal" href="#coverage-mapping-counters">coverage mapping
+counters</a>. They make up the majority of the mapping regions. They are used
+by the code coverage tool to compute the execution counts for lines,
+highlight the regions of code that were never executed, and to obtain
+the various code coverage statistics for a function.
+For example:</p>
+<p><span class="raw-html"><pre class='highlight' style='line-height:initial;'><span>int main(int argc, const char *argv[]) </span><span style='background-color:#4A789C'>{ </span> <span class='c1'>// Code Region from 1:40 to 9:2</span>
+<span style='background-color:#4A789C'> </span>
+<span style='background-color:#4A789C'> if (argc > 1) </span><span style='background-color:#85C1F5'>{ </span> <span class='c1'>// Code Region from 3:17 to 5:4</span>
+<span style='background-color:#85C1F5'> printf("%s\n", argv[1]); </span>
+<span style='background-color:#85C1F5'> }</span><span style='background-color:#4A789C'> else </span><span style='background-color:#F6D55D'>{ </span> <span class='c1'>// Code Region from 5:10 to 7:4</span>
+<span style='background-color:#F6D55D'> printf("\n"); </span>
+<span style='background-color:#F6D55D'> }</span><span style='background-color:#4A789C'> </span>
+<span style='background-color:#4A789C'> return 0; </span>
+<span style='background-color:#4A789C'>}</span>
+</pre></span></p>
+</li>
+<li><p class="first">Skipped regions are used to represent source ranges that were skipped
+by Clang’s preprocessor. They don’t associate with
+<a class="reference internal" href="#coverage-mapping-counters">coverage mapping counters</a>, as the frontend knows that they are never
+executed. They are used by the code coverage tool to mark the skipped lines
+inside a function as non-code lines that don’t have execution counts.
+For example:</p>
+<p><span class="raw-html"><pre class='highlight' style='line-height:initial;'><span>int main() </span><span style='background-color:#4A789C'>{ </span> <span class='c1'>// Code Region from 1:12 to 6:2</span>
+<span style='background-color:#85C1F5'>#ifdef DEBUG </span> <span class='c1'>// Skipped Region from 2:1 to 4:2</span>
+<span style='background-color:#85C1F5'> printf("Hello world"); </span>
+<span style='background-color:#85C1F5'>#</span><span style='background-color:#4A789C'>endif </span>
+<span style='background-color:#4A789C'> return 0; </span>
+<span style='background-color:#4A789C'>}</span>
+</pre></span></p>
+</li>
+<li><p class="first">Expansion regions are used to represent Clang’s macro expansions. They
+have an additional property - <em>expanded file id</em>. This property can be
+used by the code coverage tool to find the mapping regions that are created
+as a result of this macro expansion, by checking if their file id matches the
+expanded file id. They don’t associate with <a class="reference internal" href="#coverage-mapping-counters">coverage mapping counters</a>,
+as the code coverage tool can determine the execution count for this region
+by looking up the execution count of the first region with a corresponding
+file id.
+For example:</p>
+<p><span class="raw-html"><pre class='highlight' style='line-height:initial;'><span>int func(int x) </span><span style='background-color:#4A789C'>{ </span>
+<span style='background-color:#4A789C'> #define MAX(x,y) </span><span style='background-color:#85C1F5'>((x) > (y)? </span><span style='background-color:#F6D55D'>(x)</span><span style='background-color:#85C1F5'> : </span><span style='background-color:#F4BA70'>(y)</span><span style='background-color:#85C1F5'>)</span><span style='background-color:#4A789C'> </span>
+<span style='background-color:#4A789C'> return </span><span style='background-color:#7FCA9F'>MAX</span><span style='background-color:#4A789C'>(x, 42); </span> <span class='c1'>// Expansion Region from 3:10 to 3:13</span>
+<span style='background-color:#4A789C'>}</span>
+</pre></span></p>
+</li>
+</ul>
+<div class="section" id="source-range">
+<span id="source-code-range"></span><h4><a class="toc-backref" href="#id16">Source Range:</a><a class="headerlink" href="#source-range" title="Permalink to this headline">¶</a></h4>
+<p>The source range record contains the starting and ending location of a certain
+mapping region. Both locations include the line and the column numbers.</p>
+</div>
+<div class="section" id="file-id">
+<span id="coverage-file-id"></span><h4><a class="toc-backref" href="#id17">File ID:</a><a class="headerlink" href="#file-id" title="Permalink to this headline">¶</a></h4>
+<p>The file id an integer value that tells us
+in which source file or macro expansion is this region located.
+It enables Clang to produce mapping information for the code
+defined inside macros, like this example demonstrates:</p>
+<p><span class="raw-html"><pre class='highlight' style='line-height:initial;'><span>void func(const char *str) </span><span style='background-color:#4A789C'>{ </span> <span class='c1'>// Code Region from 1:28 to 6:2 with file id 0</span>
+<span style='background-color:#4A789C'> #define PUT </span><span style='background-color:#85C1F5'>printf("%s\n", str)</span><span style='background-color:#4A789C'> </span> <span class='c1'>// 2 Code Regions from 2:15 to 2:34 with file ids 1 and 2</span>
+<span style='background-color:#4A789C'> if(*str) </span>
+<span style='background-color:#4A789C'> </span><span style='background-color:#F6D55D'>PUT</span><span style='background-color:#4A789C'>; </span> <span class='c1'>// Expansion Region from 4:5 to 4:8 with file id 0 that expands a macro with file id 1</span>
+<span style='background-color:#4A789C'> </span><span style='background-color:#F6D55D'>PUT</span><span style='background-color:#4A789C'>; </span> <span class='c1'>// Expansion Region from 5:3 to 5:6 with file id 0 that expands a macro with file id 2</span>
+<span style='background-color:#4A789C'>}</span>
+</pre></span></p>
+</div>
+<div class="section" id="counter">
+<span id="coverage-mapping-counters"></span><span id="coverage-mapping-counter"></span><h4><a class="toc-backref" href="#id18">Counter:</a><a class="headerlink" href="#counter" title="Permalink to this headline">¶</a></h4>
+<p>A coverage mapping counter can represents a reference to the profile
+instrumentation counter. The execution count for a region with such counter
+is determined by looking up the value of the corresponding profile
+instrumentation counter.</p>
+<p>It can also represent a binary arithmetical expression that operates on
+coverage mapping counters or other expressions.
+The execution count for a region with an expression counter is determined by
+evaluating the expression’s arguments and then adding them together or
+subtracting them from one another.
+In the example below, a subtraction expression is used to compute the execution
+count for the compound statement that follows the <em>else</em> keyword:</p>
+<p><span class="raw-html"><pre class='highlight' style='line-height:initial;'><span>int main(int argc, const char *argv[]) </span><span style='background-color:#4A789C'>{ </span> <span class='c1'>// Region's counter is a reference to the profile counter #0</span>
+<span style='background-color:#4A789C'> </span>
+<span style='background-color:#4A789C'> if (argc > 1) </span><span style='background-color:#85C1F5'>{ </span> <span class='c1'>// Region's counter is a reference to the profile counter #1</span>
+<span style='background-color:#85C1F5'> printf("%s\n", argv[1]); </span><span> </span>
+<span style='background-color:#85C1F5'> }</span><span style='background-color:#4A789C'> else </span><span style='background-color:#F6D55D'>{ </span> <span class='c1'>// Region's counter is an expression (reference to the profile counter #0 - reference to the profile counter #1)</span>
+<span style='background-color:#F6D55D'> printf("\n"); </span>
+<span style='background-color:#F6D55D'> }</span><span style='background-color:#4A789C'> </span>
+<span style='background-color:#4A789C'> return 0; </span>
+<span style='background-color:#4A789C'>}</span>
+</pre></span></p>
+<p>Finally, a coverage mapping counter can also represent an execution count of
+of zero. The zero counter is used to provide coverage mapping for
+unreachable statements and expressions, like in the example below:</p>
+<p><span class="raw-html"><pre class='highlight' style='line-height:initial;'><span>int main() </span><span style='background-color:#4A789C'>{ </span>
+<span style='background-color:#4A789C'> return 0; </span>
+<span style='background-color:#4A789C'> </span><span style='background-color:#85C1F5'>printf("Hello world!\n")</span><span style='background-color:#4A789C'>; </span> <span class='c1'>// Unreachable region's counter is zero</span>
+<span style='background-color:#4A789C'>}</span>
+</pre></span></p>
+<p>The zero counters allow the code coverage tool to display proper line execution
+counts for the unreachable lines and highlight the unreachable code.
+Without them, the tool would think that those lines and regions were still
+executed, as it doesn’t possess the frontend’s knowledge.</p>
+</div>
+</div>
+</div>
+<div class="section" id="llvm-ir-representation">
+<h2><a class="toc-backref" href="#id19">LLVM IR Representation</a><a class="headerlink" href="#llvm-ir-representation" title="Permalink to this headline">¶</a></h2>
+<p>The coverage mapping data is stored in the LLVM IR using a single global
+constant structure variable called <em>__llvm_coverage_mapping</em>
+with the <em>__llvm_covmap</em> section specifier.</p>
+<p>For example, letâs consider a C file and how it gets compiled to LLVM:</p>
+<div class="highlight-c" id="coverage-mapping-sample"><div class="highlight"><pre><span class="kt">int</span> <span class="nf">foo</span><span class="p">()</span> <span class="p">{</span>
+ <span class="k">return</span> <span class="mi">42</span><span class="p">;</span>
+<span class="p">}</span>
+<span class="kt">int</span> <span class="nf">bar</span><span class="p">()</span> <span class="p">{</span>
+ <span class="k">return</span> <span class="mi">13</span><span class="p">;</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>The coverage mapping variable generated by Clang has 3 fields:</p>
+<ul class="simple">
+<li>Coverage mapping header.</li>
+<li>An array of function records.</li>
+<li>Coverage mapping data which is an array of bytes. Zero paddings are added at the end to force 8 byte alignment.</li>
+</ul>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="vg">@__llvm_coverage_mapping</span> <span class="p">=</span> <span class="k">internal</span> <span class="k">constant</span> <span class="p">{</span> <span class="p">{</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span> <span class="p">},</span> <span class="p">[</span><span class="m">2</span> <span class="k">x</span> <span class="p">{</span> <span class="k">i64</span><span class="p">,</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i64</span> <span class="p">}],</span> <span class="p">[</span><span class="m">40</span> <span class="k">x</span> <span class="k">i8</span><span class="p">]</span> <span class="p">}</span>
+<span class="p">{</span>
+ <span class="p">{</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span> <span class="p">}</span> <span class="c">; Coverage map header</span>
+ <span class="p">{</span>
+ <span class="k">i32</span> <span class="m">2</span><span class="p">,</span> <span class="c">; The number of function records</span>
+ <span class="k">i32</span> <span class="m">20</span><span class="p">,</span> <span class="c">; The length of the string that contains the encoded translation unit filenames</span>
+ <span class="k">i32</span> <span class="m">20</span><span class="p">,</span> <span class="c">; The length of the string that contains the encoded coverage mapping data</span>
+ <span class="k">i32</span> <span class="m">1</span><span class="p">,</span> <span class="c">; Coverage mapping format version</span>
+ <span class="p">},</span>
+ <span class="p">[</span><span class="m">2</span> <span class="k">x</span> <span class="p">{</span> <span class="k">i64</span><span class="p">,</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i64</span> <span class="p">}]</span> <span class="p">[</span> <span class="c">; Function records</span>
+ <span class="p">{</span> <span class="k">i64</span><span class="p">,</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i64</span> <span class="p">}</span> <span class="p">{</span>
+ <span class="k">i64</span> <span class="m">0x5cf8c24cdb18bdac</span><span class="p">,</span> <span class="c">; Function's name MD5</span>
+ <span class="k">i32</span> <span class="m">9</span><span class="p">,</span> <span class="c">; Function's encoded coverage mapping data string length</span>
+ <span class="k">i64</span> <span class="m">0</span> <span class="c">; Function's structural hash</span>
+ <span class="p">},</span>
+ <span class="p">{</span> <span class="k">i64</span><span class="p">,</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i64</span> <span class="p">}</span> <span class="p">{</span>
+ <span class="k">i64</span> <span class="m">0xe413754a191db537</span><span class="p">,</span> <span class="c">; Function's name MD5</span>
+ <span class="k">i32</span> <span class="m">9</span><span class="p">,</span> <span class="c">; Function's encoded coverage mapping data string length</span>
+ <span class="k">i64</span> <span class="m">0</span> <span class="c">; Function's structural hash</span>
+ <span class="p">}],</span>
+ <span class="p">[</span><span class="m">40</span> <span class="k">x</span> <span class="k">i8</span><span class="p">]</span> <span class="k">c</span><span class="s">"..."</span> <span class="c">; Encoded data (dissected later)</span>
+<span class="p">},</span> <span class="k">section</span> <span class="s">"__llvm_covmap"</span><span class="p">,</span> <span class="k">align</span> <span class="m">8</span>
+</pre></div>
+</div>
+<p>The function record layout has evolved since version 1. In version 1, the function record for <em>foo</em> is defined as follows:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="p">{</span> <span class="k">i8</span><span class="p">*,</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i64</span> <span class="p">}</span> <span class="p">{</span> <span class="k">i8</span><span class="p">*</span> <span class="k">getelementptr</span> <span class="k">inbounds</span> <span class="p">([</span><span class="m">3</span> <span class="k">x</span> <span class="k">i8</span><span class="p">]*</span> <span class="vg">@__profn_foo</span><span class="p">,</span> <span class="k">i32</span> <span class="m">0</span><span class="p">,</span> <span class="k">i32</span> <span class="m">0</span><span class="p">),</span> <span class="c">; Function's name</span>
+ <span class="k">i32</span> <span class="m">3</span><span class="p">,</span> <span class="c">; Function's name length</span>
+ <span class="k">i32</span> <span class="m">9</span><span class="p">,</span> <span class="c">; Function's encoded coverage mapping data string length</span>
+ <span class="k">i64</span> <span class="m">0</span> <span class="c">; Function's structural hash</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<div class="section" id="coverage-mapping-header">
+<h3><a class="toc-backref" href="#id20">Coverage Mapping Header:</a><a class="headerlink" href="#coverage-mapping-header" title="Permalink to this headline">¶</a></h3>
+<p>The coverage mapping header has the following fields:</p>
+<ul class="simple">
+<li>The number of function records.</li>
+<li>The length of the string in the third field of <em>__llvm_coverage_mapping</em> that contains the encoded translation unit filenames.</li>
+<li>The length of the string in the third field of <em>__llvm_coverage_mapping</em> that contains the encoded coverage mapping data.</li>
+<li>The format version. The current version is 2 (encoded as a 1).</li>
+</ul>
+</div>
+<div class="section" id="function-record">
+<span id="function-records"></span><h3><a class="toc-backref" href="#id21">Function record:</a><a class="headerlink" href="#function-record" title="Permalink to this headline">¶</a></h3>
+<p>A function record is a structure of the following type:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="p">{</span> <span class="k">i64</span><span class="p">,</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i64</span> <span class="p">}</span>
+</pre></div>
+</div>
+<p>It contains function name’s MD5, the length of the encoded mapping data for that function, and function’s
+structural hash value.</p>
+</div>
+<div class="section" id="encoded-data">
+<h3><a class="toc-backref" href="#id22">Encoded data:</a><a class="headerlink" href="#encoded-data" title="Permalink to this headline">¶</a></h3>
+<p>The encoded data is stored in a single string that contains
+the encoded filenames used by this translation unit and the encoded coverage
+mapping data for each function in this translation unit.</p>
+<p>The encoded data has the following structure:</p>
+<p><tt class="docutils literal"><span class="pre">[filenames,</span> <span class="pre">coverageMappingDataForFunctionRecord0,</span> <span class="pre">coverageMappingDataForFunctionRecord1,</span> <span class="pre">...,</span> <span class="pre">padding]</span></tt></p>
+<p>If necessary, the encoded data is padded with zeroes so that the size
+of the data string is rounded up to the nearest multiple of 8 bytes.</p>
+<div class="section" id="dissecting-the-sample">
+<h4><a class="toc-backref" href="#id23">Dissecting the sample:</a><a class="headerlink" href="#dissecting-the-sample" title="Permalink to this headline">¶</a></h4>
+<p>Here’s an overview of the encoded data that was stored in the
+IR for the <a class="reference internal" href="#coverage-mapping-sample">coverage mapping sample</a> that was shown earlier:</p>
+<ul>
+<li><p class="first">The IR contains the following string constant that represents the encoded
+coverage mapping data for the sample translation unit:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">c</span><span class="s">"\01\12/Users/alex/test.c\01\00\00\01\01\01\0C\02\02\01\00\00\01\01\04\0C\02\02\00\00"</span>
+</pre></div>
+</div>
+</li>
+<li><p class="first">The string contains values that are encoded in the LEB128 format, which is
+used throughout for storing integers. It also contains a string value.</p>
+</li>
+<li><p class="first">The length of the substring that contains the encoded translation unit
+filenames is the value of the second field in the <em>__llvm_coverage_mapping</em>
+structure, which is 20, thus the filenames are encoded in this string:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">c</span><span class="s">"\01\12/Users/alex/test.c"</span>
+</pre></div>
+</div>
+<p>This string contains the following data:</p>
+<ul class="simple">
+<li>Its first byte has a value of <tt class="docutils literal"><span class="pre">0x01</span></tt>. It stores the number of filenames
+contained in this string.</li>
+<li>Its second byte stores the length of the first filename in this string.</li>
+<li>The remaining 18 bytes are used to store the first filename.</li>
+</ul>
+</li>
+<li><p class="first">The length of the substring that contains the encoded coverage mapping data
+for the first function is the value of the third field in the first
+structure in an array of <a class="reference internal" href="#function-records">function records</a> stored in the
+third field of the <em>__llvm_coverage_mapping</em> structure, which is the 9.
+Therefore, the coverage mapping for the first function record is encoded
+in this string:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">c</span><span class="s">"\01\00\00\01\01\01\0C\02\02"</span>
+</pre></div>
+</div>
+<p>This string consists of the following bytes:</p>
+<table border="1" class="docutils">
+<colgroup>
+<col width="8%" />
+<col width="92%" />
+</colgroup>
+<tbody valign="top">
+<tr class="row-odd"><td><tt class="docutils literal"><span class="pre">0x01</span></tt></td>
+<td>The number of file ids used by this function. There is only one file id used by the mapping data in this function.</td>
+</tr>
+<tr class="row-even"><td><tt class="docutils literal"><span class="pre">0x00</span></tt></td>
+<td>An index into the filenames array which corresponds to the file “/Users/alex/test.c”.</td>
+</tr>
+<tr class="row-odd"><td><tt class="docutils literal"><span class="pre">0x00</span></tt></td>
+<td>The number of counter expressions used by this function. This function doesn’t use any expressions.</td>
+</tr>
+<tr class="row-even"><td><tt class="docutils literal"><span class="pre">0x01</span></tt></td>
+<td>The number of mapping regions that are stored in an array for the function’s file id #0.</td>
+</tr>
+<tr class="row-odd"><td><tt class="docutils literal"><span class="pre">0x01</span></tt></td>
+<td>The coverage mapping counter for the first region in this function. The value of 1 tells us that it’s a coverage
+mapping counter that is a reference to the profile instrumentation counter with an index of 0.</td>
+</tr>
+<tr class="row-even"><td><tt class="docutils literal"><span class="pre">0x01</span></tt></td>
+<td>The starting line of the first mapping region in this function.</td>
+</tr>
+<tr class="row-odd"><td><tt class="docutils literal"><span class="pre">0x0C</span></tt></td>
+<td>The starting column of the first mapping region in this function.</td>
+</tr>
+<tr class="row-even"><td><tt class="docutils literal"><span class="pre">0x02</span></tt></td>
+<td>The ending line of the first mapping region in this function.</td>
+</tr>
+<tr class="row-odd"><td><tt class="docutils literal"><span class="pre">0x02</span></tt></td>
+<td>The ending column of the first mapping region in this function.</td>
+</tr>
+</tbody>
+</table>
+</li>
+<li><p class="first">The length of the substring that contains the encoded coverage mapping data
+for the second function record is also 9. It’s structured like the mapping data
+for the first function record.</p>
+</li>
+<li><p class="first">The two trailing bytes are zeroes and are used to pad the coverage mapping
+data to give it the 8 byte alignment.</p>
+</li>
+</ul>
+</div>
+</div>
+</div>
+<div class="section" id="encoding">
+<h2><a class="toc-backref" href="#id24">Encoding</a><a class="headerlink" href="#encoding" title="Permalink to this headline">¶</a></h2>
+<p>The per-function coverage mapping data is encoded as a stream of bytes,
+with a simple structure. The structure consists of the encoding
+<a class="reference internal" href="#cvmtypes">types</a> like variable-length unsigned integers, that
+are used to encode <a class="reference internal" href="#file-id-mapping">File ID Mapping</a>, <a class="reference internal" href="#counter-expressions">Counter Expressions</a> and
+the <a class="reference internal" href="#mapping-regions">Mapping Regions</a>.</p>
+<p>The format of the structure follows:</p>
+<blockquote>
+<div><tt class="docutils literal"><span class="pre">[file</span> <span class="pre">id</span> <span class="pre">mapping,</span> <span class="pre">counter</span> <span class="pre">expressions,</span> <span class="pre">mapping</span> <span class="pre">regions]</span></tt></div></blockquote>
+<p>The translation unit filenames are encoded using the same encoding
+<a class="reference internal" href="#cvmtypes">types</a> as the per-function coverage mapping data, with the
+following structure:</p>
+<blockquote>
+<div><tt class="docutils literal"><span class="pre">[numFilenames</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">filename0</span> <span class="pre">:</span> <span class="pre">string,</span> <span class="pre">filename1</span> <span class="pre">:</span> <span class="pre">string,</span> <span class="pre">...]</span></tt></div></blockquote>
+<div class="section" id="types">
+<span id="cvmtypes"></span><h3><a class="toc-backref" href="#id25">Types</a><a class="headerlink" href="#types" title="Permalink to this headline">¶</a></h3>
+<p>This section describes the basic types that are used by the encoding format
+and can appear after <tt class="docutils literal"><span class="pre">:</span></tt> in the <tt class="docutils literal"><span class="pre">[foo</span> <span class="pre">:</span> <span class="pre">type]</span></tt> description.</p>
+<div class="section" id="leb128">
+<span id="id1"></span><h4><a class="toc-backref" href="#id26">LEB128</a><a class="headerlink" href="#leb128" title="Permalink to this headline">¶</a></h4>
+<p>LEB128 is an unsigned integer value that is encoded using DWARF’s LEB128
+encoding, optimizing for the case where values are small
+(1 byte for values less than 128).</p>
+</div>
+<div class="section" id="strings">
+<span id="coveragestrings"></span><h4><a class="toc-backref" href="#id27">Strings</a><a class="headerlink" href="#strings" title="Permalink to this headline">¶</a></h4>
+<p><tt class="docutils literal"><span class="pre">[length</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">characters...]</span></tt></p>
+<p>String values are encoded with a <a class="reference internal" href="#leb128">LEB value</a> for the length
+of the string and a sequence of bytes for its characters.</p>
+</div>
+</div>
+<div class="section" id="file-id-mapping">
+<span id="id2"></span><h3><a class="toc-backref" href="#id28">File ID Mapping</a><a class="headerlink" href="#file-id-mapping" title="Permalink to this headline">¶</a></h3>
+<p><tt class="docutils literal"><span class="pre">[numIndices</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">filenameIndex0</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">filenameIndex1</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">...]</span></tt></p>
+<p>File id mapping in a function’s coverage mapping stream
+contains the indices into the translation unit’s filenames array.</p>
+</div>
+<div class="section" id="id3">
+<h3><a class="toc-backref" href="#id29">Counter</a><a class="headerlink" href="#id3" title="Permalink to this headline">¶</a></h3>
+<p><tt class="docutils literal"><span class="pre">[value</span> <span class="pre">:</span> <span class="pre">LEB128]</span></tt></p>
+<p>A <a class="reference internal" href="#coverage-mapping-counter">coverage mapping counter</a> is stored in a single <a class="reference internal" href="#leb128">LEB value</a>.
+It is composed of two things — the <a class="reference internal" href="#counter-tag">tag</a>
+which is stored in the lowest 2 bits, and the <a class="reference internal" href="#counter-data">counter data</a> which is stored
+in the remaining bits.</p>
+<div class="section" id="tag">
+<span id="counter-tag"></span><h4><a class="toc-backref" href="#id30">Tag:</a><a class="headerlink" href="#tag" title="Permalink to this headline">¶</a></h4>
+<p>The counter’s tag encodes the counter’s kind
+and, if the counter is an expression, the expression’s kind.
+The possible tag values are:</p>
+<ul class="simple">
+<li>0 - The counter is zero.</li>
+<li>1 - The counter is a reference to the profile instrumentation counter.</li>
+<li>2 - The counter is a subtraction expression.</li>
+<li>3 - The counter is an addition expression.</li>
+</ul>
+</div>
+<div class="section" id="data">
+<span id="counter-data"></span><h4><a class="toc-backref" href="#id31">Data:</a><a class="headerlink" href="#data" title="Permalink to this headline">¶</a></h4>
+<p>The counter’s data is interpreted in the following manner:</p>
+<ul class="simple">
+<li>When the counter is a reference to the profile instrumentation counter,
+then the counter’s data is the id of the profile counter.</li>
+<li>When the counter is an expression, then the counter’s data
+is the index into the array of counter expressions.</li>
+</ul>
+</div>
+</div>
+<div class="section" id="counter-expressions">
+<span id="id4"></span><h3><a class="toc-backref" href="#id32">Counter Expressions</a><a class="headerlink" href="#counter-expressions" title="Permalink to this headline">¶</a></h3>
+<p><tt class="docutils literal"><span class="pre">[numExpressions</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">expr0LHS</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">expr0RHS</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">expr1LHS</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">expr1RHS</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">...]</span></tt></p>
+<p>Counter expressions consist of two counters as they
+represent binary arithmetic operations.
+The expression’s kind is determined from the <a class="reference internal" href="#counter-tag">tag</a> of the
+counter that references this expression.</p>
+</div>
+<div class="section" id="mapping-regions">
+<span id="id5"></span><h3><a class="toc-backref" href="#id33">Mapping Regions</a><a class="headerlink" href="#mapping-regions" title="Permalink to this headline">¶</a></h3>
+<p><tt class="docutils literal"><span class="pre">[numRegionArrays</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">regionsForFile0,</span> <span class="pre">regionsForFile1,</span> <span class="pre">...]</span></tt></p>
+<p>The mapping regions are stored in an array of sub-arrays where every
+region in a particular sub-array has the same file id.</p>
+<p>The file id for a sub-array of regions is the index of that
+sub-array in the main array e.g. The first sub-array will have the file id
+of 0.</p>
+<div class="section" id="sub-array-of-regions">
+<h4><a class="toc-backref" href="#id34">Sub-Array of Regions</a><a class="headerlink" href="#sub-array-of-regions" title="Permalink to this headline">¶</a></h4>
+<p><tt class="docutils literal"><span class="pre">[numRegions</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">region0,</span> <span class="pre">region1,</span> <span class="pre">...]</span></tt></p>
+<p>The mapping regions for a specific file id are stored in an array that is
+sorted in an ascending order by the region’s starting location.</p>
+</div>
+<div class="section" id="id6">
+<h4><a class="toc-backref" href="#id35">Mapping Region</a><a class="headerlink" href="#id6" title="Permalink to this headline">¶</a></h4>
+<p><tt class="docutils literal"><span class="pre">[header,</span> <span class="pre">source</span> <span class="pre">range]</span></tt></p>
+<p>The mapping region record contains two sub-records —
+the <a class="reference internal" href="#header">header</a>, which stores the counter and/or the region’s kind,
+and the <a class="reference internal" href="#id9">source range</a> that contains the starting and ending
+location of this region.</p>
+</div>
+<div class="section" id="header">
+<span id="id7"></span><h4><a class="toc-backref" href="#id36">Header</a><a class="headerlink" href="#header" title="Permalink to this headline">¶</a></h4>
+<p><tt class="docutils literal"><span class="pre">[counter]</span></tt></p>
+<p>or</p>
+<p><tt class="docutils literal"><span class="pre">[pseudo-counter]</span></tt></p>
+<p>The header encodes the region’s counter and the region’s kind.</p>
+<p>The value of the counter’s tag distinguishes between the counters and
+pseudo-counters — if the tag is zero, than this header contains a
+pseudo-counter, otherwise this header contains an ordinary counter.</p>
+<div class="section" id="id8">
+<h5><a class="toc-backref" href="#id37">Counter:</a><a class="headerlink" href="#id8" title="Permalink to this headline">¶</a></h5>
+<p>A mapping region whose header has a counter with a non-zero tag is
+a code region.</p>
+</div>
+<div class="section" id="pseudo-counter">
+<h5><a class="toc-backref" href="#id38">Pseudo-Counter:</a><a class="headerlink" href="#pseudo-counter" title="Permalink to this headline">¶</a></h5>
+<p><tt class="docutils literal"><span class="pre">[value</span> <span class="pre">:</span> <span class="pre">LEB128]</span></tt></p>
+<p>A pseudo-counter is stored in a single <a class="reference internal" href="#leb128">LEB value</a>, just like
+the ordinary counter. It has the following interpretation:</p>
+<ul>
+<li><p class="first">bits 0-1: tag, which is always 0.</p>
+</li>
+<li><p class="first">bit 2: expansionRegionTag. If this bit is set, then this mapping region
+is an expansion region.</p>
+</li>
+<li><p class="first">remaining bits: data. If this region is an expansion region, then the data
+contains the expanded file id of that region.</p>
+<p>Otherwise, the data contains the region’s kind. The possible region
+kind values are:</p>
+<ul class="simple">
+<li>0 - This mapping region is a code region with a counter of zero.</li>
+<li>2 - This mapping region is a skipped region.</li>
+</ul>
+</li>
+</ul>
+</div>
+</div>
+<div class="section" id="id9">
+<span id="id10"></span><h4><a class="toc-backref" href="#id39">Source Range</a><a class="headerlink" href="#id9" title="Permalink to this headline">¶</a></h4>
+<p><tt class="docutils literal"><span class="pre">[deltaLineStart</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">columnStart</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">numLines</span> <span class="pre">:</span> <span class="pre">LEB128,</span> <span class="pre">columnEnd</span> <span class="pre">:</span> <span class="pre">LEB128]</span></tt></p>
+<p>The source range record contains the following fields:</p>
+<ul>
+<li><p class="first"><em>deltaLineStart</em>: The difference between the starting line of the
+current mapping region and the starting line of the previous mapping region.</p>
+<p>If the current mapping region is the first region in the current
+sub-array, then it stores the starting line of that region.</p>
+</li>
+<li><p class="first"><em>columnStart</em>: The starting column of the mapping region.</p>
+</li>
+<li><p class="first"><em>numLines</em>: The difference between the ending line and the starting line
+of the current mapping region.</p>
+</li>
+<li><p class="first"><em>columnEnd</em>: The ending column of the mapping region.</p>
+</li>
+</ul>
+</div>
+</div>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="Statepoints.html" title="Garbage Collection Safepoints in LLVM"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="BigEndianNEON.html" title="Using ARM NEON instructions in big endian mode"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/DebuggingJITedCode.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/DebuggingJITedCode.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/DebuggingJITedCode.html (added)
+++ www-releases/trunk/4.0.0/docs/DebuggingJITedCode.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,221 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>Debugging JIT-ed Code With GDB — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="The LLVM gold plugin" href="GoldPlugin.html" />
+ <link rel="prev" title="TableGen Deficiencies" href="TableGen/Deficiencies.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="GoldPlugin.html" title="The LLVM gold plugin"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="TableGen/Deficiencies.html" title="TableGen Deficiencies"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="debugging-jit-ed-code-with-gdb">
+<h1>Debugging JIT-ed Code With GDB<a class="headerlink" href="#debugging-jit-ed-code-with-gdb" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="background">
+<h2>Background<a class="headerlink" href="#background" title="Permalink to this headline">¶</a></h2>
+<p>Without special runtime support, debugging dynamically generated code with
+GDB (as well as most debuggers) can be quite painful. Debuggers generally
+read debug information from the object file of the code, but for JITed
+code, there is no such file to look for.</p>
+<p>In order to communicate the necessary debug info to GDB, an interface for
+registering JITed code with debuggers has been designed and implemented for
+GDB and LLVM MCJIT. At a high level, whenever MCJIT generates new machine code,
+it does so in an in-memory object file that contains the debug information in
+DWARF format. MCJIT then adds this in-memory object file to a global list of
+dynamically generated object files and calls a special function
+(<tt class="docutils literal"><span class="pre">__jit_debug_register_code</span></tt>) marked noinline that GDB knows about. When
+GDB attaches to a process, it puts a breakpoint in this function and loads all
+of the object files in the global list. When MCJIT calls the registration
+function, GDB catches the breakpoint signal, loads the new object file from
+the inferior’s memory, and resumes the execution. In this way, GDB can get the
+necessary debug information.</p>
+</div>
+<div class="section" id="gdb-version">
+<h2>GDB Version<a class="headerlink" href="#gdb-version" title="Permalink to this headline">¶</a></h2>
+<p>In order to debug code JIT-ed by LLVM, you need GDB 7.0 or newer, which is
+available on most modern distributions of Linux. The version of GDB that
+Apple ships with Xcode has been frozen at 6.3 for a while. LLDB may be a
+better option for debugging JIT-ed code on Mac OS X.</p>
+</div>
+<div class="section" id="debugging-mcjit-ed-code">
+<h2>Debugging MCJIT-ed code<a class="headerlink" href="#debugging-mcjit-ed-code" title="Permalink to this headline">¶</a></h2>
+<p>The emerging MCJIT component of LLVM allows full debugging of JIT-ed code with
+GDB. This is due to MCJIT’s ability to use the MC emitter to provide full
+DWARF debugging information to GDB.</p>
+<p>Note that lli has to be passed the <tt class="docutils literal"><span class="pre">-use-mcjit</span></tt> flag to JIT the code with
+MCJIT instead of the old JIT.</p>
+<div class="section" id="example">
+<h3>Example<a class="headerlink" href="#example" title="Permalink to this headline">¶</a></h3>
+<p>Consider the following C code (with line numbers added to make the example
+easier to follow):</p>
+<div class="highlight-c"><div class="highlight"><pre><span class="mi">1</span> <span class="kt">int</span> <span class="n">compute_factorial</span><span class="p">(</span><span class="kt">int</span> <span class="n">n</span><span class="p">)</span>
+<span class="mi">2</span> <span class="p">{</span>
+<span class="mi">3</span> <span class="k">if</span> <span class="p">(</span><span class="n">n</span> <span class="o"><=</span> <span class="mi">1</span><span class="p">)</span>
+<span class="mi">4</span> <span class="k">return</span> <span class="mi">1</span><span class="p">;</span>
+<span class="mi">5</span>
+<span class="mi">6</span> <span class="kt">int</span> <span class="n">f</span> <span class="o">=</span> <span class="n">n</span><span class="p">;</span>
+<span class="mi">7</span> <span class="k">while</span> <span class="p">(</span><span class="o">--</span><span class="n">n</span> <span class="o">></span> <span class="mi">1</span><span class="p">)</span>
+<span class="mi">8</span> <span class="n">f</span> <span class="o">*=</span> <span class="n">n</span><span class="p">;</span>
+<span class="mi">9</span> <span class="k">return</span> <span class="n">f</span><span class="p">;</span>
+<span class="mi">10</span> <span class="p">}</span>
+<span class="mi">11</span>
+<span class="mi">12</span>
+<span class="mi">13</span> <span class="kt">int</span> <span class="n">main</span><span class="p">(</span><span class="kt">int</span> <span class="n">argc</span><span class="p">,</span> <span class="kt">char</span><span class="o">**</span> <span class="n">argv</span><span class="p">)</span>
+<span class="mi">14</span> <span class="p">{</span>
+<span class="mi">15</span> <span class="k">if</span> <span class="p">(</span><span class="n">argc</span> <span class="o"><</span> <span class="mi">2</span><span class="p">)</span>
+<span class="mi">16</span> <span class="k">return</span> <span class="o">-</span><span class="mi">1</span><span class="p">;</span>
+<span class="mi">17</span> <span class="kt">char</span> <span class="n">firstletter</span> <span class="o">=</span> <span class="n">argv</span><span class="p">[</span><span class="mi">1</span><span class="p">][</span><span class="mi">0</span><span class="p">];</span>
+<span class="mi">18</span> <span class="kt">int</span> <span class="n">result</span> <span class="o">=</span> <span class="n">compute_factorial</span><span class="p">(</span><span class="n">firstletter</span> <span class="o">-</span> <span class="sc">'0'</span><span class="p">);</span>
+<span class="mi">19</span>
+<span class="mi">20</span> <span class="c1">// Returned result is clipped at 255...</span>
+<span class="mi">21</span> <span class="k">return</span> <span class="n">result</span><span class="p">;</span>
+<span class="mi">22</span> <span class="p">}</span>
+</pre></div>
+</div>
+<p>Here is a sample command line session that shows how to build and run this
+code via <tt class="docutils literal"><span class="pre">lli</span></tt> inside GDB:</p>
+<div class="highlight-bash"><div class="highlight"><pre><span class="nv">$ $BINPATH</span>/clang -cc1 -O0 -g -emit-llvm showdebug.c
+<span class="nv">$ </span>gdb --quiet --args <span class="nv">$BINPATH</span>/lli -use-mcjit showdebug.ll 5
+Reading symbols from <span class="nv">$BINPATH</span>/lli...done.
+<span class="o">(</span>gdb<span class="o">)</span> b showdebug.c:6
+No <span class="nb">source </span>file named showdebug.c.
+Make breakpoint pending on future shared library load? <span class="o">(</span>y or <span class="o">[</span>n<span class="o">])</span> y
+Breakpoint 1 <span class="o">(</span>showdebug.c:6<span class="o">)</span> pending.
+<span class="o">(</span>gdb<span class="o">)</span> r
+Starting program: <span class="nv">$BINPATH</span>/lli -use-mcjit showdebug.ll 5
+<span class="o">[</span>Thread debugging using libthread_db enabled<span class="o">]</span>
+
+Breakpoint 1, compute_factorial <span class="o">(</span><span class="nv">n</span><span class="o">=</span>5<span class="o">)</span> at showdebug.c:6
+6 int <span class="nv">f</span> <span class="o">=</span> n;
+<span class="o">(</span>gdb<span class="o">)</span> p n
+<span class="nv">$1</span> <span class="o">=</span> 5
+<span class="o">(</span>gdb<span class="o">)</span> p f
+<span class="nv">$2</span> <span class="o">=</span> 0
+<span class="o">(</span>gdb<span class="o">)</span> n
+7 <span class="k">while</span> <span class="o">(</span>--n > 1<span class="o">)</span>
+<span class="o">(</span>gdb<span class="o">)</span> p f
+<span class="nv">$3</span> <span class="o">=</span> 5
+<span class="o">(</span>gdb<span class="o">)</span> b showdebug.c:9
+Breakpoint 2 at 0x7ffff7ed404c: file showdebug.c, line 9.
+<span class="o">(</span>gdb<span class="o">)</span> c
+Continuing.
+
+Breakpoint 2, compute_factorial <span class="o">(</span><span class="nv">n</span><span class="o">=</span>1<span class="o">)</span> at showdebug.c:9
+9 <span class="k">return </span>f;
+<span class="o">(</span>gdb<span class="o">)</span> p f
+<span class="nv">$4</span> <span class="o">=</span> 120
+<span class="o">(</span>gdb<span class="o">)</span> bt
+<span class="c">#0 compute_factorial (n=1) at showdebug.c:9</span>
+<span class="c">#1 0x00007ffff7ed40a9 in main (argc=2, argv=0x16677e0) at showdebug.c:18</span>
+<span class="c">#2 0x3500000001652748 in ?? ()</span>
+<span class="c">#3 0x00000000016677e0 in ?? ()</span>
+<span class="c">#4 0x0000000000000002 in ?? ()</span>
+<span class="c">#5 0x0000000000d953b3 in llvm::MCJIT::runFunction (this=0x16151f0, F=0x1603020, ArgValues=...) at /home/ebenders_test/llvm_svn_rw/lib/ExecutionEngine/MCJIT/MCJIT.cpp:161</span>
+<span class="c">#6 0x0000000000dc8872 in llvm::ExecutionEngine::runFunctionAsMain (this=0x16151f0, Fn=0x1603020, argv=..., envp=0x7fffffffe040)</span>
+ at /home/ebenders_test/llvm_svn_rw/lib/ExecutionEngine/ExecutionEngine.cpp:397
+<span class="c">#7 0x000000000059c583 in main (argc=4, argv=0x7fffffffe018, envp=0x7fffffffe040) at /home/ebenders_test/llvm_svn_rw/tools/lli/lli.cpp:324</span>
+<span class="o">(</span>gdb<span class="o">)</span> finish
+Run till <span class="nb">exit </span>from <span class="c">#0 compute_factorial (n=1) at showdebug.c:9</span>
+0x00007ffff7ed40a9 in main <span class="o">(</span><span class="nv">argc</span><span class="o">=</span>2, <span class="nv">argv</span><span class="o">=</span>0x16677e0<span class="o">)</span> at showdebug.c:18
+18 int <span class="nv">result</span> <span class="o">=</span> compute_factorial<span class="o">(</span>firstletter - <span class="s1">'0'</span><span class="o">)</span>;
+Value returned is <span class="nv">$5</span> <span class="o">=</span> 120
+<span class="o">(</span>gdb<span class="o">)</span> p result
+<span class="nv">$6</span> <span class="o">=</span> 23406408
+<span class="o">(</span>gdb<span class="o">)</span> n
+21 <span class="k">return </span>result;
+<span class="o">(</span>gdb<span class="o">)</span> p result
+<span class="nv">$7</span> <span class="o">=</span> 120
+<span class="o">(</span>gdb<span class="o">)</span> c
+Continuing.
+
+Program exited with code 0170.
+<span class="o">(</span>gdb<span class="o">)</span>
+</pre></div>
+</div>
+</div>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="GoldPlugin.html" title="The LLVM gold plugin"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="TableGen/Deficiencies.html" title="TableGen Deficiencies"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/DeveloperPolicy.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/DeveloperPolicy.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/DeveloperPolicy.html (added)
+++ www-releases/trunk/4.0.0/docs/DeveloperPolicy.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,729 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>LLVM Developer Policy — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="Creating an LLVM Project" href="Projects.html" />
+ <link rel="prev" title="CodeView Type Records" href="PDB/CodeViewTypes.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="Projects.html" title="Creating an LLVM Project"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="PDB/CodeViewTypes.html" title="CodeView Type Records"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-developer-policy">
+<h1>LLVM Developer Policy<a class="headerlink" href="#llvm-developer-policy" title="Permalink to this headline">¶</a></h1>
+<div class="contents local topic" id="contents">
+<ul class="simple">
+<li><a class="reference internal" href="#introduction" id="id9">Introduction</a></li>
+<li><a class="reference internal" href="#developer-policies" id="id10">Developer Policies</a><ul>
+<li><a class="reference internal" href="#stay-informed" id="id11">Stay Informed</a></li>
+<li><a class="reference internal" href="#making-and-submitting-a-patch" id="id12">Making and Submitting a Patch</a></li>
+<li><a class="reference internal" href="#code-reviews" id="id13">Code Reviews</a></li>
+<li><a class="reference internal" href="#code-owners" id="id14">Code Owners</a></li>
+<li><a class="reference internal" href="#test-cases" id="id15">Test Cases</a></li>
+<li><a class="reference internal" href="#quality" id="id16">Quality</a></li>
+<li><a class="reference internal" href="#commit-messages" id="id17">Commit messages</a></li>
+<li><a class="reference internal" href="#obtaining-commit-access" id="id18">Obtaining Commit Access</a></li>
+<li><a class="reference internal" href="#making-a-major-change" id="id19">Making a Major Change</a></li>
+<li><a class="reference internal" href="#incremental-development" id="id20">Incremental Development</a></li>
+<li><a class="reference internal" href="#attribution-of-changes" id="id21">Attribution of Changes</a></li>
+<li><a class="reference internal" href="#ir-backwards-compatibility" id="id22">IR Backwards Compatibility</a></li>
+<li><a class="reference internal" href="#c-api-changes" id="id23">C API Changes</a></li>
+<li><a class="reference internal" href="#new-targets" id="id24">New Targets</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#copyright-license-and-patents" id="id25">Copyright, License, and Patents</a><ul>
+<li><a class="reference internal" href="#copyright" id="id26">Copyright</a></li>
+<li><a class="reference internal" href="#license" id="id27">License</a></li>
+<li><a class="reference internal" href="#patents" id="id28">Patents</a></li>
+</ul>
+</li>
+</ul>
+</div>
+<div class="section" id="introduction">
+<h2><a class="toc-backref" href="#id9">Introduction</a><a class="headerlink" href="#introduction" title="Permalink to this headline">¶</a></h2>
+<p>This document contains the LLVM Developer Policy which defines the project’s
+policy towards developers and their contributions. The intent of this policy is
+to eliminate miscommunication, rework, and confusion that might arise from the
+distributed nature of LLVM’s development. By stating the policy in clear terms,
+we hope each developer can know ahead of time what to expect when making LLVM
+contributions. This policy covers all llvm.org subprojects, including Clang,
+LLDB, libc++, etc.</p>
+<p>This policy is also designed to accomplish the following objectives:</p>
+<ol class="arabic simple">
+<li>Attract both users and developers to the LLVM project.</li>
+<li>Make life as simple and easy for contributors as possible.</li>
+<li>Keep the top of Subversion trees as stable as possible.</li>
+<li>Establish awareness of the project’s <a class="reference internal" href="#copyright-license-patents"><em>copyright, license, and patent
+policies</em></a> with contributors to the project.</li>
+</ol>
+<p>This policy is aimed at frequent contributors to LLVM. People interested in
+contributing one-off patches can do so in an informal way by sending them to the
+<a class="reference external" href="http://lists.llvm.org/mailman/listinfo/llvm-commits">llvm-commits mailing list</a> and engaging another
+developer to see it through the process.</p>
+</div>
+<div class="section" id="developer-policies">
+<h2><a class="toc-backref" href="#id10">Developer Policies</a><a class="headerlink" href="#developer-policies" title="Permalink to this headline">¶</a></h2>
+<p>This section contains policies that pertain to frequent LLVM developers. We
+always welcome <a class="reference internal" href="#one-off-patches">one-off patches</a> from people who do not routinely contribute to
+LLVM, but we expect more from frequent contributors to keep the system as
+efficient as possible for everyone. Frequent LLVM contributors are expected to
+meet the following requirements in order for LLVM to maintain a high standard of
+quality.</p>
+<div class="section" id="stay-informed">
+<h3><a class="toc-backref" href="#id11">Stay Informed</a><a class="headerlink" href="#stay-informed" title="Permalink to this headline">¶</a></h3>
+<p>Developers should stay informed by reading at least the “dev” mailing list for
+the projects you are interested in, such as <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/llvm-dev">llvm-dev</a> for LLVM, <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/cfe-dev">cfe-dev</a> for Clang, or <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/lldb-dev">lldb-dev</a> for LLDB. If you are
+doing anything more than just casual work on LLVM, it is suggested that you also
+subscribe to the “commits” mailing list for the subproject you’re interested in,
+such as <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/llvm-commits">llvm-commits</a>, <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/cfe-commits">cfe-commits</a>, or <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/lldb-commits">lldb-commits</a>. Reading the
+“commits” list and paying attention to changes being made by others is a good
+way to see what other people are interested in and watching the flow of the
+project as a whole.</p>
+<p>We recommend that active developers register an email account with <a class="reference external" href="http://llvm.org/bugs/">LLVM
+Bugzilla</a> and preferably subscribe to the <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/llvm-bugs">llvm-bugs</a> email list to keep track
+of bugs and enhancements occurring in LLVM. We really appreciate people who are
+proactive at catching incoming bugs in their components and dealing with them
+promptly.</p>
+<p>Please be aware that all public LLVM mailing lists are public and archived, and
+that notices of confidentiality or non-disclosure cannot be respected.</p>
+</div>
+<div class="section" id="making-and-submitting-a-patch">
+<span id="one-off-patches"></span><span id="patch"></span><h3><a class="toc-backref" href="#id12">Making and Submitting a Patch</a><a class="headerlink" href="#making-and-submitting-a-patch" title="Permalink to this headline">¶</a></h3>
+<p>When making a patch for review, the goal is to make it as easy for the reviewer
+to read it as possible. As such, we recommend that you:</p>
+<ol class="arabic simple">
+<li>Make your patch against the Subversion trunk, not a branch, and not an old
+version of LLVM. This makes it easy to apply the patch. For information on
+how to check out SVN trunk, please see the <a class="reference external" href="GettingStarted.html#checkout">Getting Started
+Guide</a>.</li>
+<li>Similarly, patches should be submitted soon after they are generated. Old
+patches may not apply correctly if the underlying code changes between the
+time the patch was created and the time it is applied.</li>
+<li>Patches should be made with <tt class="docutils literal"><span class="pre">svn</span> <span class="pre">diff</span></tt>, or similar. If you use a
+different tool, make sure it uses the <tt class="docutils literal"><span class="pre">diff</span> <span class="pre">-u</span></tt> format and that it
+doesn’t contain clutter which makes it hard to read.</li>
+<li>If you are modifying generated files, such as the top-level <tt class="docutils literal"><span class="pre">configure</span></tt>
+script, please separate out those changes into a separate patch from the rest
+of your changes.</li>
+</ol>
+<p>Once your patch is ready, submit it by emailing it to the appropriate project’s
+commit mailing list (or commit it directly if applicable). Alternatively, some
+patches get sent to the project’s development list or component of the LLVM bug
+tracker, but the commit list is the primary place for reviews and should
+generally be preferred.</p>
+<p>When sending a patch to a mailing list, it is a good idea to send it as an
+<em>attachment</em> to the message, not embedded into the text of the message. This
+ensures that your mailer will not mangle the patch when it sends it (e.g. by
+making whitespace changes or by wrapping lines).</p>
+<p><em>For Thunderbird users:</em> Before submitting a patch, please open <em>Preferences >
+Advanced > General > Config Editor</em>, find the key
+<tt class="docutils literal"><span class="pre">mail.content_disposition_type</span></tt>, and set its value to <tt class="docutils literal"><span class="pre">1</span></tt>. Without this
+setting, Thunderbird sends your attachment using <tt class="docutils literal"><span class="pre">Content-Disposition:</span> <span class="pre">inline</span></tt>
+rather than <tt class="docutils literal"><span class="pre">Content-Disposition:</span> <span class="pre">attachment</span></tt>. Apple Mail gamely displays such
+a file inline, making it difficult to work with for reviewers using that
+program.</p>
+<p>When submitting patches, please do not add confidentiality or non-disclosure
+notices to the patches themselves. These notices conflict with the <a class="reference internal" href="#llvm-license">LLVM
+License</a> and may result in your contribution being excluded.</p>
+</div>
+<div class="section" id="code-reviews">
+<span id="code-review"></span><h3><a class="toc-backref" href="#id13">Code Reviews</a><a class="headerlink" href="#code-reviews" title="Permalink to this headline">¶</a></h3>
+<p>LLVM has a code review policy. Code review is one way to increase the quality of
+software. We generally follow these policies:</p>
+<ol class="arabic simple">
+<li>All developers are required to have significant changes reviewed before they
+are committed to the repository.</li>
+<li>Code reviews are conducted by email on the relevant project’s commit mailing
+list, or alternatively on the project’s development list or bug tracker.</li>
+<li>Code can be reviewed either before it is committed or after. We expect major
+changes to be reviewed before being committed, but smaller changes (or
+changes where the developer owns the component) can be reviewed after commit.</li>
+<li>The developer responsible for a code change is also responsible for making
+all necessary review-related changes.</li>
+<li>Code review can be an iterative process, which continues until the patch is
+ready to be committed. Specifically, once a patch is sent out for review, it
+needs an explicit “looks good” before it is submitted. Do not assume silent
+approval, or request active objections to the patch with a deadline.</li>
+</ol>
+<p>Sometimes code reviews will take longer than you would hope for, especially for
+larger features. Accepted ways to speed up review times for your patches are:</p>
+<ul class="simple">
+<li>Review other people’s patches. If you help out, everybody will be more
+willing to do the same for you; goodwill is our currency.</li>
+<li>Ping the patch. If it is urgent, provide reasons why it is important to you to
+get this patch landed and ping it every couple of days. If it is
+not urgent, the common courtesy ping rate is one week. Remember that you’re
+asking for valuable time from other professional developers.</li>
+<li>Ask for help on IRC. Developers on IRC will be able to either help you
+directly, or tell you who might be a good reviewer.</li>
+<li>Split your patch into multiple smaller patches that build on each other. The
+smaller your patch, the higher the probability that somebody will take a quick
+look at it.</li>
+</ul>
+<p>Developers should participate in code reviews as both reviewers and
+reviewees. If someone is kind enough to review your code, you should return the
+favor for someone else. Note that anyone is welcome to review and give feedback
+on a patch, but only people with Subversion write access can approve it.</p>
+<p>There is a web based code review tool that can optionally be used
+for code reviews. See <a class="reference internal" href="Phabricator.html"><em>Code Reviews with Phabricator</em></a>.</p>
+</div>
+<div class="section" id="code-owners">
+<span id="id1"></span><h3><a class="toc-backref" href="#id14">Code Owners</a><a class="headerlink" href="#code-owners" title="Permalink to this headline">¶</a></h3>
+<p>The LLVM Project relies on two features of its process to maintain rapid
+development in addition to the high quality of its source base: the combination
+of code review plus post-commit review for trusted maintainers. Having both is
+a great way for the project to take advantage of the fact that most people do
+the right thing most of the time, and only commit patches without pre-commit
+review when they are confident they are right.</p>
+<p>The trick to this is that the project has to guarantee that all patches that are
+committed are reviewed after they go in: you don’t want everyone to assume
+someone else will review it, allowing the patch to go unreviewed. To solve this
+problem, we have a notion of an ‘owner’ for a piece of the code. The sole
+responsibility of a code owner is to ensure that a commit to their area of the
+code is appropriately reviewed, either by themself or by someone else. The list
+of current code owners can be found in the file
+<a class="reference external" href="http://llvm.org/klaus/llvm/blob/master/CODE_OWNERS.TXT">CODE_OWNERS.TXT</a>
+in the root of the LLVM source tree.</p>
+<p>Note that code ownership is completely different than reviewers: anyone can
+review a piece of code, and we welcome code review from anyone who is
+interested. Code owners are the “last line of defense” to guarantee that all
+patches that are committed are actually reviewed.</p>
+<p>Being a code owner is a somewhat unglamorous position, but it is incredibly
+important for the ongoing success of the project. Because people get busy,
+interests change, and unexpected things happen, code ownership is purely opt-in,
+and anyone can choose to resign their “title” at any time. For now, we do not
+have an official policy on how one gets elected to be a code owner.</p>
+</div>
+<div class="section" id="test-cases">
+<span id="include-a-testcase"></span><h3><a class="toc-backref" href="#id15">Test Cases</a><a class="headerlink" href="#test-cases" title="Permalink to this headline">¶</a></h3>
+<p>Developers are required to create test cases for any bugs fixed and any new
+features added. Some tips for getting your testcase approved:</p>
+<ul class="simple">
+<li>All feature and regression test cases are added to the <tt class="docutils literal"><span class="pre">llvm/test</span></tt>
+directory. The appropriate sub-directory should be selected (see the
+<a class="reference internal" href="TestingGuide.html"><em>Testing Guide</em></a> for details).</li>
+<li>Test cases should be written in <a class="reference internal" href="LangRef.html"><em>LLVM assembly language</em></a>.</li>
+<li>Test cases, especially for regressions, should be reduced as much as possible,
+by <a class="reference internal" href="Bugpoint.html"><em>bugpoint</em></a> or manually. It is unacceptable to place an
+entire failing program into <tt class="docutils literal"><span class="pre">llvm/test</span></tt> as this creates a <em>time-to-test</em>
+burden on all developers. Please keep them short.</li>
+</ul>
+<p>Note that llvm/test and clang/test are designed for regression and small feature
+tests only. More extensive test cases (e.g., entire applications, benchmarks,
+etc) should be added to the <tt class="docutils literal"><span class="pre">llvm-test</span></tt> test suite. The llvm-test suite is
+for coverage (correctness, performance, etc) testing, not feature or regression
+testing.</p>
+</div>
+<div class="section" id="quality">
+<h3><a class="toc-backref" href="#id16">Quality</a><a class="headerlink" href="#quality" title="Permalink to this headline">¶</a></h3>
+<p>The minimum quality standards that any change must satisfy before being
+committed to the main development branch are:</p>
+<ol class="arabic simple">
+<li>Code must adhere to the <a class="reference external" href="CodingStandards.html">LLVM Coding Standards</a>.</li>
+<li>Code must compile cleanly (no errors, no warnings) on at least one platform.</li>
+<li>Bug fixes and new features should <a class="reference internal" href="#include-a-testcase">include a testcase</a> so we know if the
+fix/feature ever regresses in the future.</li>
+<li>Code must pass the <tt class="docutils literal"><span class="pre">llvm/test</span></tt> test suite.</li>
+<li>The code must not cause regressions on a reasonable subset of llvm-test,
+where “reasonable” depends on the contributor’s judgement and the scope of
+the change (more invasive changes require more testing). A reasonable subset
+might be something like “<tt class="docutils literal"><span class="pre">llvm-test/MultiSource/Benchmarks</span></tt>”.</li>
+</ol>
+<p>Additionally, the committer is responsible for addressing any problems found in
+the future that the change is responsible for. For example:</p>
+<ul class="simple">
+<li>The code should compile cleanly on all supported platforms.</li>
+<li>The changes should not cause any correctness regressions in the <tt class="docutils literal"><span class="pre">llvm-test</span></tt>
+suite and must not cause any major performance regressions.</li>
+<li>The change set should not cause performance or correctness regressions for the
+LLVM tools.</li>
+<li>The changes should not cause performance or correctness regressions in code
+compiled by LLVM on all applicable targets.</li>
+<li>You are expected to address any <a class="reference external" href="http://llvm.org/bugs/">Bugzilla bugs</a> that
+result from your change.</li>
+</ul>
+<p>We prefer for this to be handled before submission but understand that it isn’t
+possible to test all of this for every submission. Our build bots and nightly
+testing infrastructure normally finds these problems. A good rule of thumb is
+to check the nightly testers for regressions the day after your change. Build
+bots will directly email you if a group of commits that included yours caused a
+failure. You are expected to check the build bot messages to see if they are
+your fault and, if so, fix the breakage.</p>
+<p>Commits that violate these quality standards (e.g. are very broken) may be
+reverted. This is necessary when the change blocks other developers from making
+progress. The developer is welcome to re-commit the change after the problem has
+been fixed.</p>
+</div>
+<div class="section" id="commit-messages">
+<span id="id2"></span><h3><a class="toc-backref" href="#id17">Commit messages</a><a class="headerlink" href="#commit-messages" title="Permalink to this headline">¶</a></h3>
+<p>Although we don’t enforce the format of commit messages, we prefer that
+you follow these guidelines to help review, search in logs, email formatting
+and so on. These guidelines are very similar to rules used by other open source
+projects.</p>
+<p>Most importantly, the contents of the message should be carefully written to
+convey the rationale of the change (without delving too much in detail). It
+also should avoid being vague or overly specific. For example, “bits were not
+set right” will leave the reviewer wondering about which bits, and why they
+weren’t right, while “Correctly set overflow bits in TargetInfo” conveys almost
+all there is to the change.</p>
+<p>Below are some guidelines about the format of the message itself:</p>
+<ul class="simple">
+<li>Separate the commit message into title, body and, if you’re not the original
+author, a “Patch by” attribution line (see below).</li>
+<li>The title should be concise. Because all commits are emailed to the list with
+the first line as the subject, long titles are frowned upon. Short titles
+also look better in <cite>git log</cite>.</li>
+<li>When the changes are restricted to a specific part of the code (e.g. a
+back-end or optimization pass), it is customary to add a tag to the
+beginning of the line in square brackets. For example, “[SCEV] ...”
+or “[OpenMP] ...”. This helps email filters and searches for post-commit
+reviews.</li>
+<li>The body, if it exists, should be separated from the title by an empty line.</li>
+<li>The body should be concise, but explanatory, including a complete
+reasoning. Unless it is required to understand the change, examples,
+code snippets and gory details should be left to bug comments, web
+review or the mailing list.</li>
+<li>If the patch fixes a bug in bugzilla, please include the PR# in the message.</li>
+<li><a class="reference internal" href="#attribution-of-changes">Attribution of Changes</a> should be in a separate line, after the end of
+the body, as simple as “Patch by John Doe.”. This is how we officially
+handle attribution, and there are automated processes that rely on this
+format.</li>
+<li>Text formatting and spelling should follow the same rules as documentation
+and in-code comments, ex. capitalization, full stop, etc.</li>
+<li>If the commit is a bug fix on top of another recently committed patch, or a
+revert or reapply of a patch, include the svn revision number of the prior
+related commit. This could be as simple as “Revert rNNNN because it caused
+PR#”.</li>
+</ul>
+<p>For minor violations of these recommendations, the community normally favors
+reminding the contributor of this policy over reverting. Minor corrections and
+omissions can be handled by sending a reply to the commits mailing list.</p>
+</div>
+<div class="section" id="obtaining-commit-access">
+<h3><a class="toc-backref" href="#id18">Obtaining Commit Access</a><a class="headerlink" href="#obtaining-commit-access" title="Permalink to this headline">¶</a></h3>
+<p>We grant commit access to contributors with a track record of submitting high
+quality patches. If you would like commit access, please send an email to
+<a class="reference external" href="mailto:clattner%40llvm.org">Chris</a> with the following information:</p>
+<ol class="arabic simple">
+<li>The user name you want to commit with, e.g. “hacker”.</li>
+<li>The full name and email address you want message to llvm-commits to come
+from, e.g. “J. Random Hacker <<a class="reference external" href="mailto:hacker%40yoyodyne.com">hacker<span>@</span>yoyodyne<span>.</span>com</a>>”.</li>
+<li>A “password hash” of the password you want to use, e.g. “<tt class="docutils literal"><span class="pre">2ACR96qjUqsyM</span></tt>”.
+Note that you don’t ever tell us what your password is; you just give it to
+us in an encrypted form. To get this, run “<tt class="docutils literal"><span class="pre">htpasswd</span></tt>” (a utility that
+comes with apache) in <em>crypt</em> mode (often enabled with “<tt class="docutils literal"><span class="pre">-d</span></tt>”), or find a web
+page that will do it for you. Note that our system does not work with MD5
+hashes. These are significantly longer than a crypt hash - e.g.
+“<tt class="docutils literal"><span class="pre">$apr1$vea6bBV2$Z8IFx.AfeD8LhqlZFqJer0</span></tt>”, we only accept the shorter crypt hash.</li>
+</ol>
+<p>Once you’ve been granted commit access, you should be able to check out an LLVM
+tree with an SVN URL of “<a class="reference external" href="https://username@llvm.org/">https://username@llvm.org/</a>...” instead of the normal
+anonymous URL of “<a class="reference external" href="http://llvm.org/">http://llvm.org/</a>...”. The first time you commit you’ll have
+to type in your password. Note that you may get a warning from SVN about an
+untrusted key; you can ignore this. To verify that your commit access works,
+please do a test commit (e.g. change a comment or add a blank line). Your first
+commit to a repository may require the autogenerated email to be approved by a
+mailing list. This is normal and will be done when the mailing list owner has
+time.</p>
+<p>If you have recently been granted commit access, these policies apply:</p>
+<ol class="arabic simple">
+<li>You are granted <em>commit-after-approval</em> to all parts of LLVM. To get
+approval, submit a <a class="reference internal" href="#patch">patch</a> to <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/llvm-commits">llvm-commits</a>. When approved,
+you may commit it yourself.</li>
+<li>You are allowed to commit patches without approval which you think are
+obvious. This is clearly a subjective decision — we simply expect you to
+use good judgement. Examples include: fixing build breakage, reverting
+obviously broken patches, documentation/comment changes, any other minor
+changes.</li>
+<li>You are allowed to commit patches without approval to those portions of LLVM
+that you have contributed or maintain (i.e., have been assigned
+responsibility for), with the proviso that such commits must not break the
+build. This is a “trust but verify” policy, and commits of this nature are
+reviewed after they are committed.</li>
+<li>Multiple violations of these policies or a single egregious violation may
+cause commit access to be revoked.</li>
+</ol>
+<p>In any case, your changes are still subject to <a class="reference internal" href="#code-review">code review</a> (either before or
+after they are committed, depending on the nature of the change). You are
+encouraged to review other peoples’ patches as well, but you aren’t required
+to do so.</p>
+</div>
+<div class="section" id="making-a-major-change">
+<span id="discuss-the-change-gather-consensus"></span><h3><a class="toc-backref" href="#id19">Making a Major Change</a><a class="headerlink" href="#making-a-major-change" title="Permalink to this headline">¶</a></h3>
+<p>When a developer begins a major new project with the aim of contributing it back
+to LLVM, they should inform the community with an email to the <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/llvm-dev">llvm-dev</a> email list, to the extent
+possible. The reason for this is to:</p>
+<ol class="arabic simple">
+<li>keep the community informed about future changes to LLVM,</li>
+<li>avoid duplication of effort by preventing multiple parties working on the
+same thing and not knowing about it, and</li>
+<li>ensure that any technical issues around the proposed work are discussed and
+resolved before any significant work is done.</li>
+</ol>
+<p>The design of LLVM is carefully controlled to ensure that all the pieces fit
+together well and are as consistent as possible. If you plan to make a major
+change to the way LLVM works or want to add a major new extension, it is a good
+idea to get consensus with the development community before you start working on
+it.</p>
+<p>Once the design of the new feature is finalized, the work itself should be done
+as a series of <a class="reference internal" href="#incremental-changes">incremental changes</a>, not as a long-term development branch.</p>
+</div>
+<div class="section" id="incremental-development">
+<span id="incremental-changes"></span><h3><a class="toc-backref" href="#id20">Incremental Development</a><a class="headerlink" href="#incremental-development" title="Permalink to this headline">¶</a></h3>
+<p>In the LLVM project, we do all significant changes as a series of incremental
+patches. We have a strong dislike for huge changes or long-term development
+branches. Long-term development branches have a number of drawbacks:</p>
+<ol class="arabic simple">
+<li>Branches must have mainline merged into them periodically. If the branch
+development and mainline development occur in the same pieces of code,
+resolving merge conflicts can take a lot of time.</li>
+<li>Other people in the community tend to ignore work on branches.</li>
+<li>Huge changes (produced when a branch is merged back onto mainline) are
+extremely difficult to <a class="reference internal" href="#code-review">code review</a>.</li>
+<li>Branches are not routinely tested by our nightly tester infrastructure.</li>
+<li>Changes developed as monolithic large changes often don’t work until the
+entire set of changes is done. Breaking it down into a set of smaller
+changes increases the odds that any of the work will be committed to the main
+repository.</li>
+</ol>
+<p>To address these problems, LLVM uses an incremental development style and we
+require contributors to follow this practice when making a large/invasive
+change. Some tips:</p>
+<ul class="simple">
+<li>Large/invasive changes usually have a number of secondary changes that are
+required before the big change can be made (e.g. API cleanup, etc). These
+sorts of changes can often be done before the major change is done,
+independently of that work.</li>
+<li>The remaining inter-related work should be decomposed into unrelated sets of
+changes if possible. Once this is done, define the first increment and get
+consensus on what the end goal of the change is.</li>
+<li>Each change in the set can be stand alone (e.g. to fix a bug), or part of a
+planned series of changes that works towards the development goal.</li>
+<li>Each change should be kept as small as possible. This simplifies your work
+(into a logical progression), simplifies code review and reduces the chance
+that you will get negative feedback on the change. Small increments also
+facilitate the maintenance of a high quality code base.</li>
+<li>Often, an independent precursor to a big change is to add a new API and slowly
+migrate clients to use the new API. Each change to use the new API is often
+“obvious” and can be committed without review. Once the new API is in place
+and used, it is much easier to replace the underlying implementation of the
+API. This implementation change is logically separate from the API
+change.</li>
+</ul>
+<p>If you are interested in making a large change, and this scares you, please make
+sure to first <a class="reference internal" href="#discuss-the-change-gather-consensus">discuss the change/gather consensus</a> then ask about the best way
+to go about making the change.</p>
+</div>
+<div class="section" id="attribution-of-changes">
+<h3><a class="toc-backref" href="#id21">Attribution of Changes</a><a class="headerlink" href="#attribution-of-changes" title="Permalink to this headline">¶</a></h3>
+<p>When contributors submit a patch to an LLVM project, other developers with
+commit access may commit it for the author once appropriate (based on the
+progression of code review, etc.). When doing so, it is important to retain
+correct attribution of contributions to their contributors. However, we do not
+want the source code to be littered with random attributions “this code written
+by J. Random Hacker” (this is noisy and distracting). In practice, the revision
+control system keeps a perfect history of who changed what, and the CREDITS.txt
+file describes higher-level contributions. If you commit a patch for someone
+else, please follow the attribution of changes in the simple manner as outlined
+by the <a class="reference internal" href="#commit-messages">commit messages</a> section. Overall, please do not add contributor names
+to the source code.</p>
+<p>Also, don’t commit patches authored by others unless they have submitted the
+patch to the project or you have been authorized to submit them on their behalf
+(you work together and your company authorized you to contribute the patches,
+etc.). The author should first submit them to the relevant project’s commit
+list, development list, or LLVM bug tracker component. If someone sends you
+a patch privately, encourage them to submit it to the appropriate list first.</p>
+</div>
+<div class="section" id="ir-backwards-compatibility">
+<span id="id5"></span><h3><a class="toc-backref" href="#id22">IR Backwards Compatibility</a><a class="headerlink" href="#ir-backwards-compatibility" title="Permalink to this headline">¶</a></h3>
+<p>When the IR format has to be changed, keep in mind that we try to maintain some
+backwards compatibility. The rules are intended as a balance between convenience
+for llvm users and not imposing a big burden on llvm developers:</p>
+<ul class="simple">
+<li>The textual format is not backwards compatible. We don’t change it too often,
+but there are no specific promises.</li>
+<li>Additions and changes to the IR should be reflected in
+<tt class="docutils literal"><span class="pre">test/Bitcode/compatibility.ll</span></tt>.</li>
+<li>The current LLVM version supports loading any bitcode since version 3.0.</li>
+<li>After each X.Y release, <tt class="docutils literal"><span class="pre">compatibility.ll</span></tt> must be copied to
+<tt class="docutils literal"><span class="pre">compatibility-X.Y.ll</span></tt>. The corresponding bitcode file should be assembled
+using the X.Y build and committed as <tt class="docutils literal"><span class="pre">compatibility-X.Y.ll.bc</span></tt>.</li>
+<li>Newer releases can ignore features from older releases, but they cannot
+miscompile them. For example, if nsw is ever replaced with something else,
+dropping it would be a valid way to upgrade the IR.</li>
+<li>Debug metadata is special in that it is currently dropped during upgrades.</li>
+<li>Non-debug metadata is defined to be safe to drop, so a valid way to upgrade
+it is to drop it. That is not very user friendly and a bit more effort is
+expected, but no promises are made.</li>
+</ul>
+</div>
+<div class="section" id="c-api-changes">
+<h3><a class="toc-backref" href="#id23">C API Changes</a><a class="headerlink" href="#c-api-changes" title="Permalink to this headline">¶</a></h3>
+<ul class="simple">
+<li>Stability Guarantees: The C API is, in general, a “best effort” for stability.
+This means that we make every attempt to keep the C API stable, but that
+stability will be limited by the abstractness of the interface and the
+stability of the C++ API that it wraps. In practice, this means that things
+like “create debug info” or “create this type of instruction” are likely to be
+less stable than “take this IR file and JIT it for my current machine”.</li>
+<li>Release stability: We won’t break the C API on the release branch with patches
+that go on that branch, with the exception that we will fix an unintentional
+C API break that will keep the release consistent with both the previous and
+next release.</li>
+<li>Testing: Patches to the C API are expected to come with tests just like any
+other patch.</li>
+<li>Including new things into the API: If an LLVM subcomponent has a C API already
+included, then expanding that C API is acceptable. Adding C API for
+subcomponents that don’t currently have one needs to be discussed on the
+mailing list for design and maintainability feedback prior to implementation.</li>
+<li>Documentation: Any changes to the C API are required to be documented in the
+release notes so that it’s clear to external users who do not follow the
+project how the C API is changing and evolving.</li>
+</ul>
+</div>
+<div class="section" id="new-targets">
+<h3><a class="toc-backref" href="#id24">New Targets</a><a class="headerlink" href="#new-targets" title="Permalink to this headline">¶</a></h3>
+<p>LLVM is very receptive to new targets, even experimental ones, but a number of
+problems can appear when adding new large portions of code, and back-ends are
+normally added in bulk. We have found that landing large pieces of new code
+and then trying to fix emergent problems in-tree is problematic for a variety
+of reasons.</p>
+<p>For these reasons, new targets are <em>always</em> added as <em>experimental</em> until
+they can be proven stable, and later moved to non-experimental. The difference
+between both classes is that experimental targets are not built by default
+(need to be added to -DLLVM_TARGETS_TO_BUILD at CMake time).</p>
+<p>The basic rules for a back-end to be upstreamed in <strong>experimental</strong> mode are:</p>
+<ul class="simple">
+<li>Every target must have a <a class="reference internal" href="#code-owners"><em>code owner</em></a>. The <cite>CODE_OWNERS.TXT</cite>
+file has to be updated as part of the first merge. The code owner makes sure
+that changes to the target get reviewed and steers the overall effort.</li>
+<li>There must be an active community behind the target. This community
+will help maintain the target by providing buildbots, fixing
+bugs, answering the LLVM community’s questions and making sure the new
+target doesn’t break any of the other targets, or generic code. This
+behavior is expected to continue throughout the lifetime of the
+target’s code.</li>
+<li>The code must be free of contentious issues, for example, large
+changes in how the IR behaves or should be formed by the front-ends,
+unless agreed by the majority of the community via refactoring of the
+(<a class="reference internal" href="LangRef.html"><em>IR standard</em></a>) <strong>before</strong> the merge of the new target changes,
+following the <a class="reference internal" href="#ir-backwards-compatibility"><em>IR Backwards Compatibility</em></a>.</li>
+<li>The code conforms to all of the policies laid out in this developer policy
+document, including license, patent, and coding standards.</li>
+<li>The target should have either reasonable documentation on how it
+works (ISA, ABI, etc.) or a publicly available simulator/hardware
+(either free or cheap enough) - preferably both. This allows
+developers to validate assumptions, understand constraints and review code
+that can affect the target.</li>
+</ul>
+<p>In addition, the rules for a back-end to be promoted to <strong>official</strong> are:</p>
+<ul class="simple">
+<li>The target must have addressed every other minimum requirement and
+have been stable in tree for at least 3 months. This cool down
+period is to make sure that the back-end and the target community can
+endure continuous upstream development for the foreseeable future.</li>
+<li>The target’s code must have been completely adapted to this policy
+as well as the <a class="reference internal" href="CodingStandards.html"><em>coding standards</em></a>. Any exceptions that
+were made to move into experimental mode must have been fixed <strong>before</strong>
+becoming official.</li>
+<li>The test coverage needs to be broad and well written (small tests,
+well documented). The build target <tt class="docutils literal"><span class="pre">check-all</span></tt> must pass with the
+new target built, and where applicable, the <tt class="docutils literal"><span class="pre">test-suite</span></tt> must also
+pass without errors, in at least one configuration (publicly
+demonstrated, for example, via buildbots).</li>
+<li>Public buildbots need to be created and actively maintained, unless
+the target requires no additional buildbots (ex. <tt class="docutils literal"><span class="pre">check-all</span></tt> covers
+all tests). The more relevant and public the new target’s CI infrastructure
+is, the more the LLVM community will embrace it.</li>
+</ul>
+<p>To <strong>continue</strong> as a supported and official target:</p>
+<ul class="simple">
+<li>The maintainer(s) must continue following these rules throughout the lifetime
+of the target. Continuous violations of aforementioned rules and policies
+could lead to complete removal of the target from the code base.</li>
+<li>Degradation in support, documentation or test coverage will make the target as
+nuisance to other targets and be considered a candidate for deprecation and
+ultimately removed.</li>
+</ul>
+<p>In essences, these rules are necessary for targets to gain and retain their
+status, but also markers to define bit-rot, and will be used to clean up the
+tree from unmaintained targets.</p>
+</div>
+</div>
+<div class="section" id="copyright-license-and-patents">
+<span id="copyright-license-patents"></span><h2><a class="toc-backref" href="#id25">Copyright, License, and Patents</a><a class="headerlink" href="#copyright-license-and-patents" title="Permalink to this headline">¶</a></h2>
+<div class="admonition note">
+<p class="first admonition-title">Note</p>
+<p class="last">This section deals with legal matters but does not provide legal advice. We
+are not lawyers — please seek legal counsel from an attorney.</p>
+</div>
+<p>This section addresses the issues of copyright, license and patents for the LLVM
+project. The copyright for the code is held by the individual contributors of
+the code and the terms of its license to LLVM users and developers is the
+<a class="reference external" href="http://www.opensource.org/licenses/UoI-NCSA.php">University of Illinois/NCSA Open Source License</a> (with portions dual licensed
+under the <a class="reference external" href="http://www.opensource.org/licenses/mit-license.php">MIT License</a>,
+see below). As contributor to the LLVM project, you agree to allow any
+contributions to the project to licensed under these terms.</p>
+<div class="section" id="copyright">
+<h3><a class="toc-backref" href="#id26">Copyright</a><a class="headerlink" href="#copyright" title="Permalink to this headline">¶</a></h3>
+<p>The LLVM project does not require copyright assignments, which means that the
+copyright for the code in the project is held by its respective contributors who
+have each agreed to release their contributed code under the terms of the <a class="reference internal" href="#llvm-license">LLVM
+License</a>.</p>
+<p>An implication of this is that the LLVM license is unlikely to ever change:
+changing it would require tracking down all the contributors to LLVM and getting
+them to agree that a license change is acceptable for their contribution. Since
+there are no plans to change the license, this is not a cause for concern.</p>
+<p>As a contributor to the project, this means that you (or your company) retain
+ownership of the code you contribute, that it cannot be used in a way that
+contradicts the license (which is a liberal BSD-style license), and that the
+license for your contributions won’t change without your approval in the
+future.</p>
+</div>
+<div class="section" id="license">
+<span id="llvm-license"></span><h3><a class="toc-backref" href="#id27">License</a><a class="headerlink" href="#license" title="Permalink to this headline">¶</a></h3>
+<p>We intend to keep LLVM perpetually open source and to use a liberal open source
+license. <strong>As a contributor to the project, you agree that any contributions be
+licensed under the terms of the corresponding subproject.</strong> All of the code in
+LLVM is available under the <a class="reference external" href="http://www.opensource.org/licenses/UoI-NCSA.php">University of Illinois/NCSA Open Source License</a>, which boils down to
+this:</p>
+<ul class="simple">
+<li>You can freely distribute LLVM.</li>
+<li>You must retain the copyright notice if you redistribute LLVM.</li>
+<li>Binaries derived from LLVM must reproduce the copyright notice (e.g. in an
+included readme file).</li>
+<li>You can’t use our names to promote your LLVM derived products.</li>
+<li>There’s no warranty on LLVM at all.</li>
+</ul>
+<p>We believe this fosters the widest adoption of LLVM because it <strong>allows
+commercial products to be derived from LLVM</strong> with few restrictions and without
+a requirement for making any derived works also open source (i.e. LLVM’s
+license is not a “copyleft” license like the GPL). We suggest that you read the
+<a class="reference external" href="http://www.opensource.org/licenses/UoI-NCSA.php">License</a> if further
+clarification is needed.</p>
+<p>In addition to the UIUC license, the runtime library components of LLVM
+(<strong>compiler_rt, libc++, and libclc</strong>) are also licensed under the <a class="reference external" href="http://www.opensource.org/licenses/mit-license.php">MIT License</a>, which does not contain
+the binary redistribution clause. As a user of these runtime libraries, it
+means that you can choose to use the code under either license (and thus don’t
+need the binary redistribution clause), and as a contributor to the code that
+you agree that any contributions to these libraries be licensed under both
+licenses. We feel that this is important for runtime libraries, because they
+are implicitly linked into applications and therefore should not subject those
+applications to the binary redistribution clause. This also means that it is ok
+to move code from (e.g.) libc++ to the LLVM core without concern, but that code
+cannot be moved from the LLVM core to libc++ without the copyright owner’s
+permission.</p>
+<p>Note that the LLVM Project does distribute dragonegg, <strong>which is
+GPL.</strong> This means that anything “linked” into dragonegg must itself be compatible
+with the GPL, and must be releasable under the terms of the GPL. This implies
+that <strong>any code linked into dragonegg and distributed to others may be subject to
+the viral aspects of the GPL</strong> (for example, a proprietary code generator linked
+into dragonegg must be made available under the GPL). This is not a problem for
+code already distributed under a more liberal license (like the UIUC license),
+and GPL-containing subprojects are kept in separate SVN repositories whose
+LICENSE.txt files specifically indicate that they contain GPL code.</p>
+<p>We have no plans to change the license of LLVM. If you have questions or
+comments about the license, please contact the <a class="reference external" href="mailto:llvm-dev%40lists.llvm.org">LLVM Developer's Mailing
+List</a>.</p>
+</div>
+<div class="section" id="patents">
+<h3><a class="toc-backref" href="#id28">Patents</a><a class="headerlink" href="#patents" title="Permalink to this headline">¶</a></h3>
+<p>To the best of our knowledge, LLVM does not infringe on any patents (we have
+actually removed code from LLVM in the past that was found to infringe). Having
+code in LLVM that infringes on patents would violate an important goal of the
+project by making it hard or impossible to reuse the code for arbitrary purposes
+(including commercial use).</p>
+<p>When contributing code, we expect contributors to notify us of any potential for
+patent-related trouble with their changes (including from third parties). If
+you or your employer own the rights to a patent and would like to contribute
+code to LLVM that relies on it, we require that the copyright owner sign an
+agreement that allows any other user of LLVM to freely use your patent. Please
+contact the <a class="reference external" href="mailto:board%40llvm.org">LLVM Foundation Board of Directors</a> for more
+details.</p>
+</div>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="Projects.html" title="Creating an LLVM Project"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="PDB/CodeViewTypes.html" title="CodeView Type Records"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/ExceptionHandling.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/ExceptionHandling.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/ExceptionHandling.html (added)
+++ www-releases/trunk/4.0.0/docs/ExceptionHandling.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,868 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>Exception Handling in LLVM — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="LLVM Link Time Optimization: Design and Implementation" href="LinkTimeOptimization.html" />
+ <link rel="prev" title="The LLVM Target-Independent Code Generator" href="CodeGenerator.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="LinkTimeOptimization.html" title="LLVM Link Time Optimization: Design and Implementation"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="CodeGenerator.html" title="The LLVM Target-Independent Code Generator"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="exception-handling-in-llvm">
+<h1>Exception Handling in LLVM<a class="headerlink" href="#exception-handling-in-llvm" title="Permalink to this headline">¶</a></h1>
+<div class="contents local topic" id="contents">
+<ul class="simple">
+<li><a class="reference internal" href="#introduction" id="id7">Introduction</a><ul>
+<li><a class="reference internal" href="#itanium-abi-zero-cost-exception-handling" id="id8">Itanium ABI Zero-cost Exception Handling</a></li>
+<li><a class="reference internal" href="#setjmp-longjmp-exception-handling" id="id9">Setjmp/Longjmp Exception Handling</a></li>
+<li><a class="reference internal" href="#windows-runtime-exception-handling" id="id10">Windows Runtime Exception Handling</a></li>
+<li><a class="reference internal" href="#overview" id="id11">Overview</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#llvm-code-generation" id="id12">LLVM Code Generation</a><ul>
+<li><a class="reference internal" href="#throw" id="id13">Throw</a></li>
+<li><a class="reference internal" href="#try-catch" id="id14">Try/Catch</a></li>
+<li><a class="reference internal" href="#cleanups" id="id15">Cleanups</a></li>
+<li><a class="reference internal" href="#throw-filters" id="id16">Throw Filters</a></li>
+<li><a class="reference internal" href="#restrictions" id="id17">Restrictions</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#exception-handling-intrinsics" id="id18">Exception Handling Intrinsics</a><ul>
+<li><a class="reference internal" href="#llvm-eh-typeid-for" id="id19"><tt class="docutils literal"><span class="pre">llvm.eh.typeid.for</span></tt></a></li>
+<li><a class="reference internal" href="#llvm-eh-begincatch" id="id20"><tt class="docutils literal"><span class="pre">llvm.eh.begincatch</span></tt></a></li>
+<li><a class="reference internal" href="#llvm-eh-endcatch" id="id21"><tt class="docutils literal"><span class="pre">llvm.eh.endcatch</span></tt></a></li>
+<li><a class="reference internal" href="#llvm-eh-exceptionpointer" id="id22"><tt class="docutils literal"><span class="pre">llvm.eh.exceptionpointer</span></tt></a></li>
+<li><a class="reference internal" href="#sjlj-intrinsics" id="id23">SJLJ Intrinsics</a><ul>
+<li><a class="reference internal" href="#llvm-eh-sjlj-setjmp" id="id24"><tt class="docutils literal"><span class="pre">llvm.eh.sjlj.setjmp</span></tt></a></li>
+<li><a class="reference internal" href="#llvm-eh-sjlj-longjmp" id="id25"><tt class="docutils literal"><span class="pre">llvm.eh.sjlj.longjmp</span></tt></a></li>
+<li><a class="reference internal" href="#llvm-eh-sjlj-lsda" id="id26"><tt class="docutils literal"><span class="pre">llvm.eh.sjlj.lsda</span></tt></a></li>
+<li><a class="reference internal" href="#llvm-eh-sjlj-callsite" id="id27"><tt class="docutils literal"><span class="pre">llvm.eh.sjlj.callsite</span></tt></a></li>
+</ul>
+</li>
+</ul>
+</li>
+<li><a class="reference internal" href="#asm-table-formats" id="id28">Asm Table Formats</a><ul>
+<li><a class="reference internal" href="#exception-handling-frame" id="id29">Exception Handling Frame</a></li>
+<li><a class="reference internal" href="#exception-tables" id="id30">Exception Tables</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#exception-handling-using-the-windows-runtime" id="id31">Exception Handling using the Windows Runtime</a><ul>
+<li><a class="reference internal" href="#background-on-windows-exceptions" id="id32">Background on Windows exceptions</a></li>
+<li><a class="reference internal" href="#seh-filter-expressions" id="id33">SEH filter expressions</a></li>
+<li><a class="reference internal" href="#new-exception-handling-instructions" id="id34">New exception handling instructions</a></li>
+<li><a class="reference internal" href="#funclet-parent-tokens" id="id35">Funclet parent tokens</a></li>
+<li><a class="reference internal" href="#funclet-transitions" id="id36">Funclet transitions</a></li>
+</ul>
+</li>
+</ul>
+</div>
+<div class="section" id="introduction">
+<h2><a class="toc-backref" href="#id7">Introduction</a><a class="headerlink" href="#introduction" title="Permalink to this headline">¶</a></h2>
+<p>This document is the central repository for all information pertaining to
+exception handling in LLVM. It describes the format that LLVM exception
+handling information takes, which is useful for those interested in creating
+front-ends or dealing directly with the information. Further, this document
+provides specific examples of what exception handling information is used for in
+C and C++.</p>
+<div class="section" id="itanium-abi-zero-cost-exception-handling">
+<h3><a class="toc-backref" href="#id8">Itanium ABI Zero-cost Exception Handling</a><a class="headerlink" href="#itanium-abi-zero-cost-exception-handling" title="Permalink to this headline">¶</a></h3>
+<p>Exception handling for most programming languages is designed to recover from
+conditions that rarely occur during general use of an application. To that end,
+exception handling should not interfere with the main flow of an application’s
+algorithm by performing checkpointing tasks, such as saving the current pc or
+register state.</p>
+<p>The Itanium ABI Exception Handling Specification defines a methodology for
+providing outlying data in the form of exception tables without inlining
+speculative exception handling code in the flow of an application’s main
+algorithm. Thus, the specification is said to add “zero-cost” to the normal
+execution of an application.</p>
+<p>A more complete description of the Itanium ABI exception handling runtime
+support of can be found at <a class="reference external" href="http://mentorembedded.github.com/cxx-abi/abi-eh.html">Itanium C++ ABI: Exception Handling</a>. A description of the
+exception frame format can be found at <a class="reference external" href="http://refspecs.linuxfoundation.org/LSB_3.0.0/LSB-Core-generic/LSB-Core-generic/ehframechpt.html">Exception Frames</a>,
+with details of the DWARF 4 specification at <a class="reference external" href="http://dwarfstd.org/Dwarf4Std.php">DWARF 4 Standard</a>. A description for the C++ exception
+table formats can be found at <a class="reference external" href="http://mentorembedded.github.com/cxx-abi/exceptions.pdf">Exception Handling Tables</a>.</p>
+</div>
+<div class="section" id="setjmp-longjmp-exception-handling">
+<h3><a class="toc-backref" href="#id9">Setjmp/Longjmp Exception Handling</a><a class="headerlink" href="#setjmp-longjmp-exception-handling" title="Permalink to this headline">¶</a></h3>
+<p>Setjmp/Longjmp (SJLJ) based exception handling uses LLVM intrinsics
+<a class="reference internal" href="#llvm-eh-sjlj-setjmp">llvm.eh.sjlj.setjmp</a> and <a class="reference internal" href="#llvm-eh-sjlj-longjmp">llvm.eh.sjlj.longjmp</a> to handle control flow for
+exception handling.</p>
+<p>For each function which does exception processing — be it <tt class="docutils literal"><span class="pre">try</span></tt>/<tt class="docutils literal"><span class="pre">catch</span></tt>
+blocks or cleanups — that function registers itself on a global frame
+list. When exceptions are unwinding, the runtime uses this list to identify
+which functions need processing.</p>
+<p>Landing pad selection is encoded in the call site entry of the function
+context. The runtime returns to the function via <a class="reference internal" href="#llvm-eh-sjlj-longjmp">llvm.eh.sjlj.longjmp</a>, where
+a switch table transfers control to the appropriate landing pad based on the
+index stored in the function context.</p>
+<p>In contrast to DWARF exception handling, which encodes exception regions and
+frame information in out-of-line tables, SJLJ exception handling builds and
+removes the unwind frame context at runtime. This results in faster exception
+handling at the expense of slower execution when no exceptions are thrown. As
+exceptions are, by their nature, intended for uncommon code paths, DWARF
+exception handling is generally preferred to SJLJ.</p>
+</div>
+<div class="section" id="windows-runtime-exception-handling">
+<h3><a class="toc-backref" href="#id10">Windows Runtime Exception Handling</a><a class="headerlink" href="#windows-runtime-exception-handling" title="Permalink to this headline">¶</a></h3>
+<p>LLVM supports handling exceptions produced by the Windows runtime, but it
+requires a very different intermediate representation. It is not based on the
+“<a class="reference internal" href="LangRef.html#i-landingpad"><em>landingpad</em></a>” instruction like the other two models, and is
+described later in this document under <a class="reference internal" href="#wineh"><em>Exception Handling using the Windows Runtime</em></a>.</p>
+</div>
+<div class="section" id="overview">
+<h3><a class="toc-backref" href="#id11">Overview</a><a class="headerlink" href="#overview" title="Permalink to this headline">¶</a></h3>
+<p>When an exception is thrown in LLVM code, the runtime does its best to find a
+handler suited to processing the circumstance.</p>
+<p>The runtime first attempts to find an <em>exception frame</em> corresponding to the
+function where the exception was thrown. If the programming language supports
+exception handling (e.g. C++), the exception frame contains a reference to an
+exception table describing how to process the exception. If the language does
+not support exception handling (e.g. C), or if the exception needs to be
+forwarded to a prior activation, the exception frame contains information about
+how to unwind the current activation and restore the state of the prior
+activation. This process is repeated until the exception is handled. If the
+exception is not handled and no activations remain, then the application is
+terminated with an appropriate error message.</p>
+<p>Because different programming languages have different behaviors when handling
+exceptions, the exception handling ABI provides a mechanism for
+supplying <em>personalities</em>. An exception handling personality is defined by
+way of a <em>personality function</em> (e.g. <tt class="docutils literal"><span class="pre">__gxx_personality_v0</span></tt> in C++),
+which receives the context of the exception, an <em>exception structure</em>
+containing the exception object type and value, and a reference to the exception
+table for the current function. The personality function for the current
+compile unit is specified in a <em>common exception frame</em>.</p>
+<p>The organization of an exception table is language dependent. For C++, an
+exception table is organized as a series of code ranges defining what to do if
+an exception occurs in that range. Typically, the information associated with a
+range defines which types of exception objects (using C++ <em>type info</em>) that are
+handled in that range, and an associated action that should take place. Actions
+typically pass control to a <em>landing pad</em>.</p>
+<p>A landing pad corresponds roughly to the code found in the <tt class="docutils literal"><span class="pre">catch</span></tt> portion of
+a <tt class="docutils literal"><span class="pre">try</span></tt>/<tt class="docutils literal"><span class="pre">catch</span></tt> sequence. When execution resumes at a landing pad, it
+receives an <em>exception structure</em> and a <em>selector value</em> corresponding to the
+<em>type</em> of exception thrown. The selector is then used to determine which <em>catch</em>
+should actually process the exception.</p>
+</div>
+</div>
+<div class="section" id="llvm-code-generation">
+<h2><a class="toc-backref" href="#id12">LLVM Code Generation</a><a class="headerlink" href="#llvm-code-generation" title="Permalink to this headline">¶</a></h2>
+<p>From a C++ developer’s perspective, exceptions are defined in terms of the
+<tt class="docutils literal"><span class="pre">throw</span></tt> and <tt class="docutils literal"><span class="pre">try</span></tt>/<tt class="docutils literal"><span class="pre">catch</span></tt> statements. In this section we will describe the
+implementation of LLVM exception handling in terms of C++ examples.</p>
+<div class="section" id="throw">
+<h3><a class="toc-backref" href="#id13">Throw</a><a class="headerlink" href="#throw" title="Permalink to this headline">¶</a></h3>
+<p>Languages that support exception handling typically provide a <tt class="docutils literal"><span class="pre">throw</span></tt>
+operation to initiate the exception process. Internally, a <tt class="docutils literal"><span class="pre">throw</span></tt> operation
+breaks down into two steps.</p>
+<ol class="arabic simple">
+<li>A request is made to allocate exception space for an exception structure.
+This structure needs to survive beyond the current activation. This structure
+will contain the type and value of the object being thrown.</li>
+<li>A call is made to the runtime to raise the exception, passing the exception
+structure as an argument.</li>
+</ol>
+<p>In C++, the allocation of the exception structure is done by the
+<tt class="docutils literal"><span class="pre">__cxa_allocate_exception</span></tt> runtime function. The exception raising is handled
+by <tt class="docutils literal"><span class="pre">__cxa_throw</span></tt>. The type of the exception is represented using a C++ RTTI
+structure.</p>
+</div>
+<div class="section" id="try-catch">
+<h3><a class="toc-backref" href="#id14">Try/Catch</a><a class="headerlink" href="#try-catch" title="Permalink to this headline">¶</a></h3>
+<p>A call within the scope of a <em>try</em> statement can potentially raise an
+exception. In those circumstances, the LLVM C++ front-end replaces the call with
+an <tt class="docutils literal"><span class="pre">invoke</span></tt> instruction. Unlike a call, the <tt class="docutils literal"><span class="pre">invoke</span></tt> has two potential
+continuation points:</p>
+<ol class="arabic simple">
+<li>where to continue when the call succeeds as per normal, and</li>
+<li>where to continue if the call raises an exception, either by a throw or the
+unwinding of a throw</li>
+</ol>
+<p>The term used to define the place where an <tt class="docutils literal"><span class="pre">invoke</span></tt> continues after an
+exception is called a <em>landing pad</em>. LLVM landing pads are conceptually
+alternative function entry points where an exception structure reference and a
+type info index are passed in as arguments. The landing pad saves the exception
+structure reference and then proceeds to select the catch block that corresponds
+to the type info of the exception object.</p>
+<p>The LLVM <a class="reference internal" href="LangRef.html#i-landingpad"><em>‘landingpad’ Instruction</em></a> is used to convey information about the landing
+pad to the back end. For C++, the <tt class="docutils literal"><span class="pre">landingpad</span></tt> instruction returns a pointer
+and integer pair corresponding to the pointer to the <em>exception structure</em> and
+the <em>selector value</em> respectively.</p>
+<p>The <tt class="docutils literal"><span class="pre">landingpad</span></tt> instruction looks for a reference to the personality
+function to be used for this <tt class="docutils literal"><span class="pre">try</span></tt>/<tt class="docutils literal"><span class="pre">catch</span></tt> sequence in the parent
+function’s attribute list. The instruction contains a list of <em>cleanup</em>,
+<em>catch</em>, and <em>filter</em> clauses. The exception is tested against the clauses
+sequentially from first to last. The clauses have the following meanings:</p>
+<ul>
+<li><p class="first"><tt class="docutils literal"><span class="pre">catch</span> <span class="pre"><type></span> <span class="pre">@ExcType</span></tt></p>
+<ul class="simple">
+<li>This clause means that the landingpad block should be entered if the
+exception being thrown is of type <tt class="docutils literal"><span class="pre">@ExcType</span></tt> or a subtype of
+<tt class="docutils literal"><span class="pre">@ExcType</span></tt>. For C++, <tt class="docutils literal"><span class="pre">@ExcType</span></tt> is a pointer to the <tt class="docutils literal"><span class="pre">std::type_info</span></tt>
+object (an RTTI object) representing the C++ exception type.</li>
+<li>If <tt class="docutils literal"><span class="pre">@ExcType</span></tt> is <tt class="docutils literal"><span class="pre">null</span></tt>, any exception matches, so the landingpad
+should always be entered. This is used for C++ catch-all blocks (“<tt class="docutils literal"><span class="pre">catch</span>
+<span class="pre">(...)</span></tt>”).</li>
+<li>When this clause is matched, the selector value will be equal to the value
+returned by “<tt class="docutils literal"><span class="pre">@llvm.eh.typeid.for(i8*</span> <span class="pre">@ExcType)</span></tt>”. This will always be a
+positive value.</li>
+</ul>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">filter</span> <span class="pre"><type></span> <span class="pre">[<type></span> <span class="pre">@ExcType1,</span> <span class="pre">...,</span> <span class="pre"><type></span> <span class="pre">@ExcTypeN]</span></tt></p>
+<ul class="simple">
+<li>This clause means that the landingpad should be entered if the exception
+being thrown does <em>not</em> match any of the types in the list (which, for C++,
+are again specified as <tt class="docutils literal"><span class="pre">std::type_info</span></tt> pointers).</li>
+<li>C++ front-ends use this to implement C++ exception specifications, such as
+“<tt class="docutils literal"><span class="pre">void</span> <span class="pre">foo()</span> <span class="pre">throw</span> <span class="pre">(ExcType1,</span> <span class="pre">...,</span> <span class="pre">ExcTypeN)</span> <span class="pre">{</span> <span class="pre">...</span> <span class="pre">}</span></tt>”.</li>
+<li>When this clause is matched, the selector value will be negative.</li>
+<li>The array argument to <tt class="docutils literal"><span class="pre">filter</span></tt> may be empty; for example, “<tt class="docutils literal"><span class="pre">[0</span> <span class="pre">x</span> <span class="pre">i8**]</span>
+<span class="pre">undef</span></tt>”. This means that the landingpad should always be entered. (Note
+that such a <tt class="docutils literal"><span class="pre">filter</span></tt> would not be equivalent to “<tt class="docutils literal"><span class="pre">catch</span> <span class="pre">i8*</span> <span class="pre">null</span></tt>”,
+because <tt class="docutils literal"><span class="pre">filter</span></tt> and <tt class="docutils literal"><span class="pre">catch</span></tt> produce negative and positive selector
+values respectively.)</li>
+</ul>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">cleanup</span></tt></p>
+<ul>
+<li><p class="first">This clause means that the landingpad should always be entered.</p>
+</li>
+<li><p class="first">C++ front-ends use this for calling objects’ destructors.</p>
+</li>
+<li><p class="first">When this clause is matched, the selector value will be zero.</p>
+</li>
+<li><p class="first">The runtime may treat “<tt class="docutils literal"><span class="pre">cleanup</span></tt>” differently from “<tt class="docutils literal"><span class="pre">catch</span> <span class="pre"><type></span>
+<span class="pre">null</span></tt>”.</p>
+<p>In C++, if an unhandled exception occurs, the language runtime will call
+<tt class="docutils literal"><span class="pre">std::terminate()</span></tt>, but it is implementation-defined whether the runtime
+unwinds the stack and calls object destructors first. For example, the GNU
+C++ unwinder does not call object destructors when an unhandled exception
+occurs. The reason for this is to improve debuggability: it ensures that
+<tt class="docutils literal"><span class="pre">std::terminate()</span></tt> is called from the context of the <tt class="docutils literal"><span class="pre">throw</span></tt>, so that
+this context is not lost by unwinding the stack. A runtime will typically
+implement this by searching for a matching non-<tt class="docutils literal"><span class="pre">cleanup</span></tt> clause, and
+aborting if it does not find one, before entering any landingpad blocks.</p>
+</li>
+</ul>
+</li>
+</ul>
+<p>Once the landing pad has the type info selector, the code branches to the code
+for the first catch. The catch then checks the value of the type info selector
+against the index of type info for that catch. Since the type info index is not
+known until all the type infos have been gathered in the backend, the catch code
+must call the <a class="reference internal" href="#llvm-eh-typeid-for">llvm.eh.typeid.for</a> intrinsic to determine the index for a given
+type info. If the catch fails to match the selector then control is passed on to
+the next catch.</p>
+<p>Finally, the entry and exit of catch code is bracketed with calls to
+<tt class="docutils literal"><span class="pre">__cxa_begin_catch</span></tt> and <tt class="docutils literal"><span class="pre">__cxa_end_catch</span></tt>.</p>
+<ul>
+<li><p class="first"><tt class="docutils literal"><span class="pre">__cxa_begin_catch</span></tt> takes an exception structure reference as an argument
+and returns the value of the exception object.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">__cxa_end_catch</span></tt> takes no arguments. This function:</p>
+<ol class="arabic simple">
+<li>Locates the most recently caught exception and decrements its handler
+count,</li>
+<li>Removes the exception from the <em>caught</em> stack if the handler count goes to
+zero, and</li>
+<li>Destroys the exception if the handler count goes to zero and the exception
+was not re-thrown by throw.</li>
+</ol>
+<div class="admonition note">
+<p class="first admonition-title">Note</p>
+<p class="last">a rethrow from within the catch may replace this call with a
+<tt class="docutils literal"><span class="pre">__cxa_rethrow</span></tt>.</p>
+</div>
+</li>
+</ul>
+</div>
+<div class="section" id="cleanups">
+<h3><a class="toc-backref" href="#id15">Cleanups</a><a class="headerlink" href="#cleanups" title="Permalink to this headline">¶</a></h3>
+<p>A cleanup is extra code which needs to be run as part of unwinding a scope. C++
+destructors are a typical example, but other languages and language extensions
+provide a variety of different kinds of cleanups. In general, a landing pad may
+need to run arbitrary amounts of cleanup code before actually entering a catch
+block. To indicate the presence of cleanups, a <a class="reference internal" href="LangRef.html#i-landingpad"><em>‘landingpad’ Instruction</em></a> should have
+a <em>cleanup</em> clause. Otherwise, the unwinder will not stop at the landing pad if
+there are no catches or filters that require it to.</p>
+<div class="admonition note">
+<p class="first admonition-title">Note</p>
+<p class="last">Do not allow a new exception to propagate out of the execution of a
+cleanup. This can corrupt the internal state of the unwinder. Different
+languages describe different high-level semantics for these situations: for
+example, C++ requires that the process be terminated, whereas Ada cancels both
+exceptions and throws a third.</p>
+</div>
+<p>When all cleanups are finished, if the exception is not handled by the current
+function, resume unwinding by calling the <a class="reference internal" href="LangRef.html#i-resume"><em>resume instruction</em></a>,
+passing in the result of the <tt class="docutils literal"><span class="pre">landingpad</span></tt> instruction for the original
+landing pad.</p>
+</div>
+<div class="section" id="throw-filters">
+<h3><a class="toc-backref" href="#id16">Throw Filters</a><a class="headerlink" href="#throw-filters" title="Permalink to this headline">¶</a></h3>
+<p>C++ allows the specification of which exception types may be thrown from a
+function. To represent this, a top level landing pad may exist to filter out
+invalid types. To express this in LLVM code the <a class="reference internal" href="LangRef.html#i-landingpad"><em>‘landingpad’ Instruction</em></a> will have a
+filter clause. The clause consists of an array of type infos.
+<tt class="docutils literal"><span class="pre">landingpad</span></tt> will return a negative value
+if the exception does not match any of the type infos. If no match is found then
+a call to <tt class="docutils literal"><span class="pre">__cxa_call_unexpected</span></tt> should be made, otherwise
+<tt class="docutils literal"><span class="pre">_Unwind_Resume</span></tt>. Each of these functions requires a reference to the
+exception structure. Note that the most general form of a <tt class="docutils literal"><span class="pre">landingpad</span></tt>
+instruction can have any number of catch, cleanup, and filter clauses (though
+having more than one cleanup is pointless). The LLVM C++ front-end can generate
+such <tt class="docutils literal"><span class="pre">landingpad</span></tt> instructions due to inlining creating nested exception
+handling scopes.</p>
+</div>
+<div class="section" id="restrictions">
+<span id="undefined"></span><h3><a class="toc-backref" href="#id17">Restrictions</a><a class="headerlink" href="#restrictions" title="Permalink to this headline">¶</a></h3>
+<p>The unwinder delegates the decision of whether to stop in a call frame to that
+call frame’s language-specific personality function. Not all unwinders guarantee
+that they will stop to perform cleanups. For example, the GNU C++ unwinder
+doesn’t do so unless the exception is actually caught somewhere further up the
+stack.</p>
+<p>In order for inlining to behave correctly, landing pads must be prepared to
+handle selector results that they did not originally advertise. Suppose that a
+function catches exceptions of type <tt class="docutils literal"><span class="pre">A</span></tt>, and it’s inlined into a function that
+catches exceptions of type <tt class="docutils literal"><span class="pre">B</span></tt>. The inliner will update the <tt class="docutils literal"><span class="pre">landingpad</span></tt>
+instruction for the inlined landing pad to include the fact that <tt class="docutils literal"><span class="pre">B</span></tt> is also
+caught. If that landing pad assumes that it will only be entered to catch an
+<tt class="docutils literal"><span class="pre">A</span></tt>, it’s in for a rude awakening. Consequently, landing pads must test for
+the selector results they understand and then resume exception propagation with
+the <a class="reference external" href="LangRef.html#i_resume">resume instruction</a> if none of the conditions
+match.</p>
+</div>
+</div>
+<div class="section" id="exception-handling-intrinsics">
+<h2><a class="toc-backref" href="#id18">Exception Handling Intrinsics</a><a class="headerlink" href="#exception-handling-intrinsics" title="Permalink to this headline">¶</a></h2>
+<p>In addition to the <tt class="docutils literal"><span class="pre">landingpad</span></tt> and <tt class="docutils literal"><span class="pre">resume</span></tt> instructions, LLVM uses several
+intrinsic functions (name prefixed with <tt class="docutils literal"><span class="pre">llvm.eh</span></tt>) to provide exception
+handling information at various points in generated code.</p>
+<div class="section" id="llvm-eh-typeid-for">
+<span id="id1"></span><h3><a class="toc-backref" href="#id19"><tt class="docutils literal"><span class="pre">llvm.eh.typeid.for</span></tt></a><a class="headerlink" href="#llvm-eh-typeid-for" title="Permalink to this headline">¶</a></h3>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">i32</span> <span class="vg">@llvm.eh.typeid.for</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%type_info</span><span class="p">)</span>
+</pre></div>
+</div>
+<p>This intrinsic returns the type info index in the exception table of the current
+function. This value can be used to compare against the result of
+<tt class="docutils literal"><span class="pre">landingpad</span></tt> instruction. The single argument is a reference to a type info.</p>
+<p>Uses of this intrinsic are generated by the C++ front-end.</p>
+</div>
+<div class="section" id="llvm-eh-begincatch">
+<span id="id2"></span><h3><a class="toc-backref" href="#id20"><tt class="docutils literal"><span class="pre">llvm.eh.begincatch</span></tt></a><a class="headerlink" href="#llvm-eh-begincatch" title="Permalink to this headline">¶</a></h3>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="kt">void</span> <span class="vg">@llvm.eh.begincatch</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%ehptr</span><span class="p">,</span> <span class="k">i8</span><span class="p">*</span> <span class="nv">%ehobj</span><span class="p">)</span>
+</pre></div>
+</div>
+<p>This intrinsic marks the beginning of catch handling code within the blocks
+following a <tt class="docutils literal"><span class="pre">landingpad</span></tt> instruction. The exact behavior of this function
+depends on the compilation target and the personality function associated
+with the <tt class="docutils literal"><span class="pre">landingpad</span></tt> instruction.</p>
+<p>The first argument to this intrinsic is a pointer that was previously extracted
+from the aggregate return value of the <tt class="docutils literal"><span class="pre">landingpad</span></tt> instruction. The second
+argument to the intrinsic is a pointer to stack space where the exception object
+should be stored. The runtime handles the details of copying the exception
+object into the slot. If the second parameter is null, no copy occurs.</p>
+<p>Uses of this intrinsic are generated by the C++ front-end. Many targets will
+use implementation-specific functions (such as <tt class="docutils literal"><span class="pre">__cxa_begin_catch</span></tt>) instead
+of this intrinsic. The intrinsic is provided for targets that require a more
+abstract interface.</p>
+<p>When used in the native Windows C++ exception handling implementation, this
+intrinsic serves as a placeholder to delimit code before a catch handler is
+outlined. When the handler is is outlined, this intrinsic will be replaced
+by instructions that retrieve the exception object pointer from the frame
+allocation block.</p>
+</div>
+<div class="section" id="llvm-eh-endcatch">
+<span id="id3"></span><h3><a class="toc-backref" href="#id21"><tt class="docutils literal"><span class="pre">llvm.eh.endcatch</span></tt></a><a class="headerlink" href="#llvm-eh-endcatch" title="Permalink to this headline">¶</a></h3>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="kt">void</span> <span class="vg">@llvm.eh.endcatch</span><span class="p">()</span>
+</pre></div>
+</div>
+<p>This intrinsic marks the end of catch handling code within the current block,
+which will be a successor of a block which called <tt class="docutils literal"><span class="pre">llvm.eh.begincatch''.</span>
+<span class="pre">The</span> <span class="pre">exact</span> <span class="pre">behavior</span> <span class="pre">of</span> <span class="pre">this</span> <span class="pre">function</span> <span class="pre">depends</span> <span class="pre">on</span> <span class="pre">the</span> <span class="pre">compilation</span> <span class="pre">target</span> <span class="pre">and</span> <span class="pre">the</span>
+<span class="pre">personality</span> <span class="pre">function</span> <span class="pre">associated</span> <span class="pre">with</span> <span class="pre">the</span> <span class="pre">corresponding</span> <span class="pre">``landingpad</span></tt>
+instruction.</p>
+<p>There may be more than one call to <tt class="docutils literal"><span class="pre">llvm.eh.endcatch</span></tt> for any given call to
+<tt class="docutils literal"><span class="pre">llvm.eh.begincatch</span></tt> with each <tt class="docutils literal"><span class="pre">llvm.eh.endcatch</span></tt> call corresponding to the
+end of a different control path. All control paths following a call to
+<tt class="docutils literal"><span class="pre">llvm.eh.begincatch</span></tt> must reach a call to <tt class="docutils literal"><span class="pre">llvm.eh.endcatch</span></tt>.</p>
+<p>Uses of this intrinsic are generated by the C++ front-end. Many targets will
+use implementation-specific functions (such as <tt class="docutils literal"><span class="pre">__cxa_begin_catch</span></tt>) instead
+of this intrinsic. The intrinsic is provided for targets that require a more
+abstract interface.</p>
+<p>When used in the native Windows C++ exception handling implementation, this
+intrinsic serves as a placeholder to delimit code before a catch handler is
+outlined. After the handler is outlined, this intrinsic is simply removed.</p>
+</div>
+<div class="section" id="llvm-eh-exceptionpointer">
+<span id="id4"></span><h3><a class="toc-backref" href="#id22"><tt class="docutils literal"><span class="pre">llvm.eh.exceptionpointer</span></tt></a><a class="headerlink" href="#llvm-eh-exceptionpointer" title="Permalink to this headline">¶</a></h3>
+<div class="highlight-text"><div class="highlight"><pre>i8 addrspace(N)* @llvm.eh.padparam.pNi8(token %catchpad)
+</pre></div>
+</div>
+<p>This intrinsic retrieves a pointer to the exception caught by the given
+<tt class="docutils literal"><span class="pre">catchpad</span></tt>.</p>
+</div>
+<div class="section" id="sjlj-intrinsics">
+<h3><a class="toc-backref" href="#id23">SJLJ Intrinsics</a><a class="headerlink" href="#sjlj-intrinsics" title="Permalink to this headline">¶</a></h3>
+<p>The <tt class="docutils literal"><span class="pre">llvm.eh.sjlj</span></tt> intrinsics are used internally within LLVM’s
+backend. Uses of them are generated by the backend’s
+<tt class="docutils literal"><span class="pre">SjLjEHPrepare</span></tt> pass.</p>
+<div class="section" id="llvm-eh-sjlj-setjmp">
+<span id="id5"></span><h4><a class="toc-backref" href="#id24"><tt class="docutils literal"><span class="pre">llvm.eh.sjlj.setjmp</span></tt></a><a class="headerlink" href="#llvm-eh-sjlj-setjmp" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-text"><div class="highlight"><pre>i32 @llvm.eh.sjlj.setjmp(i8* %setjmp_buf)
+</pre></div>
+</div>
+<p>For SJLJ based exception handling, this intrinsic forces register saving for the
+current function and stores the address of the following instruction for use as
+a destination address by <a class="reference internal" href="#llvm-eh-sjlj-longjmp">llvm.eh.sjlj.longjmp</a>. The buffer format and the
+overall functioning of this intrinsic is compatible with the GCC
+<tt class="docutils literal"><span class="pre">__builtin_setjmp</span></tt> implementation allowing code built with the clang and GCC
+to interoperate.</p>
+<p>The single parameter is a pointer to a five word buffer in which the calling
+context is saved. The front end places the frame pointer in the first word, and
+the target implementation of this intrinsic should place the destination address
+for a <a class="reference internal" href="#llvm-eh-sjlj-longjmp">llvm.eh.sjlj.longjmp</a> in the second word. The following three words are
+available for use in a target-specific manner.</p>
+</div>
+<div class="section" id="llvm-eh-sjlj-longjmp">
+<span id="id6"></span><h4><a class="toc-backref" href="#id25"><tt class="docutils literal"><span class="pre">llvm.eh.sjlj.longjmp</span></tt></a><a class="headerlink" href="#llvm-eh-sjlj-longjmp" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="kt">void</span> <span class="vg">@llvm.eh.sjlj.longjmp</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%setjmp_buf</span><span class="p">)</span>
+</pre></div>
+</div>
+<p>For SJLJ based exception handling, the <tt class="docutils literal"><span class="pre">llvm.eh.sjlj.longjmp</span></tt> intrinsic is
+used to implement <tt class="docutils literal"><span class="pre">__builtin_longjmp()</span></tt>. The single parameter is a pointer to
+a buffer populated by <a class="reference internal" href="#llvm-eh-sjlj-setjmp">llvm.eh.sjlj.setjmp</a>. The frame pointer and stack
+pointer are restored from the buffer, then control is transferred to the
+destination address.</p>
+</div>
+<div class="section" id="llvm-eh-sjlj-lsda">
+<h4><a class="toc-backref" href="#id26"><tt class="docutils literal"><span class="pre">llvm.eh.sjlj.lsda</span></tt></a><a class="headerlink" href="#llvm-eh-sjlj-lsda" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">i8</span><span class="p">*</span> <span class="vg">@llvm.eh.sjlj.lsda</span><span class="p">()</span>
+</pre></div>
+</div>
+<p>For SJLJ based exception handling, the <tt class="docutils literal"><span class="pre">llvm.eh.sjlj.lsda</span></tt> intrinsic returns
+the address of the Language Specific Data Area (LSDA) for the current
+function. The SJLJ front-end code stores this address in the exception handling
+function context for use by the runtime.</p>
+</div>
+<div class="section" id="llvm-eh-sjlj-callsite">
+<h4><a class="toc-backref" href="#id27"><tt class="docutils literal"><span class="pre">llvm.eh.sjlj.callsite</span></tt></a><a class="headerlink" href="#llvm-eh-sjlj-callsite" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="kt">void</span> <span class="vg">@llvm.eh.sjlj.callsite</span><span class="p">(</span><span class="k">i32</span> <span class="nv">%call_site_num</span><span class="p">)</span>
+</pre></div>
+</div>
+<p>For SJLJ based exception handling, the <tt class="docutils literal"><span class="pre">llvm.eh.sjlj.callsite</span></tt> intrinsic
+identifies the callsite value associated with the following <tt class="docutils literal"><span class="pre">invoke</span></tt>
+instruction. This is used to ensure that landing pad entries in the LSDA are
+generated in matching order.</p>
+</div>
+</div>
+</div>
+<div class="section" id="asm-table-formats">
+<h2><a class="toc-backref" href="#id28">Asm Table Formats</a><a class="headerlink" href="#asm-table-formats" title="Permalink to this headline">¶</a></h2>
+<p>There are two tables that are used by the exception handling runtime to
+determine which actions should be taken when an exception is thrown.</p>
+<div class="section" id="exception-handling-frame">
+<h3><a class="toc-backref" href="#id29">Exception Handling Frame</a><a class="headerlink" href="#exception-handling-frame" title="Permalink to this headline">¶</a></h3>
+<p>An exception handling frame <tt class="docutils literal"><span class="pre">eh_frame</span></tt> is very similar to the unwind frame
+used by DWARF debug info. The frame contains all the information necessary to
+tear down the current frame and restore the state of the prior frame. There is
+an exception handling frame for each function in a compile unit, plus a common
+exception handling frame that defines information common to all functions in the
+unit.</p>
+<p>The format of this call frame information (CFI) is often platform-dependent,
+however. ARM, for example, defines their own format. Apple has their own compact
+unwind info format. On Windows, another format is used for all architectures
+since 32-bit x86. LLVM will emit whatever information is required by the
+target.</p>
+</div>
+<div class="section" id="exception-tables">
+<h3><a class="toc-backref" href="#id30">Exception Tables</a><a class="headerlink" href="#exception-tables" title="Permalink to this headline">¶</a></h3>
+<p>An exception table contains information about what actions to take when an
+exception is thrown in a particular part of a function’s code. This is typically
+referred to as the language-specific data area (LSDA). The format of the LSDA
+table is specific to the personality function, but the majority of personalities
+out there use a variation of the tables consumed by <tt class="docutils literal"><span class="pre">__gxx_personality_v0</span></tt>.
+There is one exception table per function, except leaf functions and functions
+that have calls only to non-throwing functions. They do not need an exception
+table.</p>
+</div>
+</div>
+<div class="section" id="exception-handling-using-the-windows-runtime">
+<span id="wineh"></span><h2><a class="toc-backref" href="#id31">Exception Handling using the Windows Runtime</a><a class="headerlink" href="#exception-handling-using-the-windows-runtime" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="background-on-windows-exceptions">
+<h3><a class="toc-backref" href="#id32">Background on Windows exceptions</a><a class="headerlink" href="#background-on-windows-exceptions" title="Permalink to this headline">¶</a></h3>
+<p>Interacting with exceptions on Windows is significantly more complicated than
+on Itanium C++ ABI platforms. The fundamental difference between the two models
+is that Itanium EH is designed around the idea of “successive unwinding,” while
+Windows EH is not.</p>
+<p>Under Itanium, throwing an exception typically involes allocating thread local
+memory to hold the exception, and calling into the EH runtime. The runtime
+identifies frames with appropriate exception handling actions, and successively
+resets the register context of the current thread to the most recently active
+frame with actions to run. In LLVM, execution resumes at a <tt class="docutils literal"><span class="pre">landingpad</span></tt>
+instruction, which produces register values provided by the runtime. If a
+function is only cleaning up allocated resources, the function is responsible
+for calling <tt class="docutils literal"><span class="pre">_Unwind_Resume</span></tt> to transition to the next most recently active
+frame after it is finished cleaning up. Eventually, the frame responsible for
+handling the exception calls <tt class="docutils literal"><span class="pre">__cxa_end_catch</span></tt> to destroy the exception,
+release its memory, and resume normal control flow.</p>
+<p>The Windows EH model does not use these successive register context resets.
+Instead, the active exception is typically described by a frame on the stack.
+In the case of C++ exceptions, the exception object is allocated in stack memory
+and its address is passed to <tt class="docutils literal"><span class="pre">__CxxThrowException</span></tt>. General purpose structured
+exceptions (SEH) are more analogous to Linux signals, and they are dispatched by
+userspace DLLs provided with Windows. Each frame on the stack has an assigned EH
+personality routine, which decides what actions to take to handle the exception.
+There are a few major personalities for C and C++ code: the C++ personality
+(<tt class="docutils literal"><span class="pre">__CxxFrameHandler3</span></tt>) and the SEH personalities (<tt class="docutils literal"><span class="pre">_except_handler3</span></tt>,
+<tt class="docutils literal"><span class="pre">_except_handler4</span></tt>, and <tt class="docutils literal"><span class="pre">__C_specific_handler</span></tt>). All of them implement
+cleanups by calling back into a “funclet” contained in the parent function.</p>
+<p>Funclets, in this context, are regions of the parent function that can be called
+as though they were a function pointer with a very special calling convention.
+The frame pointer of the parent frame is passed into the funclet either using
+the standard EBP register or as the first parameter register, depending on the
+architecture. The funclet implements the EH action by accessing local variables
+in memory through the frame pointer, and returning some appropriate value,
+continuing the EH process. No variables live in to or out of the funclet can be
+allocated in registers.</p>
+<p>The C++ personality also uses funclets to contain the code for catch blocks
+(i.e. all user code between the braces in <tt class="docutils literal"><span class="pre">catch</span> <span class="pre">(Type</span> <span class="pre">obj)</span> <span class="pre">{</span> <span class="pre">...</span> <span class="pre">}</span></tt>). The
+runtime must use funclets for catch bodies because the C++ exception object is
+allocated in a child stack frame of the function handling the exception. If the
+runtime rewound the stack back to frame of the catch, the memory holding the
+exception would be overwritten quickly by subsequent function calls. The use of
+funclets also allows <tt class="docutils literal"><span class="pre">__CxxFrameHandler3</span></tt> to implement rethrow without
+resorting to TLS. Instead, the runtime throws a special exception, and then uses
+SEH (<tt class="docutils literal"><span class="pre">__try</span> <span class="pre">/</span> <span class="pre">__except</span></tt>) to resume execution with new information in the child
+frame.</p>
+<p>In other words, the successive unwinding approach is incompatible with Visual
+C++ exceptions and general purpose Windows exception handling. Because the C++
+exception object lives in stack memory, LLVM cannot provide a custom personality
+function that uses landingpads. Similarly, SEH does not provide any mechanism
+to rethrow an exception or continue unwinding. Therefore, LLVM must use the IR
+constructs described later in this document to implement compatible exception
+handling.</p>
+</div>
+<div class="section" id="seh-filter-expressions">
+<h3><a class="toc-backref" href="#id33">SEH filter expressions</a><a class="headerlink" href="#seh-filter-expressions" title="Permalink to this headline">¶</a></h3>
+<p>The SEH personality functions also use funclets to implement filter expressions,
+which allow executing arbitrary user code to decide which exceptions to catch.
+Filter expressions should not be confused with the <tt class="docutils literal"><span class="pre">filter</span></tt> clause of the LLVM
+<tt class="docutils literal"><span class="pre">landingpad</span></tt> instruction. Typically filter expressions are used to determine
+if the exception came from a particular DLL or code region, or if code faulted
+while accessing a particular memory address range. LLVM does not currently have
+IR to represent filter expressions because it is difficult to represent their
+control dependencies. Filter expressions run during the first phase of EH,
+before cleanups run, making it very difficult to build a faithful control flow
+graph. For now, the new EH instructions cannot represent SEH filter
+expressions, and frontends must outline them ahead of time. Local variables of
+the parent function can be escaped and accessed using the <tt class="docutils literal"><span class="pre">llvm.localescape</span></tt>
+and <tt class="docutils literal"><span class="pre">llvm.localrecover</span></tt> intrinsics.</p>
+</div>
+<div class="section" id="new-exception-handling-instructions">
+<h3><a class="toc-backref" href="#id34">New exception handling instructions</a><a class="headerlink" href="#new-exception-handling-instructions" title="Permalink to this headline">¶</a></h3>
+<p>The primary design goal of the new EH instructions is to support funclet
+generation while preserving information about the CFG so that SSA formation
+still works. As a secondary goal, they are designed to be generic across MSVC
+and Itanium C++ exceptions. They make very few assumptions about the data
+required by the personality, so long as it uses the familiar core EH actions:
+catch, cleanup, and terminate. However, the new instructions are hard to modify
+without knowing details of the EH personality. While they can be used to
+represent Itanium EH, the landingpad model is strictly better for optimization
+purposes.</p>
+<p>The following new instructions are considered “exception handling pads”, in that
+they must be the first non-phi instruction of a basic block that may be the
+unwind destination of an EH flow edge:
+<tt class="docutils literal"><span class="pre">catchswitch</span></tt>, <tt class="docutils literal"><span class="pre">catchpad</span></tt>, and <tt class="docutils literal"><span class="pre">cleanuppad</span></tt>.
+As with landingpads, when entering a try scope, if the
+frontend encounters a call site that may throw an exception, it should emit an
+invoke that unwinds to a <tt class="docutils literal"><span class="pre">catchswitch</span></tt> block. Similarly, inside the scope of a
+C++ object with a destructor, invokes should unwind to a <tt class="docutils literal"><span class="pre">cleanuppad</span></tt>.</p>
+<p>New instructions are also used to mark the points where control is transferred
+out of a catch/cleanup handler (which will correspond to exits from the
+generated funclet). A catch handler which reaches its end by normal execution
+executes a <tt class="docutils literal"><span class="pre">catchret</span></tt> instruction, which is a terminator indicating where in
+the function control is returned to. A cleanup handler which reaches its end
+by normal execution executes a <tt class="docutils literal"><span class="pre">cleanupret</span></tt> instruction, which is a terminator
+indicating where the active exception will unwind to next.</p>
+<p>Each of these new EH pad instructions has a way to identify which action should
+be considered after this action. The <tt class="docutils literal"><span class="pre">catchswitch</span></tt> instruction is a terminator
+and has an unwind destination operand analogous to the unwind destination of an
+invoke. The <tt class="docutils literal"><span class="pre">cleanuppad</span></tt> instruction is not
+a terminator, so the unwind destination is stored on the <tt class="docutils literal"><span class="pre">cleanupret</span></tt>
+instruction instead. Successfully executing a catch handler should resume
+normal control flow, so neither <tt class="docutils literal"><span class="pre">catchpad</span></tt> nor <tt class="docutils literal"><span class="pre">catchret</span></tt> instructions can
+unwind. All of these “unwind edges” may refer to a basic block that contains an
+EH pad instruction, or they may unwind to the caller. Unwinding to the caller
+has roughly the same semantics as the <tt class="docutils literal"><span class="pre">resume</span></tt> instruction in the landingpad
+model. When inlining through an invoke, instructions that unwind to the caller
+are hooked up to unwind to the unwind destination of the call site.</p>
+<p>Putting things together, here is a hypothetical lowering of some C++ that uses
+all of the new IR instructions:</p>
+<div class="highlight-c"><div class="highlight"><pre><span class="k">struct</span> <span class="n">Cleanup</span> <span class="p">{</span>
+ <span class="n">Cleanup</span><span class="p">();</span>
+ <span class="o">~</span><span class="n">Cleanup</span><span class="p">();</span>
+ <span class="kt">int</span> <span class="n">m</span><span class="p">;</span>
+<span class="p">};</span>
+<span class="kt">void</span> <span class="nf">may_throw</span><span class="p">();</span>
+<span class="kt">int</span> <span class="n">f</span><span class="p">()</span> <span class="n">noexcept</span> <span class="p">{</span>
+ <span class="n">try</span> <span class="p">{</span>
+ <span class="n">Cleanup</span> <span class="n">obj</span><span class="p">;</span>
+ <span class="n">may_throw</span><span class="p">();</span>
+ <span class="p">}</span> <span class="n">catch</span> <span class="p">(</span><span class="kt">int</span> <span class="n">e</span><span class="p">)</span> <span class="p">{</span>
+ <span class="n">may_throw</span><span class="p">();</span>
+ <span class="k">return</span> <span class="n">e</span><span class="p">;</span>
+ <span class="p">}</span>
+ <span class="k">return</span> <span class="mi">0</span><span class="p">;</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<div class="highlight-text"><div class="highlight"><pre>define i32 @f() nounwind personality i32 (...)* @__CxxFrameHandler3 {
+entry:
+ %obj = alloca %struct.Cleanup, align 4
+ %e = alloca i32, align 4
+ %call = invoke %struct.Cleanup* @"\01??0Cleanup@@QEAA at XZ"(%struct.Cleanup* nonnull %obj)
+ to label %invoke.cont unwind label %lpad.catch
+
+invoke.cont: ; preds = %entry
+ invoke void @"\01?may_throw@@YAXXZ"()
+ to label %invoke.cont.2 unwind label %lpad.cleanup
+
+invoke.cont.2: ; preds = %invoke.cont
+ call void @"\01??_DCleanup@@QEAA at XZ"(%struct.Cleanup* nonnull %obj) nounwind
+ br label %return
+
+return: ; preds = %invoke.cont.3, %invoke.cont.2
+ %retval.0 = phi i32 [ 0, %invoke.cont.2 ], [ %3, %invoke.cont.3 ]
+ ret i32 %retval.0
+
+lpad.cleanup: ; preds = %invoke.cont.2
+ %0 = cleanuppad within none []
+ call void @"\01??1Cleanup@@QEAA at XZ"(%struct.Cleanup* nonnull %obj) nounwind
+ cleanupret %0 unwind label %lpad.catch
+
+lpad.catch: ; preds = %lpad.cleanup, %entry
+ %1 = catchswitch within none [label %catch.body] unwind label %lpad.terminate
+
+catch.body: ; preds = %lpad.catch
+ %catch = catchpad within %1 [%rtti.TypeDescriptor2* @"\01??_R0H at 8", i32 0, i32* %e]
+ invoke void @"\01?may_throw@@YAXXZ"()
+ to label %invoke.cont.3 unwind label %lpad.terminate
+
+invoke.cont.3: ; preds = %catch.body
+ %3 = load i32, i32* %e, align 4
+ catchret from %catch to label %return
+
+lpad.terminate: ; preds = %catch.body, %lpad.catch
+ cleanuppad within none []
+ call void @"\01?terminate@@YAXXZ"
+ unreachable
+}
+</pre></div>
+</div>
+</div>
+<div class="section" id="funclet-parent-tokens">
+<h3><a class="toc-backref" href="#id35">Funclet parent tokens</a><a class="headerlink" href="#funclet-parent-tokens" title="Permalink to this headline">¶</a></h3>
+<p>In order to produce tables for EH personalities that use funclets, it is
+necessary to recover the nesting that was present in the source. This funclet
+parent relationship is encoded in the IR using tokens produced by the new “pad”
+instructions. The token operand of a “pad” or “ret” instruction indicates which
+funclet it is in, or “none” if it is not nested within another funclet.</p>
+<p>The <tt class="docutils literal"><span class="pre">catchpad</span></tt> and <tt class="docutils literal"><span class="pre">cleanuppad</span></tt> instructions establish new funclets, and
+their tokens are consumed by other “pad” instructions to establish membership.
+The <tt class="docutils literal"><span class="pre">catchswitch</span></tt> instruction does not create a funclet, but it produces a
+token that is always consumed by its immediate successor <tt class="docutils literal"><span class="pre">catchpad</span></tt>
+instructions. This ensures that every catch handler modelled by a <tt class="docutils literal"><span class="pre">catchpad</span></tt>
+belongs to exactly one <tt class="docutils literal"><span class="pre">catchswitch</span></tt>, which models the dispatch point after a
+C++ try.</p>
+<p>Here is an example of what this nesting looks like using some hypothetical
+C++ code:</p>
+<div class="highlight-c"><div class="highlight"><pre><span class="kt">void</span> <span class="nf">f</span><span class="p">()</span> <span class="p">{</span>
+ <span class="n">try</span> <span class="p">{</span>
+ <span class="n">throw</span><span class="p">;</span>
+ <span class="p">}</span> <span class="n">catch</span> <span class="p">(...)</span> <span class="p">{</span>
+ <span class="n">try</span> <span class="p">{</span>
+ <span class="n">throw</span><span class="p">;</span>
+ <span class="p">}</span> <span class="n">catch</span> <span class="p">(...)</span> <span class="p">{</span>
+ <span class="p">}</span>
+ <span class="p">}</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<div class="highlight-text"><div class="highlight"><pre>define void @f() #0 personality i8* bitcast (i32 (...)* @__CxxFrameHandler3 to i8*) {
+entry:
+ invoke void @_CxxThrowException(i8* null, %eh.ThrowInfo* null) #1
+ to label %unreachable unwind label %catch.dispatch
+
+catch.dispatch: ; preds = %entry
+ %0 = catchswitch within none [label %catch] unwind to caller
+
+catch: ; preds = %catch.dispatch
+ %1 = catchpad within %0 [i8* null, i32 64, i8* null]
+ invoke void @_CxxThrowException(i8* null, %eh.ThrowInfo* null) #1
+ to label %unreachable unwind label %catch.dispatch2
+
+catch.dispatch2: ; preds = %catch
+ %2 = catchswitch within %1 [label %catch3] unwind to caller
+
+catch3: ; preds = %catch.dispatch2
+ %3 = catchpad within %2 [i8* null, i32 64, i8* null]
+ catchret from %3 to label %try.cont
+
+try.cont: ; preds = %catch3
+ catchret from %1 to label %try.cont6
+
+try.cont6: ; preds = %try.cont
+ ret void
+
+unreachable: ; preds = %catch, %entry
+ unreachable
+}
+</pre></div>
+</div>
+<p>The “inner” <tt class="docutils literal"><span class="pre">catchswitch</span></tt> consumes <tt class="docutils literal"><span class="pre">%1</span></tt> which is produced by the outer
+catchswitch.</p>
+</div>
+<div class="section" id="funclet-transitions">
+<span id="wineh-constraints"></span><h3><a class="toc-backref" href="#id36">Funclet transitions</a><a class="headerlink" href="#funclet-transitions" title="Permalink to this headline">¶</a></h3>
+<p>The EH tables for personalities that use funclets make implicit use of the
+funclet nesting relationship to encode unwind destinations, and so are
+constrained in the set of funclet transitions they can represent. The related
+LLVM IR instructions accordingly have constraints that ensure encodability of
+the EH edges in the flow graph.</p>
+<p>A <tt class="docutils literal"><span class="pre">catchswitch</span></tt>, <tt class="docutils literal"><span class="pre">catchpad</span></tt>, or <tt class="docutils literal"><span class="pre">cleanuppad</span></tt> is said to be “entered”
+when it executes. It may subsequently be “exited” by any of the following
+means:</p>
+<ul class="simple">
+<li>A <tt class="docutils literal"><span class="pre">catchswitch</span></tt> is immediately exited when none of its constituent
+<tt class="docutils literal"><span class="pre">catchpad</span></tt>s are appropriate for the in-flight exception and it unwinds
+to its unwind destination or the caller.</li>
+<li>A <tt class="docutils literal"><span class="pre">catchpad</span></tt> and its parent <tt class="docutils literal"><span class="pre">catchswitch</span></tt> are both exited when a
+<tt class="docutils literal"><span class="pre">catchret</span></tt> from the <tt class="docutils literal"><span class="pre">catchpad</span></tt> is executed.</li>
+<li>A <tt class="docutils literal"><span class="pre">cleanuppad</span></tt> is exited when a <tt class="docutils literal"><span class="pre">cleanupret</span></tt> from it is executed.</li>
+<li>Any of these pads is exited when control unwinds to the function’s caller,
+either by a <tt class="docutils literal"><span class="pre">call</span></tt> which unwinds all the way to the function’s caller,
+a nested <tt class="docutils literal"><span class="pre">catchswitch</span></tt> marked “<tt class="docutils literal"><span class="pre">unwinds</span> <span class="pre">to</span> <span class="pre">caller</span></tt>”, or a nested
+<tt class="docutils literal"><span class="pre">cleanuppad</span></tt>‘s <tt class="docutils literal"><span class="pre">cleanupret</span></tt> marked “<tt class="docutils literal"><span class="pre">unwinds</span> <span class="pre">to</span> <span class="pre">caller"</span></tt>.</li>
+<li>Any of these pads is exited when an unwind edge (from an <tt class="docutils literal"><span class="pre">invoke</span></tt>,
+nested <tt class="docutils literal"><span class="pre">catchswitch</span></tt>, or nested <tt class="docutils literal"><span class="pre">cleanuppad</span></tt>‘s <tt class="docutils literal"><span class="pre">cleanupret</span></tt>)
+unwinds to a destination pad that is not a descendant of the given pad.</li>
+</ul>
+<p>Note that the <tt class="docutils literal"><span class="pre">ret</span></tt> instruction is <em>not</em> a valid way to exit a funclet pad;
+it is undefined behavior to execute a <tt class="docutils literal"><span class="pre">ret</span></tt> when a pad has been entered but
+not exited.</p>
+<p>A single unwind edge may exit any number of pads (with the restrictions that
+the edge from a <tt class="docutils literal"><span class="pre">catchswitch</span></tt> must exit at least itself, and the edge from
+a <tt class="docutils literal"><span class="pre">cleanupret</span></tt> must exit at least its <tt class="docutils literal"><span class="pre">cleanuppad</span></tt>), and then must enter
+exactly one pad, which must be distinct from all the exited pads. The parent
+of the pad that an unwind edge enters must be the most-recently-entered
+not-yet-exited pad (after exiting from any pads that the unwind edge exits),
+or “none” if there is no such pad. This ensures that the stack of executing
+funclets at run-time always corresponds to some path in the funclet pad tree
+that the parent tokens encode.</p>
+<p>All unwind edges which exit any given funclet pad (including <tt class="docutils literal"><span class="pre">cleanupret</span></tt>
+edges exiting their <tt class="docutils literal"><span class="pre">cleanuppad</span></tt> and <tt class="docutils literal"><span class="pre">catchswitch</span></tt> edges exiting their
+<tt class="docutils literal"><span class="pre">catchswitch</span></tt>) must share the same unwind destination. Similarly, any
+funclet pad which may be exited by unwind to caller must not be exited by
+any exception edges which unwind anywhere other than the caller. This
+ensures that each funclet as a whole has only one unwind destination, which
+EH tables for funclet personalities may require. Note that any unwind edge
+which exits a <tt class="docutils literal"><span class="pre">catchpad</span></tt> also exits its parent <tt class="docutils literal"><span class="pre">catchswitch</span></tt>, so this
+implies that for any given <tt class="docutils literal"><span class="pre">catchswitch</span></tt>, its unwind destination must also
+be the unwind destination of any unwind edge that exits any of its constituent
+<tt class="docutils literal"><span class="pre">catchpad</span></tt>s. Because <tt class="docutils literal"><span class="pre">catchswitch</span></tt> has no <tt class="docutils literal"><span class="pre">nounwind</span></tt> variant, and
+because IR producers are not <em>required</em> to annotate calls which will not
+unwind as <tt class="docutils literal"><span class="pre">nounwind</span></tt>, it is legal to nest a <tt class="docutils literal"><span class="pre">call</span></tt> or an “<tt class="docutils literal"><span class="pre">unwind</span> <span class="pre">to</span>
+<span class="pre">caller</span></tt>” <tt class="docutils literal"><span class="pre">catchswitch</span></tt> within a funclet pad that has an unwind
+destination other than caller; it is undefined behavior for such a <tt class="docutils literal"><span class="pre">call</span></tt>
+or <tt class="docutils literal"><span class="pre">catchswitch</span></tt> to unwind.</p>
+<p>Finally, the funclet pads’ unwind destinations cannot form a cycle. This
+ensures that EH lowering can construct “try regions” with a tree-like
+structure, which funclet-based personalities may require.</p>
+</div>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="LinkTimeOptimization.html" title="LLVM Link Time Optimization: Design and Implementation"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="CodeGenerator.html" title="The LLVM Target-Independent Code Generator"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/ExtendingLLVM.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/ExtendingLLVM.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/ExtendingLLVM.html (added)
+++ www-releases/trunk/4.0.0/docs/ExtendingLLVM.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,380 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>Extending LLVM: Adding instructions, intrinsics, types, etc. — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="How to set up LLVM-style RTTI for your class hierarchy" href="HowToSetUpLLVMStyleRTTI.html" />
+ <link rel="prev" title="Architecture & Platform Information for Compiler Writers" href="CompilerWriterInfo.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="HowToSetUpLLVMStyleRTTI.html" title="How to set up LLVM-style RTTI for your class hierarchy"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="CompilerWriterInfo.html" title="Architecture & Platform Information for Compiler Writers"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="extending-llvm-adding-instructions-intrinsics-types-etc">
+<h1>Extending LLVM: Adding instructions, intrinsics, types, etc.<a class="headerlink" href="#extending-llvm-adding-instructions-intrinsics-types-etc" title="Permalink to this headline">¶</a></h1>
+<div class="section" id="introduction-and-warning">
+<h2>Introduction and Warning<a class="headerlink" href="#introduction-and-warning" title="Permalink to this headline">¶</a></h2>
+<p>During the course of using LLVM, you may wish to customize it for your research
+project or for experimentation. At this point, you may realize that you need to
+add something to LLVM, whether it be a new fundamental type, a new intrinsic
+function, or a whole new instruction.</p>
+<p>When you come to this realization, stop and think. Do you really need to extend
+LLVM? Is it a new fundamental capability that LLVM does not support at its
+current incarnation or can it be synthesized from already pre-existing LLVM
+elements? If you are not sure, ask on the <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/llvm-dev">LLVM-dev</a> list. The reason is that
+extending LLVM will get involved as you need to update all the different passes
+that you intend to use with your extension, and there are <tt class="docutils literal"><span class="pre">many</span></tt> LLVM analyses
+and transformations, so it may be quite a bit of work.</p>
+<p>Adding an <a class="reference internal" href="#intrinsic-function">intrinsic function</a> is far easier than adding an
+instruction, and is transparent to optimization passes. If your added
+functionality can be expressed as a function call, an intrinsic function is the
+method of choice for LLVM extension.</p>
+<p>Before you invest a significant amount of effort into a non-trivial extension,
+<strong>ask on the list</strong> if what you are looking to do can be done with
+already-existing infrastructure, or if maybe someone else is already working on
+it. You will save yourself a lot of time and effort by doing so.</p>
+</div>
+<div class="section" id="adding-a-new-intrinsic-function">
+<span id="intrinsic-function"></span><h2>Adding a new intrinsic function<a class="headerlink" href="#adding-a-new-intrinsic-function" title="Permalink to this headline">¶</a></h2>
+<p>Adding a new intrinsic function to LLVM is much easier than adding a new
+instruction. Almost all extensions to LLVM should start as an intrinsic
+function and then be turned into an instruction if warranted.</p>
+<ol class="arabic">
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/docs/LangRef.html</span></tt>:</p>
+<p>Document the intrinsic. Decide whether it is code generator specific and
+what the restrictions are. Talk to other people about it so that you are
+sure it’s a good idea.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/IR/Intrinsics*.td</span></tt>:</p>
+<p>Add an entry for your intrinsic. Describe its memory access characteristics
+for optimization (this controls whether it will be DCE’d, CSE’d, etc). Note
+that any intrinsic using one of the <tt class="docutils literal"><span class="pre">llvm_any*_ty</span></tt> types for an argument or
+return type will be deemed by <tt class="docutils literal"><span class="pre">tblgen</span></tt> as overloaded and the corresponding
+suffix will be required on the intrinsic’s name.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Analysis/ConstantFolding.cpp</span></tt>:</p>
+<p>If it is possible to constant fold your intrinsic, add support to it in the
+<tt class="docutils literal"><span class="pre">canConstantFoldCallTo</span></tt> and <tt class="docutils literal"><span class="pre">ConstantFoldCall</span></tt> functions.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/test/*</span></tt>:</p>
+<p>Add test cases for your test cases to the test suite</p>
+</li>
+</ol>
+<p>Once the intrinsic has been added to the system, you must add code generator
+support for it. Generally you must do the following steps:</p>
+<p>Add support to the .td file for the target(s) of your choice in
+<tt class="docutils literal"><span class="pre">lib/Target/*/*.td</span></tt>.</p>
+<blockquote>
+<div>This is usually a matter of adding a pattern to the .td file that matches the
+intrinsic, though it may obviously require adding the instructions you want to
+generate as well. There are lots of examples in the PowerPC and X86 backend
+to follow.</div></blockquote>
+</div>
+<div class="section" id="adding-a-new-selectiondag-node">
+<h2>Adding a new SelectionDAG node<a class="headerlink" href="#adding-a-new-selectiondag-node" title="Permalink to this headline">¶</a></h2>
+<p>As with intrinsics, adding a new SelectionDAG node to LLVM is much easier than
+adding a new instruction. New nodes are often added to help represent
+instructions common to many targets. These nodes often map to an LLVM
+instruction (add, sub) or intrinsic (byteswap, population count). In other
+cases, new nodes have been added to allow many targets to perform a common task
+(converting between floating point and integer representation) or capture more
+complicated behavior in a single node (rotate).</p>
+<ol class="arabic">
+<li><p class="first"><tt class="docutils literal"><span class="pre">include/llvm/CodeGen/ISDOpcodes.h</span></tt>:</p>
+<p>Add an enum value for the new SelectionDAG node.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">lib/CodeGen/SelectionDAG/SelectionDAG.cpp</span></tt>:</p>
+<dl class="docutils">
+<dt>Add code to print the node to <tt class="docutils literal"><span class="pre">getOperationName</span></tt>. If your new node can be</dt>
+<dd><p class="first last">evaluated at compile time when given constant arguments (such as an add of a
+constant with another constant), find the <tt class="docutils literal"><span class="pre">getNode</span></tt> method that takes the
+appropriate number of arguments, and add a case for your node to the switch
+statement that performs constant folding for nodes that take the same number
+of arguments as your new node.</p>
+</dd>
+</dl>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">lib/CodeGen/SelectionDAG/LegalizeDAG.cpp</span></tt>:</p>
+<p>Add code to <a class="reference external" href="CodeGenerator.html#selectiondag_legalize">legalize, promote, and expand</a> the node as necessary. At a
+minimum, you will need to add a case statement for your node in
+<tt class="docutils literal"><span class="pre">LegalizeOp</span></tt> which calls LegalizeOp on the node’s operands, and returns a
+new node if any of the operands changed as a result of being legalized. It
+is likely that not all targets supported by the SelectionDAG framework will
+natively support the new node. In this case, you must also add code in your
+node’s case statement in <tt class="docutils literal"><span class="pre">LegalizeOp</span></tt> to Expand your node into simpler,
+legal operations. The case for <tt class="docutils literal"><span class="pre">ISD::UREM</span></tt> for expanding a remainder into
+a divide, multiply, and a subtract is a good example.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">lib/CodeGen/SelectionDAG/LegalizeDAG.cpp</span></tt>:</p>
+<dl class="docutils">
+<dt>If targets may support the new node being added only at certain sizes, you</dt>
+<dd><p class="first last">will also need to add code to your node’s case statement in <tt class="docutils literal"><span class="pre">LegalizeOp</span></tt>
+to Promote your node’s operands to a larger size, and perform the correct
+operation. You will also need to add code to <tt class="docutils literal"><span class="pre">PromoteOp</span></tt> to do this as
+well. For a good example, see <tt class="docutils literal"><span class="pre">ISD::BSWAP</span></tt>, which promotes its operand to
+a wider size, performs the byteswap, and then shifts the correct bytes right
+to emulate the narrower byteswap in the wider type.</p>
+</dd>
+</dl>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">lib/CodeGen/SelectionDAG/LegalizeDAG.cpp</span></tt>:</p>
+<p>Add a case for your node in <tt class="docutils literal"><span class="pre">ExpandOp</span></tt> to teach the legalizer how to
+perform the action represented by the new node on a value that has been split
+into high and low halves. This case will be used to support your node with a
+64 bit operand on a 32 bit target.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">lib/CodeGen/SelectionDAG/DAGCombiner.cpp</span></tt>:</p>
+<p>If your node can be combined with itself, or other existing nodes in a
+peephole-like fashion, add a visit function for it, and call that function
+from. There are several good examples for simple combines you can do;
+<tt class="docutils literal"><span class="pre">visitFABS</span></tt> and <tt class="docutils literal"><span class="pre">visitSRL</span></tt> are good starting places.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">lib/Target/PowerPC/PPCISelLowering.cpp</span></tt>:</p>
+<p>Each target has an implementation of the <tt class="docutils literal"><span class="pre">TargetLowering</span></tt> class, usually in
+its own file (although some targets include it in the same file as the
+DAGToDAGISel). The default behavior for a target is to assume that your new
+node is legal for all types that are legal for that target. If this target
+does not natively support your node, then tell the target to either Promote
+it (if it is supported at a larger type) or Expand it. This will cause the
+code you wrote in <tt class="docutils literal"><span class="pre">LegalizeOp</span></tt> above to decompose your new node into other
+legal nodes for this target.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">lib/Target/TargetSelectionDAG.td</span></tt>:</p>
+<p>Most current targets supported by LLVM generate code using the DAGToDAG
+method, where SelectionDAG nodes are pattern matched to target-specific
+nodes, which represent individual instructions. In order for the targets to
+match an instruction to your new node, you must add a def for that node to
+the list in this file, with the appropriate type constraints. Look at
+<tt class="docutils literal"><span class="pre">add</span></tt>, <tt class="docutils literal"><span class="pre">bswap</span></tt>, and <tt class="docutils literal"><span class="pre">fadd</span></tt> for examples.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">lib/Target/PowerPC/PPCInstrInfo.td</span></tt>:</p>
+<p>Each target has a tablegen file that describes the target’s instruction set.
+For targets that use the DAGToDAG instruction selection framework, add a
+pattern for your new node that uses one or more target nodes. Documentation
+for this is a bit sparse right now, but there are several decent examples.
+See the patterns for <tt class="docutils literal"><span class="pre">rotl</span></tt> in <tt class="docutils literal"><span class="pre">PPCInstrInfo.td</span></tt>.</p>
+</li>
+<li><p class="first">TODO: document complex patterns.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/test/CodeGen/*</span></tt>:</p>
+<p>Add test cases for your new node to the test suite.
+<tt class="docutils literal"><span class="pre">llvm/test/CodeGen/X86/bswap.ll</span></tt> is a good example.</p>
+</li>
+</ol>
+</div>
+<div class="section" id="adding-a-new-instruction">
+<h2>Adding a new instruction<a class="headerlink" href="#adding-a-new-instruction" title="Permalink to this headline">¶</a></h2>
+<div class="admonition warning">
+<p class="first admonition-title">Warning</p>
+<p class="last">Adding instructions changes the bitcode format, and it will take some effort
+to maintain compatibility with the previous version. Only add an instruction
+if it is absolutely necessary.</p>
+</div>
+<ol class="arabic">
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/IR/Instruction.def</span></tt>:</p>
+<p>add a number for your instruction and an enum name</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/IR/Instructions.h</span></tt>:</p>
+<p>add a definition for the class that will represent your instruction</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/IR/InstVisitor.h</span></tt>:</p>
+<p>add a prototype for a visitor to your new instruction type</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/AsmParser/LLLexer.cpp</span></tt>:</p>
+<p>add a new token to parse your instruction from assembly text file</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/AsmParser/LLParser.cpp</span></tt>:</p>
+<p>add the grammar on how your instruction can be read and what it will
+construct as a result</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Bitcode/Reader/BitcodeReader.cpp</span></tt>:</p>
+<p>add a case for your instruction and how it will be parsed from bitcode</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Bitcode/Writer/BitcodeWriter.cpp</span></tt>:</p>
+<p>add a case for your instruction and how it will be parsed from bitcode</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/IR/Instruction.cpp</span></tt>:</p>
+<p>add a case for how your instruction will be printed out to assembly</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/IR/Instructions.cpp</span></tt>:</p>
+<p>implement the class you defined in <tt class="docutils literal"><span class="pre">llvm/include/llvm/Instructions.h</span></tt></p>
+</li>
+<li><p class="first">Test your instruction</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Target/*</span></tt>:</p>
+<p>add support for your instruction to code generators, or add a lowering pass.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/test/*</span></tt>:</p>
+<p>add your test cases to the test suite.</p>
+</li>
+</ol>
+<p>Also, you need to implement (or modify) any analyses or passes that you want to
+understand this new instruction.</p>
+</div>
+<div class="section" id="adding-a-new-type">
+<h2>Adding a new type<a class="headerlink" href="#adding-a-new-type" title="Permalink to this headline">¶</a></h2>
+<div class="admonition warning">
+<p class="first admonition-title">Warning</p>
+<p class="last">Adding new types changes the bitcode format, and will break compatibility with
+currently-existing LLVM installations. Only add new types if it is absolutely
+necessary.</p>
+</div>
+<div class="section" id="adding-a-fundamental-type">
+<h3>Adding a fundamental type<a class="headerlink" href="#adding-a-fundamental-type" title="Permalink to this headline">¶</a></h3>
+<ol class="arabic">
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/IR/Type.h</span></tt>:</p>
+<p>add enum for the new type; add static <tt class="docutils literal"><span class="pre">Type*</span></tt> for this type</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/IR/Type.cpp</span></tt> and <tt class="docutils literal"><span class="pre">llvm/lib/IR/ValueTypes.cpp</span></tt>:</p>
+<p>add mapping from <tt class="docutils literal"><span class="pre">TypeID</span></tt> => <tt class="docutils literal"><span class="pre">Type*</span></tt>; initialize the static <tt class="docutils literal"><span class="pre">Type*</span></tt></p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/llvm/llvm-c/Core.cpp</span></tt>:</p>
+<p>add enum <tt class="docutils literal"><span class="pre">LLVMTypeKind</span></tt> and modify
+<tt class="docutils literal"><span class="pre">LLVMTypeKind</span> <span class="pre">LLVMGetTypeKind(LLVMTypeRef</span> <span class="pre">Ty)</span></tt> for the new type</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/IR/TypeBuilder.h</span></tt>:</p>
+<p>add new class to represent new type in the hierarchy</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/AsmParser/LLLexer.cpp</span></tt>:</p>
+<p>add ability to parse in the type from text assembly</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/AsmParser/LLParser.cpp</span></tt>:</p>
+<p>add a token for that type</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Bitcode/Writer/BitcodeWriter.cpp</span></tt>:</p>
+<p>modify <tt class="docutils literal"><span class="pre">static</span> <span class="pre">void</span> <span class="pre">WriteTypeTable(const</span> <span class="pre">ValueEnumerator</span> <span class="pre">&VE,</span>
+<span class="pre">BitstreamWriter</span> <span class="pre">&Stream)</span></tt> to serialize your type</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Bitcode/Reader/BitcodeReader.cpp</span></tt>:</p>
+<p>modify <tt class="docutils literal"><span class="pre">bool</span> <span class="pre">BitcodeReader::ParseTypeType()</span></tt> to read your data type</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">include/llvm/Bitcode/LLVMBitCodes.h</span></tt>:</p>
+<p>add enum <tt class="docutils literal"><span class="pre">TypeCodes</span></tt> for the new type</p>
+</li>
+</ol>
+</div>
+<div class="section" id="adding-a-derived-type">
+<h3>Adding a derived type<a class="headerlink" href="#adding-a-derived-type" title="Permalink to this headline">¶</a></h3>
+<ol class="arabic">
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/IR/Type.h</span></tt>:</p>
+<p>add enum for the new type; add a forward declaration of the type also</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/IR/DerivedTypes.h</span></tt>:</p>
+<p>add new class to represent new class in the hierarchy; add forward
+declaration to the TypeMap value type</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/IR/Type.cpp</span></tt> and <tt class="docutils literal"><span class="pre">llvm/lib/IR/ValueTypes.cpp</span></tt>:</p>
+<p>add support for derived type, notably <cite>enum TypeID</cite> and <cite>is</cite>, <cite>get</cite> methods.</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/llvm/llvm-c/Core.cpp</span></tt>:</p>
+<p>add enum <tt class="docutils literal"><span class="pre">LLVMTypeKind</span></tt> and modify
+<cite>LLVMTypeKind LLVMGetTypeKind(LLVMTypeRef Ty)</cite> for the new type</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/IR/TypeBuilder.h</span></tt>:</p>
+<p>add new class to represent new class in the hierarchy</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/AsmParser/LLLexer.cpp</span></tt>:</p>
+<p>modify <tt class="docutils literal"><span class="pre">lltok::Kind</span> <span class="pre">LLLexer::LexIdentifier()</span></tt> to add ability to
+parse in the type from text assembly</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Bitcode/Writer/BitcodeWriter.cpp</span></tt>:</p>
+<p>modify <tt class="docutils literal"><span class="pre">static</span> <span class="pre">void</span> <span class="pre">WriteTypeTable(const</span> <span class="pre">ValueEnumerator</span> <span class="pre">&VE,</span>
+<span class="pre">BitstreamWriter</span> <span class="pre">&Stream)</span></tt> to serialize your type</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Bitcode/Reader/BitcodeReader.cpp</span></tt>:</p>
+<p>modify <tt class="docutils literal"><span class="pre">bool</span> <span class="pre">BitcodeReader::ParseTypeType()</span></tt> to read your data type</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">include/llvm/Bitcode/LLVMBitCodes.h</span></tt>:</p>
+<p>add enum <tt class="docutils literal"><span class="pre">TypeCodes</span></tt> for the new type</p>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/IR/AsmWriter.cpp</span></tt>:</p>
+<p>modify <tt class="docutils literal"><span class="pre">void</span> <span class="pre">TypePrinting::print(Type</span> <span class="pre">*Ty,</span> <span class="pre">raw_ostream</span> <span class="pre">&OS)</span></tt>
+to output the new derived type</p>
+</li>
+</ol>
+</div>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="HowToSetUpLLVMStyleRTTI.html" title="How to set up LLVM-style RTTI for your class hierarchy"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="CompilerWriterInfo.html" title="Architecture & Platform Information for Compiler Writers"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/Extensions.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/Extensions.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/Extensions.html (added)
+++ www-releases/trunk/4.0.0/docs/Extensions.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,335 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>LLVM Extensions — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="libFuzzer â a library for coverage-guided fuzz testing." href="LibFuzzer.html" />
+ <link rel="prev" title="LLVM Programmerâs Manual" href="ProgrammersManual.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="LibFuzzer.html" title="libFuzzer â a library for coverage-guided fuzz testing."
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="ProgrammersManual.html" title="LLVM Programmerâs Manual"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="llvm-extensions">
+<h1>LLVM Extensions<a class="headerlink" href="#llvm-extensions" title="Permalink to this headline">¶</a></h1>
+<div class="contents local topic" id="contents">
+<ul class="simple">
+<li><a class="reference internal" href="#introduction" id="id2">Introduction</a></li>
+<li><a class="reference internal" href="#general-assembly-syntax" id="id3">General Assembly Syntax</a><ul>
+<li><a class="reference internal" href="#c99-style-hexadecimal-floating-point-constants" id="id4">C99-style Hexadecimal Floating-point Constants</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#machine-specific-assembly-syntax" id="id5">Machine-specific Assembly Syntax</a><ul>
+<li><a class="reference internal" href="#x86-coff-dependent" id="id6">X86/COFF-Dependent</a><ul>
+<li><a class="reference internal" href="#relocations" id="id7">Relocations</a></li>
+<li><a class="reference internal" href="#linkonce-directive" id="id8"><tt class="docutils literal"><span class="pre">.linkonce</span></tt> Directive</a></li>
+<li><a class="reference internal" href="#section-directive" id="id9"><tt class="docutils literal"><span class="pre">.section</span></tt> Directive</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#elf-dependent" id="id10">ELF-Dependent</a><ul>
+<li><a class="reference internal" href="#id1" id="id11"><tt class="docutils literal"><span class="pre">.section</span></tt> Directive</a></li>
+</ul>
+</li>
+</ul>
+</li>
+<li><a class="reference internal" href="#target-specific-behaviour" id="id12">Target Specific Behaviour</a><ul>
+<li><a class="reference internal" href="#windows-on-arm" id="id13">Windows on ARM</a><ul>
+<li><a class="reference internal" href="#stack-probe-emission" id="id14">Stack Probe Emission</a></li>
+<li><a class="reference internal" href="#variable-length-arrays" id="id15">Variable Length Arrays</a></li>
+</ul>
+</li>
+</ul>
+</li>
+</ul>
+</div>
+<div class="toctree-wrapper compound">
+</div>
+<div class="section" id="introduction">
+<h2><a class="toc-backref" href="#id2">Introduction</a><a class="headerlink" href="#introduction" title="Permalink to this headline">¶</a></h2>
+<p>This document describes extensions to tools and formats LLVM seeks compatibility
+with.</p>
+</div>
+<div class="section" id="general-assembly-syntax">
+<h2><a class="toc-backref" href="#id3">General Assembly Syntax</a><a class="headerlink" href="#general-assembly-syntax" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="c99-style-hexadecimal-floating-point-constants">
+<h3><a class="toc-backref" href="#id4">C99-style Hexadecimal Floating-point Constants</a><a class="headerlink" href="#c99-style-hexadecimal-floating-point-constants" title="Permalink to this headline">¶</a></h3>
+<p>LLVM’s assemblers allow floating-point constants to be written in C99’s
+hexadecimal format instead of decimal if desired.</p>
+<div class="highlight-gas"><div class="highlight"><pre><span class="na">.section</span> <span class="no">.data</span>
+<span class="na">.float</span> <span class="mi">0x1c2</span><span class="no">.2ap3</span>
+</pre></div>
+</div>
+</div>
+</div>
+<div class="section" id="machine-specific-assembly-syntax">
+<h2><a class="toc-backref" href="#id5">Machine-specific Assembly Syntax</a><a class="headerlink" href="#machine-specific-assembly-syntax" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="x86-coff-dependent">
+<h3><a class="toc-backref" href="#id6">X86/COFF-Dependent</a><a class="headerlink" href="#x86-coff-dependent" title="Permalink to this headline">¶</a></h3>
+<div class="section" id="relocations">
+<h4><a class="toc-backref" href="#id7">Relocations</a><a class="headerlink" href="#relocations" title="Permalink to this headline">¶</a></h4>
+<p>The following additional relocation types are supported:</p>
+<p><strong>@IMGREL</strong> (AT&T syntax only) generates an image-relative relocation that
+corresponds to the COFF relocation types <tt class="docutils literal"><span class="pre">IMAGE_REL_I386_DIR32NB</span></tt> (32-bit) or
+<tt class="docutils literal"><span class="pre">IMAGE_REL_AMD64_ADDR32NB</span></tt> (64-bit).</p>
+<div class="highlight-text"><div class="highlight"><pre>.text
+fun:
+ mov foo at IMGREL(%ebx, %ecx, 4), %eax
+
+.section .pdata
+ .long fun at IMGREL
+ .long (fun at imgrel + 0x3F)
+ .long $unwind$fun at imgrel
+</pre></div>
+</div>
+<p><strong>.secrel32</strong> generates a relocation that corresponds to the COFF relocation
+types <tt class="docutils literal"><span class="pre">IMAGE_REL_I386_SECREL</span></tt> (32-bit) or <tt class="docutils literal"><span class="pre">IMAGE_REL_AMD64_SECREL</span></tt> (64-bit).</p>
+<p><strong>.secidx</strong> relocation generates an index of the section that contains
+the target. It corresponds to the COFF relocation types
+<tt class="docutils literal"><span class="pre">IMAGE_REL_I386_SECTION</span></tt> (32-bit) or <tt class="docutils literal"><span class="pre">IMAGE_REL_AMD64_SECTION</span></tt> (64-bit).</p>
+<div class="highlight-none"><div class="highlight"><pre>.section .debug$S,"rn"
+ .long 4
+ .long 242
+ .long 40
+ .secrel32 _function_name + 0
+ .secidx _function_name
+ ...
+</pre></div>
+</div>
+</div>
+<div class="section" id="linkonce-directive">
+<h4><a class="toc-backref" href="#id8"><tt class="docutils literal"><span class="pre">.linkonce</span></tt> Directive</a><a class="headerlink" href="#linkonce-directive" title="Permalink to this headline">¶</a></h4>
+<p>Syntax:</p>
+<blockquote>
+<div><tt class="docutils literal"><span class="pre">.linkonce</span> <span class="pre">[</span> <span class="pre">comdat</span> <span class="pre">type</span> <span class="pre">]</span></tt></div></blockquote>
+<p>Supported COMDAT types:</p>
+<dl class="docutils">
+<dt><tt class="docutils literal"><span class="pre">discard</span></tt></dt>
+<dd>Discards duplicate sections with the same COMDAT symbol. This is the default
+if no type is specified.</dd>
+<dt><tt class="docutils literal"><span class="pre">one_only</span></tt></dt>
+<dd>If the symbol is defined multiple times, the linker issues an error.</dd>
+<dt><tt class="docutils literal"><span class="pre">same_size</span></tt></dt>
+<dd>Duplicates are discarded, but the linker issues an error if any have
+different sizes.</dd>
+<dt><tt class="docutils literal"><span class="pre">same_contents</span></tt></dt>
+<dd>Duplicates are discarded, but the linker issues an error if any duplicates
+do not have exactly the same content.</dd>
+<dt><tt class="docutils literal"><span class="pre">largest</span></tt></dt>
+<dd>Links the largest section from among the duplicates.</dd>
+<dt><tt class="docutils literal"><span class="pre">newest</span></tt></dt>
+<dd>Links the newest section from among the duplicates.</dd>
+</dl>
+<div class="highlight-gas"><div class="highlight"><pre><span class="na">.section</span> <span class="no">.text$foo</span>
+<span class="na">.linkonce</span>
+ <span class="na">...</span>
+</pre></div>
+</div>
+</div>
+<div class="section" id="section-directive">
+<h4><a class="toc-backref" href="#id9"><tt class="docutils literal"><span class="pre">.section</span></tt> Directive</a><a class="headerlink" href="#section-directive" title="Permalink to this headline">¶</a></h4>
+<p>MC supports passing the information in <tt class="docutils literal"><span class="pre">.linkonce</span></tt> at the end of
+<tt class="docutils literal"><span class="pre">.section</span></tt>. For example, these two codes are equivalent</p>
+<div class="highlight-gas"><div class="highlight"><pre><span class="na">.section</span> <span class="no">secName</span><span class="p">,</span> <span class="s">"dr"</span><span class="p">,</span> <span class="no">discard</span><span class="p">,</span> <span class="s">"Symbol1"</span>
+<span class="na">.globl</span> <span class="no">Symbol1</span>
+<span class="nl">Symbol1:</span>
+<span class="na">.long</span> <span class="mi">1</span>
+</pre></div>
+</div>
+<div class="highlight-gas"><div class="highlight"><pre><span class="na">.section</span> <span class="no">secName</span><span class="p">,</span> <span class="s">"dr"</span>
+<span class="na">.linkonce</span> <span class="no">discard</span>
+<span class="na">.globl</span> <span class="no">Symbol1</span>
+<span class="nl">Symbol1:</span>
+<span class="na">.long</span> <span class="mi">1</span>
+</pre></div>
+</div>
+<p>Note that in the combined form the COMDAT symbol is explicit. This
+extension exists to support multiple sections with the same name in
+different COMDATs:</p>
+<div class="highlight-gas"><div class="highlight"><pre><span class="na">.section</span> <span class="no">secName</span><span class="p">,</span> <span class="s">"dr"</span><span class="p">,</span> <span class="no">discard</span><span class="p">,</span> <span class="s">"Symbol1"</span>
+<span class="na">.globl</span> <span class="no">Symbol1</span>
+<span class="nl">Symbol1:</span>
+<span class="na">.long</span> <span class="mi">1</span>
+
+<span class="na">.section</span> <span class="no">secName</span><span class="p">,</span> <span class="s">"dr"</span><span class="p">,</span> <span class="no">discard</span><span class="p">,</span> <span class="s">"Symbol2"</span>
+<span class="na">.globl</span> <span class="no">Symbol2</span>
+<span class="nl">Symbol2:</span>
+<span class="na">.long</span> <span class="mi">1</span>
+</pre></div>
+</div>
+<p>In addition to the types allowed with <tt class="docutils literal"><span class="pre">.linkonce</span></tt>, <tt class="docutils literal"><span class="pre">.section</span></tt> also accepts
+<tt class="docutils literal"><span class="pre">associative</span></tt>. The meaning is that the section is linked if a certain other
+COMDAT section is linked. This other section is indicated by the comdat symbol
+in this directive. It can be any symbol defined in the associated section, but
+is usually the associated section’s comdat.</p>
+<blockquote>
+<div><p>The following restrictions apply to the associated section:</p>
+<ol class="arabic simple">
+<li>It must be a COMDAT section.</li>
+<li>It cannot be another associative COMDAT section.</li>
+</ol>
+</div></blockquote>
+<p>In the following example the symobl <tt class="docutils literal"><span class="pre">sym</span></tt> is the comdat symbol of <tt class="docutils literal"><span class="pre">.foo</span></tt>
+and <tt class="docutils literal"><span class="pre">.bar</span></tt> is associated to <tt class="docutils literal"><span class="pre">.foo</span></tt>.</p>
+<div class="highlight-gas"><div class="highlight"><pre><span class="na">.section</span> <span class="no">.foo</span><span class="p">,</span><span class="s">"bw"</span><span class="p">,</span><span class="no">discard</span><span class="p">,</span> <span class="s">"sym"</span>
+<span class="na">.section</span> <span class="no">.bar</span><span class="p">,</span><span class="s">"rd"</span><span class="p">,</span><span class="no">associative</span><span class="p">,</span> <span class="s">"sym"</span>
+</pre></div>
+</div>
+<p>MC supports these flags in the COFF <tt class="docutils literal"><span class="pre">.section</span></tt> directive:</p>
+<blockquote>
+<div><ul class="simple">
+<li><tt class="docutils literal"><span class="pre">b</span></tt>: BSS section (<tt class="docutils literal"><span class="pre">IMAGE_SCN_CNT_INITIALIZED_DATA</span></tt>)</li>
+<li><tt class="docutils literal"><span class="pre">d</span></tt>: Data section (<tt class="docutils literal"><span class="pre">IMAGE_SCN_CNT_UNINITIALIZED_DATA</span></tt>)</li>
+<li><tt class="docutils literal"><span class="pre">n</span></tt>: Section is not loaded (<tt class="docutils literal"><span class="pre">IMAGE_SCN_LNK_REMOVE</span></tt>)</li>
+<li><tt class="docutils literal"><span class="pre">r</span></tt>: Read-only</li>
+<li><tt class="docutils literal"><span class="pre">s</span></tt>: Shared section</li>
+<li><tt class="docutils literal"><span class="pre">w</span></tt>: Writable</li>
+<li><tt class="docutils literal"><span class="pre">x</span></tt>: Executable section</li>
+<li><tt class="docutils literal"><span class="pre">y</span></tt>: Not readable</li>
+<li><tt class="docutils literal"><span class="pre">D</span></tt>: Discardable (<tt class="docutils literal"><span class="pre">IMAGE_SCN_MEM_DISCARDABLE</span></tt>)</li>
+</ul>
+</div></blockquote>
+<p>These flags are all compatible with gas, with the exception of the <tt class="docutils literal"><span class="pre">D</span></tt> flag,
+which gnu as does not support. For gas compatibility, sections with a name
+starting with ”.debug” are implicitly discardable.</p>
+</div>
+</div>
+<div class="section" id="elf-dependent">
+<h3><a class="toc-backref" href="#id10">ELF-Dependent</a><a class="headerlink" href="#elf-dependent" title="Permalink to this headline">¶</a></h3>
+<div class="section" id="id1">
+<h4><a class="toc-backref" href="#id11"><tt class="docutils literal"><span class="pre">.section</span></tt> Directive</a><a class="headerlink" href="#id1" title="Permalink to this headline">¶</a></h4>
+<p>In order to support creating multiple sections with the same name and comdat,
+it is possible to add an unique number at the end of the <tt class="docutils literal"><span class="pre">.seciton</span></tt> directive.
+For example, the following code creates two sections named <tt class="docutils literal"><span class="pre">.text</span></tt>.</p>
+<div class="highlight-gas"><div class="highlight"><pre><span class="na">.section</span> <span class="no">.text</span><span class="p">,</span><span class="s">"ax"</span><span class="p">,</span><span class="na">@progbits</span><span class="p">,</span><span class="no">unique</span><span class="p">,</span><span class="mi">1</span>
+<span class="nf">nop</span>
+
+<span class="na">.section</span> <span class="no">.text</span><span class="p">,</span><span class="s">"ax"</span><span class="p">,</span><span class="na">@progbits</span><span class="p">,</span><span class="no">unique</span><span class="p">,</span><span class="mi">2</span>
+<span class="nf">nop</span>
+</pre></div>
+</div>
+<p>The unique number is not present in the resulting object at all. It is just used
+in the assembler to differentiate the sections.</p>
+</div>
+</div>
+</div>
+<div class="section" id="target-specific-behaviour">
+<h2><a class="toc-backref" href="#id12">Target Specific Behaviour</a><a class="headerlink" href="#target-specific-behaviour" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="windows-on-arm">
+<h3><a class="toc-backref" href="#id13">Windows on ARM</a><a class="headerlink" href="#windows-on-arm" title="Permalink to this headline">¶</a></h3>
+<div class="section" id="stack-probe-emission">
+<h4><a class="toc-backref" href="#id14">Stack Probe Emission</a><a class="headerlink" href="#stack-probe-emission" title="Permalink to this headline">¶</a></h4>
+<p>The reference implementation (Microsoft Visual Studio 2012) emits stack probes
+in the following fashion:</p>
+<div class="highlight-gas"><div class="highlight"><pre><span class="nf">movw</span> <span class="no">r4</span><span class="p">,</span> <span class="c">#constant</span>
+<span class="nf">bl</span> <span class="no">__chkstk</span>
+<span class="nf">sub.w</span> <span class="no">sp</span><span class="p">,</span> <span class="no">sp</span><span class="p">,</span> <span class="no">r4</span>
+</pre></div>
+</div>
+<p>However, this has the limitation of 32 MiB (±16MiB). In order to accommodate
+larger binaries, LLVM supports the use of <tt class="docutils literal"><span class="pre">-mcode-model=large</span></tt> to allow a 4GiB
+range via a slight deviation. It will generate an indirect jump as follows:</p>
+<div class="highlight-gas"><div class="highlight"><pre><span class="nf">movw</span> <span class="no">r4</span><span class="p">,</span> <span class="c">#constant</span>
+<span class="nf">movw</span> <span class="no">r12</span><span class="p">,</span> <span class="p">:</span><span class="no">lower16</span><span class="p">:</span><span class="no">__chkstk</span>
+<span class="nf">movt</span> <span class="no">r12</span><span class="p">,</span> <span class="p">:</span><span class="no">upper16</span><span class="p">:</span><span class="no">__chkstk</span>
+<span class="nf">blx</span> <span class="no">r12</span>
+<span class="nf">sub.w</span> <span class="no">sp</span><span class="p">,</span> <span class="no">sp</span><span class="p">,</span> <span class="no">r4</span>
+</pre></div>
+</div>
+</div>
+<div class="section" id="variable-length-arrays">
+<h4><a class="toc-backref" href="#id15">Variable Length Arrays</a><a class="headerlink" href="#variable-length-arrays" title="Permalink to this headline">¶</a></h4>
+<p>The reference implementation (Microsoft Visual Studio 2012) does not permit the
+emission of Variable Length Arrays (VLAs).</p>
+<p>The Windows ARM Itanium ABI extends the base ABI by adding support for emitting
+a dynamic stack allocation. When emitting a variable stack allocation, a call
+to <tt class="docutils literal"><span class="pre">__chkstk</span></tt> is emitted unconditionally to ensure that guard pages are setup
+properly. The emission of this stack probe emission is handled similar to the
+standard stack probe emission.</p>
+<p>The MSVC environment does not emit code for VLAs currently.</p>
+</div>
+</div>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="LibFuzzer.html" title="libFuzzer â a library for coverage-guided fuzz testing."
+ >next</a> |</li>
+ <li class="right" >
+ <a href="ProgrammersManual.html" title="LLVM Programmerâs Manual"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/FAQ.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/FAQ.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/FAQ.html (added)
+++ www-releases/trunk/4.0.0/docs/FAQ.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,436 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>Frequently Asked Questions (FAQ) — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="The LLVM Lexicon" href="Lexicon.html" />
+ <link rel="prev" title="Getting Started with the LLVM System using Microsoft Visual Studio" href="GettingStartedVS.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="Lexicon.html" title="The LLVM Lexicon"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="GettingStartedVS.html" title="Getting Started with the LLVM System using Microsoft Visual Studio"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="frequently-asked-questions-faq">
+<h1>Frequently Asked Questions (FAQ)<a class="headerlink" href="#frequently-asked-questions-faq" title="Permalink to this headline">¶</a></h1>
+<div class="contents local topic" id="contents">
+<ul class="simple">
+<li><a class="reference internal" href="#license" id="id1">License</a><ul>
+<li><a class="reference internal" href="#does-the-university-of-illinois-open-source-license-really-qualify-as-an-open-source-license" id="id2">Does the University of Illinois Open Source License really qualify as an “open source” license?</a></li>
+<li><a class="reference internal" href="#can-i-modify-llvm-source-code-and-redistribute-the-modified-source" id="id3">Can I modify LLVM source code and redistribute the modified source?</a></li>
+<li><a class="reference internal" href="#can-i-modify-the-llvm-source-code-and-redistribute-binaries-or-other-tools-based-on-it-without-redistributing-the-source" id="id4">Can I modify the LLVM source code and redistribute binaries or other tools based on it, without redistributing the source?</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#source-code" id="id5">Source Code</a><ul>
+<li><a class="reference internal" href="#in-what-language-is-llvm-written" id="id6">In what language is LLVM written?</a></li>
+<li><a class="reference internal" href="#how-portable-is-the-llvm-source-code" id="id7">How portable is the LLVM source code?</a></li>
+<li><a class="reference internal" href="#what-api-do-i-use-to-store-a-value-to-one-of-the-virtual-registers-in-llvm-ir-s-ssa-representation" id="id8">What API do I use to store a value to one of the virtual registers in LLVM IR’s SSA representation?</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#source-languages" id="id9">Source Languages</a><ul>
+<li><a class="reference internal" href="#what-source-languages-are-supported" id="id10">What source languages are supported?</a></li>
+<li><a class="reference internal" href="#i-d-like-to-write-a-self-hosting-llvm-compiler-how-should-i-interface-with-the-llvm-middle-end-optimizers-and-back-end-code-generators" id="id11">I’d like to write a self-hosting LLVM compiler. How should I interface with the LLVM middle-end optimizers and back-end code generators?</a></li>
+<li><a class="reference internal" href="#what-support-is-there-for-a-higher-level-source-language-constructs-for-building-a-compiler" id="id12">What support is there for a higher level source language constructs for building a compiler?</a></li>
+<li><a class="reference internal" href="#i-don-t-understand-the-getelementptr-instruction-help" id="id13">I don’t understand the <tt class="docutils literal"><span class="pre">GetElementPtr</span></tt> instruction. Help!</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#using-the-c-and-c-front-ends" id="id14">Using the C and C++ Front Ends</a><ul>
+<li><a class="reference internal" href="#can-i-compile-c-or-c-code-to-platform-independent-llvm-bitcode" id="id15">Can I compile C or C++ code to platform-independent LLVM bitcode?</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#questions-about-code-generated-by-the-demo-page" id="id16">Questions about code generated by the demo page</a><ul>
+<li><a class="reference internal" href="#what-is-this-llvm-global-ctors-and-global-i-a-stuff-that-happens-when-i-include-iostream" id="id17">What is this <tt class="docutils literal"><span class="pre">llvm.global_ctors</span></tt> and <tt class="docutils literal"><span class="pre">_GLOBAL__I_a...</span></tt> stuff that happens when I <tt class="docutils literal"><span class="pre">#include</span> <span class="pre"><iostream></span></tt>?</a></li>
+<li><a class="reference internal" href="#where-did-all-of-my-code-go" id="id18">Where did all of my code go??</a></li>
+<li><a class="reference internal" href="#what-is-this-undef-thing-that-shows-up-in-my-code" id="id19">What is this “<tt class="docutils literal"><span class="pre">undef</span></tt>” thing that shows up in my code?</a></li>
+<li><a class="reference internal" href="#why-does-instcombine-simplifycfg-turn-a-call-to-a-function-with-a-mismatched-calling-convention-into-unreachable-why-not-make-the-verifier-reject-it" id="id20">Why does instcombine + simplifycfg turn a call to a function with a mismatched calling convention into “unreachable”? Why not make the verifier reject it?</a></li>
+</ul>
+</li>
+</ul>
+</div>
+<div class="section" id="license">
+<h2><a class="toc-backref" href="#id1">License</a><a class="headerlink" href="#license" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="does-the-university-of-illinois-open-source-license-really-qualify-as-an-open-source-license">
+<h3><a class="toc-backref" href="#id2">Does the University of Illinois Open Source License really qualify as an “open source” license?</a><a class="headerlink" href="#does-the-university-of-illinois-open-source-license-really-qualify-as-an-open-source-license" title="Permalink to this headline">¶</a></h3>
+<p>Yes, the license is <a class="reference external" href="http://www.opensource.org/licenses/UoI-NCSA.php">certified</a> by the Open Source
+Initiative (OSI).</p>
+</div>
+<div class="section" id="can-i-modify-llvm-source-code-and-redistribute-the-modified-source">
+<h3><a class="toc-backref" href="#id3">Can I modify LLVM source code and redistribute the modified source?</a><a class="headerlink" href="#can-i-modify-llvm-source-code-and-redistribute-the-modified-source" title="Permalink to this headline">¶</a></h3>
+<p>Yes. The modified source distribution must retain the copyright notice and
+follow the three bulleted conditions listed in the <a class="reference external" href="http://llvm.org/svn/llvm-project/llvm/trunk/LICENSE.TXT">LLVM license</a>.</p>
+</div>
+<div class="section" id="can-i-modify-the-llvm-source-code-and-redistribute-binaries-or-other-tools-based-on-it-without-redistributing-the-source">
+<h3><a class="toc-backref" href="#id4">Can I modify the LLVM source code and redistribute binaries or other tools based on it, without redistributing the source?</a><a class="headerlink" href="#can-i-modify-the-llvm-source-code-and-redistribute-binaries-or-other-tools-based-on-it-without-redistributing-the-source" title="Permalink to this headline">¶</a></h3>
+<p>Yes. This is why we distribute LLVM under a less restrictive license than GPL,
+as explained in the first question above.</p>
+</div>
+</div>
+<div class="section" id="source-code">
+<h2><a class="toc-backref" href="#id5">Source Code</a><a class="headerlink" href="#source-code" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="in-what-language-is-llvm-written">
+<h3><a class="toc-backref" href="#id6">In what language is LLVM written?</a><a class="headerlink" href="#in-what-language-is-llvm-written" title="Permalink to this headline">¶</a></h3>
+<p>All of the LLVM tools and libraries are written in C++ with extensive use of
+the STL.</p>
+</div>
+<div class="section" id="how-portable-is-the-llvm-source-code">
+<h3><a class="toc-backref" href="#id7">How portable is the LLVM source code?</a><a class="headerlink" href="#how-portable-is-the-llvm-source-code" title="Permalink to this headline">¶</a></h3>
+<p>The LLVM source code should be portable to most modern Unix-like operating
+systems. Most of the code is written in standard C++ with operating system
+services abstracted to a support library. The tools required to build and
+test LLVM have been ported to a plethora of platforms.</p>
+<p>Some porting problems may exist in the following areas:</p>
+<ul class="simple">
+<li>The autoconf/makefile build system relies heavily on UNIX shell tools,
+like the Bourne Shell and sed. Porting to systems without these tools
+(MacOS 9, Plan 9) will require more effort.</li>
+</ul>
+</div>
+<div class="section" id="what-api-do-i-use-to-store-a-value-to-one-of-the-virtual-registers-in-llvm-ir-s-ssa-representation">
+<h3><a class="toc-backref" href="#id8">What API do I use to store a value to one of the virtual registers in LLVM IR’s SSA representation?</a><a class="headerlink" href="#what-api-do-i-use-to-store-a-value-to-one-of-the-virtual-registers-in-llvm-ir-s-ssa-representation" title="Permalink to this headline">¶</a></h3>
+<p>In short: you can’t. It’s actually kind of a silly question once you grok
+what’s going on. Basically, in code like:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="nv">%result</span> <span class="p">=</span> <span class="k">add</span> <span class="k">i32</span> <span class="nv">%foo</span><span class="p">,</span> <span class="nv">%bar</span>
+</pre></div>
+</div>
+<p>, <tt class="docutils literal"><span class="pre">%result</span></tt> is just a name given to the <tt class="docutils literal"><span class="pre">Value</span></tt> of the <tt class="docutils literal"><span class="pre">add</span></tt>
+instruction. In other words, <tt class="docutils literal"><span class="pre">%result</span></tt> <em>is</em> the add instruction. The
+“assignment” doesn’t explicitly “store” anything to any “virtual register”;
+the “<tt class="docutils literal"><span class="pre">=</span></tt>” is more like the mathematical sense of equality.</p>
+<p>Longer explanation: In order to generate a textual representation of the
+IR, some kind of name has to be given to each instruction so that other
+instructions can textually reference it. However, the isomorphic in-memory
+representation that you manipulate from C++ has no such restriction since
+instructions can simply keep pointers to any other <tt class="docutils literal"><span class="pre">Value</span></tt>‘s that they
+reference. In fact, the names of dummy numbered temporaries like <tt class="docutils literal"><span class="pre">%1</span></tt> are
+not explicitly represented in the in-memory representation at all (see
+<tt class="docutils literal"><span class="pre">Value::getName()</span></tt>).</p>
+</div>
+</div>
+<div class="section" id="source-languages">
+<h2><a class="toc-backref" href="#id9">Source Languages</a><a class="headerlink" href="#source-languages" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="what-source-languages-are-supported">
+<h3><a class="toc-backref" href="#id10">What source languages are supported?</a><a class="headerlink" href="#what-source-languages-are-supported" title="Permalink to this headline">¶</a></h3>
+<p>LLVM currently has full support for C and C++ source languages through
+<a class="reference external" href="http://clang.llvm.org/">Clang</a>. Many other language frontends have
+been written using LLVM, and an incomplete list is available at
+<a class="reference external" href="http://llvm.org/ProjectsWithLLVM/">projects with LLVM</a>.</p>
+</div>
+<div class="section" id="i-d-like-to-write-a-self-hosting-llvm-compiler-how-should-i-interface-with-the-llvm-middle-end-optimizers-and-back-end-code-generators">
+<h3><a class="toc-backref" href="#id11">I’d like to write a self-hosting LLVM compiler. How should I interface with the LLVM middle-end optimizers and back-end code generators?</a><a class="headerlink" href="#i-d-like-to-write-a-self-hosting-llvm-compiler-how-should-i-interface-with-the-llvm-middle-end-optimizers-and-back-end-code-generators" title="Permalink to this headline">¶</a></h3>
+<p>Your compiler front-end will communicate with LLVM by creating a module in the
+LLVM intermediate representation (IR) format. Assuming you want to write your
+language’s compiler in the language itself (rather than C++), there are 3
+major ways to tackle generating LLVM IR from a front-end:</p>
+<ol class="arabic simple">
+<li><strong>Call into the LLVM libraries code using your language’s FFI (foreign
+function interface).</strong></li>
+</ol>
+<blockquote>
+<div><ul class="simple">
+<li><em>for:</em> best tracks changes to the LLVM IR, .ll syntax, and .bc format</li>
+<li><em>for:</em> enables running LLVM optimization passes without a emit/parse
+overhead</li>
+<li><em>for:</em> adapts well to a JIT context</li>
+<li><em>against:</em> lots of ugly glue code to write</li>
+</ul>
+</div></blockquote>
+<ol class="arabic simple" start="2">
+<li><strong>Emit LLVM assembly from your compiler’s native language.</strong></li>
+</ol>
+<blockquote>
+<div><ul class="simple">
+<li><em>for:</em> very straightforward to get started</li>
+<li><em>against:</em> the .ll parser is slower than the bitcode reader when
+interfacing to the middle end</li>
+<li><em>against:</em> it may be harder to track changes to the IR</li>
+</ul>
+</div></blockquote>
+<ol class="arabic simple" start="3">
+<li><strong>Emit LLVM bitcode from your compiler’s native language.</strong></li>
+</ol>
+<blockquote>
+<div><ul class="simple">
+<li><em>for:</em> can use the more-efficient bitcode reader when interfacing to the
+middle end</li>
+<li><em>against:</em> you’ll have to re-engineer the LLVM IR object model and bitcode
+writer in your language</li>
+<li><em>against:</em> it may be harder to track changes to the IR</li>
+</ul>
+</div></blockquote>
+<p>If you go with the first option, the C bindings in include/llvm-c should help
+a lot, since most languages have strong support for interfacing with C. The
+most common hurdle with calling C from managed code is interfacing with the
+garbage collector. The C interface was designed to require very little memory
+management, and so is straightforward in this regard.</p>
+</div>
+<div class="section" id="what-support-is-there-for-a-higher-level-source-language-constructs-for-building-a-compiler">
+<h3><a class="toc-backref" href="#id12">What support is there for a higher level source language constructs for building a compiler?</a><a class="headerlink" href="#what-support-is-there-for-a-higher-level-source-language-constructs-for-building-a-compiler" title="Permalink to this headline">¶</a></h3>
+<p>Currently, there isn’t much. LLVM supports an intermediate representation
+which is useful for code representation but will not support the high level
+(abstract syntax tree) representation needed by most compilers. There are no
+facilities for lexical nor semantic analysis.</p>
+</div>
+<div class="section" id="i-don-t-understand-the-getelementptr-instruction-help">
+<h3><a class="toc-backref" href="#id13">I don’t understand the <tt class="docutils literal"><span class="pre">GetElementPtr</span></tt> instruction. Help!</a><a class="headerlink" href="#i-don-t-understand-the-getelementptr-instruction-help" title="Permalink to this headline">¶</a></h3>
+<p>See <a class="reference external" href="GetElementPtr.html">The Often Misunderstood GEP Instruction</a>.</p>
+</div>
+</div>
+<div class="section" id="using-the-c-and-c-front-ends">
+<h2><a class="toc-backref" href="#id14">Using the C and C++ Front Ends</a><a class="headerlink" href="#using-the-c-and-c-front-ends" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="can-i-compile-c-or-c-code-to-platform-independent-llvm-bitcode">
+<h3><a class="toc-backref" href="#id15">Can I compile C or C++ code to platform-independent LLVM bitcode?</a><a class="headerlink" href="#can-i-compile-c-or-c-code-to-platform-independent-llvm-bitcode" title="Permalink to this headline">¶</a></h3>
+<p>No. C and C++ are inherently platform-dependent languages. The most obvious
+example of this is the preprocessor. A very common way that C code is made
+portable is by using the preprocessor to include platform-specific code. In
+practice, information about other platforms is lost after preprocessing, so
+the result is inherently dependent on the platform that the preprocessing was
+targeting.</p>
+<p>Another example is <tt class="docutils literal"><span class="pre">sizeof</span></tt>. It’s common for <tt class="docutils literal"><span class="pre">sizeof(long)</span></tt> to vary
+between platforms. In most C front-ends, <tt class="docutils literal"><span class="pre">sizeof</span></tt> is expanded to a
+constant immediately, thus hard-wiring a platform-specific detail.</p>
+<p>Also, since many platforms define their ABIs in terms of C, and since LLVM is
+lower-level than C, front-ends currently must emit platform-specific IR in
+order to have the result conform to the platform ABI.</p>
+</div>
+</div>
+<div class="section" id="questions-about-code-generated-by-the-demo-page">
+<h2><a class="toc-backref" href="#id16">Questions about code generated by the demo page</a><a class="headerlink" href="#questions-about-code-generated-by-the-demo-page" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="what-is-this-llvm-global-ctors-and-global-i-a-stuff-that-happens-when-i-include-iostream">
+<h3><a class="toc-backref" href="#id17">What is this <tt class="docutils literal"><span class="pre">llvm.global_ctors</span></tt> and <tt class="docutils literal"><span class="pre">_GLOBAL__I_a...</span></tt> stuff that happens when I <tt class="docutils literal"><span class="pre">#include</span> <span class="pre"><iostream></span></tt>?</a><a class="headerlink" href="#what-is-this-llvm-global-ctors-and-global-i-a-stuff-that-happens-when-i-include-iostream" title="Permalink to this headline">¶</a></h3>
+<p>If you <tt class="docutils literal"><span class="pre">#include</span></tt> the <tt class="docutils literal"><span class="pre"><iostream></span></tt> header into a C++ translation unit,
+the file will probably use the <tt class="docutils literal"><span class="pre">std::cin</span></tt>/<tt class="docutils literal"><span class="pre">std::cout</span></tt>/... global objects.
+However, C++ does not guarantee an order of initialization between static
+objects in different translation units, so if a static ctor/dtor in your .cpp
+file used <tt class="docutils literal"><span class="pre">std::cout</span></tt>, for example, the object would not necessarily be
+automatically initialized before your use.</p>
+<p>To make <tt class="docutils literal"><span class="pre">std::cout</span></tt> and friends work correctly in these scenarios, the STL
+that we use declares a static object that gets created in every translation
+unit that includes <tt class="docutils literal"><span class="pre"><iostream></span></tt>. This object has a static constructor
+and destructor that initializes and destroys the global iostream objects
+before they could possibly be used in the file. The code that you see in the
+<tt class="docutils literal"><span class="pre">.ll</span></tt> file corresponds to the constructor and destructor registration code.</p>
+<p>If you would like to make it easier to <em>understand</em> the LLVM code generated
+by the compiler in the demo page, consider using <tt class="docutils literal"><span class="pre">printf()</span></tt> instead of
+<tt class="docutils literal"><span class="pre">iostream</span></tt>s to print values.</p>
+</div>
+<div class="section" id="where-did-all-of-my-code-go">
+<h3><a class="toc-backref" href="#id18">Where did all of my code go??</a><a class="headerlink" href="#where-did-all-of-my-code-go" title="Permalink to this headline">¶</a></h3>
+<p>If you are using the LLVM demo page, you may often wonder what happened to
+all of the code that you typed in. Remember that the demo script is running
+the code through the LLVM optimizers, so if your code doesn’t actually do
+anything useful, it might all be deleted.</p>
+<p>To prevent this, make sure that the code is actually needed. For example, if
+you are computing some expression, return the value from the function instead
+of leaving it in a local variable. If you really want to constrain the
+optimizer, you can read from and assign to <tt class="docutils literal"><span class="pre">volatile</span></tt> global variables.</p>
+</div>
+<div class="section" id="what-is-this-undef-thing-that-shows-up-in-my-code">
+<h3><a class="toc-backref" href="#id19">What is this “<tt class="docutils literal"><span class="pre">undef</span></tt>” thing that shows up in my code?</a><a class="headerlink" href="#what-is-this-undef-thing-that-shows-up-in-my-code" title="Permalink to this headline">¶</a></h3>
+<p><tt class="docutils literal"><span class="pre">undef</span></tt> is the LLVM way of representing a value that is not defined. You
+can get these if you do not initialize a variable before you use it. For
+example, the C function:</p>
+<div class="highlight-c"><div class="highlight"><pre><span class="kt">int</span> <span class="nf">X</span><span class="p">()</span> <span class="p">{</span> <span class="kt">int</span> <span class="n">i</span><span class="p">;</span> <span class="k">return</span> <span class="n">i</span><span class="p">;</span> <span class="p">}</span>
+</pre></div>
+</div>
+<p>Is compiled to “<tt class="docutils literal"><span class="pre">ret</span> <span class="pre">i32</span> <span class="pre">undef</span></tt>” because “<tt class="docutils literal"><span class="pre">i</span></tt>” never has a value specified
+for it.</p>
+</div>
+<div class="section" id="why-does-instcombine-simplifycfg-turn-a-call-to-a-function-with-a-mismatched-calling-convention-into-unreachable-why-not-make-the-verifier-reject-it">
+<h3><a class="toc-backref" href="#id20">Why does instcombine + simplifycfg turn a call to a function with a mismatched calling convention into “unreachable”? Why not make the verifier reject it?</a><a class="headerlink" href="#why-does-instcombine-simplifycfg-turn-a-call-to-a-function-with-a-mismatched-calling-convention-into-unreachable-why-not-make-the-verifier-reject-it" title="Permalink to this headline">¶</a></h3>
+<p>This is a common problem run into by authors of front-ends that are using
+custom calling conventions: you need to make sure to set the right calling
+convention on both the function and on each call to the function. For
+example, this code:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span> <span class="p">{</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+<span class="k">define</span> <span class="kt">void</span> <span class="vg">@bar</span><span class="p">()</span> <span class="p">{</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>Is optimized to:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span> <span class="p">{</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+<span class="k">define</span> <span class="kt">void</span> <span class="vg">@bar</span><span class="p">()</span> <span class="p">{</span>
+ <span class="k">unreachable</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>... with “<tt class="docutils literal"><span class="pre">opt</span> <span class="pre">-instcombine</span> <span class="pre">-simplifycfg</span></tt>”. This often bites people because
+“all their code disappears”. Setting the calling convention on the caller and
+callee is required for indirect calls to work, so people often ask why not
+make the verifier reject this sort of thing.</p>
+<p>The answer is that this code has undefined behavior, but it is not illegal.
+If we made it illegal, then every transformation that could potentially create
+this would have to ensure that it doesn’t, and there is valid code that can
+create this sort of construct (in dead code). The sorts of things that can
+cause this to happen are fairly contrived, but we still need to accept them.
+Here’s an example:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span> <span class="p">{</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+<span class="k">define</span> <span class="k">internal</span> <span class="kt">void</span> <span class="vg">@bar</span><span class="p">(</span><span class="kt">void</span><span class="p">()*</span> <span class="nv">%FP</span><span class="p">,</span> <span class="k">i1</span> <span class="nv">%cond</span><span class="p">)</span> <span class="p">{</span>
+ <span class="k">br</span> <span class="k">i1</span> <span class="nv">%cond</span><span class="p">,</span> <span class="kt">label</span> <span class="nv">%T</span><span class="p">,</span> <span class="kt">label</span> <span class="nv">%F</span>
+<span class="nl">T:</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="nv">%FP</span><span class="p">()</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="nl">F:</span>
+ <span class="k">call</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="nv">%FP</span><span class="p">()</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+<span class="k">define</span> <span class="kt">void</span> <span class="vg">@test</span><span class="p">()</span> <span class="p">{</span>
+ <span class="nv">%X</span> <span class="p">=</span> <span class="k">or</span> <span class="k">i1</span> <span class="k">false</span><span class="p">,</span> <span class="k">false</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@bar</span><span class="p">(</span><span class="kt">void</span><span class="p">()*</span> <span class="vg">@foo</span><span class="p">,</span> <span class="k">i1</span> <span class="nv">%X</span><span class="p">)</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>In this example, “test” always passes <tt class="docutils literal"><span class="pre">@foo</span></tt>/<tt class="docutils literal"><span class="pre">false</span></tt> into <tt class="docutils literal"><span class="pre">bar</span></tt>, which
+ensures that it is dynamically called with the right calling conv (thus, the
+code is perfectly well defined). If you run this through the inliner, you
+get this (the explicit “or” is there so that the inliner doesn’t dead code
+eliminate a bunch of stuff):</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span> <span class="p">{</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+<span class="k">define</span> <span class="kt">void</span> <span class="vg">@test</span><span class="p">()</span> <span class="p">{</span>
+ <span class="nv">%X</span> <span class="p">=</span> <span class="k">or</span> <span class="k">i1</span> <span class="k">false</span><span class="p">,</span> <span class="k">false</span>
+ <span class="k">br</span> <span class="k">i1</span> <span class="nv">%X</span><span class="p">,</span> <span class="kt">label</span> <span class="nv">%T.i</span><span class="p">,</span> <span class="kt">label</span> <span class="nv">%F.i</span>
+<span class="nl">T.i:</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span>
+ <span class="k">br</span> <span class="kt">label</span> <span class="nv">%bar.exit</span>
+<span class="nl">F.i:</span>
+ <span class="k">call</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span>
+ <span class="k">br</span> <span class="kt">label</span> <span class="nv">%bar.exit</span>
+<span class="nl">bar.exit:</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>Here you can see that the inlining pass made an undefined call to <tt class="docutils literal"><span class="pre">@foo</span></tt>
+with the wrong calling convention. We really don’t want to make the inliner
+have to know about this sort of thing, so it needs to be valid code. In this
+case, dead code elimination can trivially remove the undefined code. However,
+if <tt class="docutils literal"><span class="pre">%X</span></tt> was an input argument to <tt class="docutils literal"><span class="pre">@test</span></tt>, the inliner would produce this:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span> <span class="p">{</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+
+<span class="k">define</span> <span class="kt">void</span> <span class="vg">@test</span><span class="p">(</span><span class="k">i1</span> <span class="nv">%X</span><span class="p">)</span> <span class="p">{</span>
+ <span class="k">br</span> <span class="k">i1</span> <span class="nv">%X</span><span class="p">,</span> <span class="kt">label</span> <span class="nv">%T.i</span><span class="p">,</span> <span class="kt">label</span> <span class="nv">%F.i</span>
+<span class="nl">T.i:</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span>
+ <span class="k">br</span> <span class="kt">label</span> <span class="nv">%bar.exit</span>
+<span class="nl">F.i:</span>
+ <span class="k">call</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span>
+ <span class="k">br</span> <span class="kt">label</span> <span class="nv">%bar.exit</span>
+<span class="nl">bar.exit:</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>The interesting thing about this is that <tt class="docutils literal"><span class="pre">%X</span></tt> <em>must</em> be false for the
+code to be well-defined, but no amount of dead code elimination will be able
+to delete the broken call as unreachable. However, since
+<tt class="docutils literal"><span class="pre">instcombine</span></tt>/<tt class="docutils literal"><span class="pre">simplifycfg</span></tt> turns the undefined call into unreachable, we
+end up with a branch on a condition that goes to unreachable: a branch to
+unreachable can never happen, so “<tt class="docutils literal"><span class="pre">-inline</span> <span class="pre">-instcombine</span> <span class="pre">-simplifycfg</span></tt>” is
+able to produce:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span> <span class="p">{</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+<span class="k">define</span> <span class="kt">void</span> <span class="vg">@test</span><span class="p">(</span><span class="k">i1</span> <span class="nv">%X</span><span class="p">)</span> <span class="p">{</span>
+<span class="nl">F.i:</span>
+ <span class="k">call</span> <span class="k">fastcc</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span>
+ <span class="k">ret</span> <span class="kt">void</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+</div>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="Lexicon.html" title="The LLVM Lexicon"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="GettingStartedVS.html" title="Getting Started with the LLVM System using Microsoft Visual Studio"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/FaultMaps.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/FaultMaps.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/FaultMaps.html (added)
+++ www-releases/trunk/4.0.0/docs/FaultMaps.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,215 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>FaultMaps and implicit checks — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="Machine IR (MIR) Format Reference Manual" href="MIRLangRef.html" />
+ <link rel="prev" title="Type Metadata" href="TypeMetadata.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="MIRLangRef.html" title="Machine IR (MIR) Format Reference Manual"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="TypeMetadata.html" title="Type Metadata"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="faultmaps-and-implicit-checks">
+<h1>FaultMaps and implicit checks<a class="headerlink" href="#faultmaps-and-implicit-checks" title="Permalink to this headline">¶</a></h1>
+<div class="contents local topic" id="contents">
+<ul class="simple">
+<li><a class="reference internal" href="#motivation" id="id1">Motivation</a></li>
+<li><a class="reference internal" href="#the-fault-map-section" id="id2">The Fault Map Section</a></li>
+<li><a class="reference internal" href="#the-implicitnullchecks-pass" id="id3">The <tt class="docutils literal"><span class="pre">ImplicitNullChecks</span></tt> pass</a><ul>
+<li><a class="reference internal" href="#make-implicit-metadata" id="id4"><tt class="docutils literal"><span class="pre">make.implicit</span></tt> metadata</a></li>
+</ul>
+</li>
+</ul>
+</div>
+<div class="section" id="motivation">
+<h2><a class="toc-backref" href="#id1">Motivation</a><a class="headerlink" href="#motivation" title="Permalink to this headline">¶</a></h2>
+<p>Code generated by managed language runtimes tend to have checks that
+are required for safety but never fail in practice. In such cases, it
+is profitable to make the non-failing case cheaper even if it makes
+the failing case significantly more expensive. This asymmetry can be
+exploited by folding such safety checks into operations that can be
+made to fault reliably if the check would have failed, and recovering
+from such a fault by using a signal handler.</p>
+<p>For example, Java requires null checks on objects before they are read
+from or written to. If the object is <tt class="docutils literal"><span class="pre">null</span></tt> then a
+<tt class="docutils literal"><span class="pre">NullPointerException</span></tt> has to be thrown, interrupting normal
+execution. In practice, however, dereferencing a <tt class="docutils literal"><span class="pre">null</span></tt> pointer is
+extremely rare in well-behaved Java programs, and typically the null
+check can be folded into a nearby memory operation that operates on
+the same memory location.</p>
+</div>
+<div class="section" id="the-fault-map-section">
+<h2><a class="toc-backref" href="#id2">The Fault Map Section</a><a class="headerlink" href="#the-fault-map-section" title="Permalink to this headline">¶</a></h2>
+<p>Information about implicit checks generated by LLVM are put in a
+special “fault map” section. On Darwin this section is named
+<tt class="docutils literal"><span class="pre">__llvm_faultmaps</span></tt>.</p>
+<p>The format of this section is</p>
+<div class="highlight-none"><div class="highlight"><pre>Header {
+ uint8 : Fault Map Version (current version is 1)
+ uint8 : Reserved (expected to be 0)
+ uint16 : Reserved (expected to be 0)
+}
+uint32 : NumFunctions
+FunctionInfo[NumFunctions] {
+ uint64 : FunctionAddress
+ uint32 : NumFaultingPCs
+ uint32 : Reserved (expected to be 0)
+ FunctionFaultInfo[NumFaultingPCs] {
+ uint32 : FaultKind = FaultMaps::FaultingLoad (only legal value currently)
+ uint32 : FaultingPCOffset
+ uint32 : HandlerPCOffset
+ }
+}
+</pre></div>
+</div>
+</div>
+<div class="section" id="the-implicitnullchecks-pass">
+<h2><a class="toc-backref" href="#id3">The <tt class="docutils literal"><span class="pre">ImplicitNullChecks</span></tt> pass</a><a class="headerlink" href="#the-implicitnullchecks-pass" title="Permalink to this headline">¶</a></h2>
+<p>The <tt class="docutils literal"><span class="pre">ImplicitNullChecks</span></tt> pass transforms explicit control flow for
+checking if a pointer is <tt class="docutils literal"><span class="pre">null</span></tt>, like:</p>
+<div class="highlight-llvm"><div class="highlight"><pre> <span class="nv">%ptr</span> <span class="p">=</span> <span class="k">call</span> <span class="k">i32</span><span class="p">*</span> <span class="vg">@get_ptr</span><span class="p">()</span>
+ <span class="nv">%ptr_is_null</span> <span class="p">=</span> <span class="k">icmp</span> <span class="k">i32</span><span class="p">*</span> <span class="nv">%ptr</span><span class="p">,</span> <span class="k">null</span>
+ <span class="k">br</span> <span class="k">i1</span> <span class="nv">%ptr_is_null</span><span class="p">,</span> <span class="kt">label</span> <span class="nv">%is_null</span><span class="p">,</span> <span class="kt">label</span> <span class="nv">%not_null</span><span class="p">,</span> <span class="nv">!make.implicit</span> <span class="nv-Anonymous">!0</span>
+
+<span class="nl">not_null:</span>
+ <span class="nv">%t</span> <span class="p">=</span> <span class="k">load</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span><span class="p">*</span> <span class="nv">%ptr</span>
+ <span class="k">br</span> <span class="kt">label</span> <span class="nv">%do_something_with_t</span>
+
+<span class="nl">is_null:</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@HFC</span><span class="p">()</span>
+ <span class="k">unreachable</span>
+
+<span class="nv-Anonymous">!0</span> <span class="p">=</span> <span class="p">!{}</span>
+</pre></div>
+</div>
+<p>to control flow implicit in the instruction loading or storing through
+the pointer being null checked:</p>
+<div class="highlight-llvm"><div class="highlight"><pre> <span class="nv">%ptr</span> <span class="p">=</span> <span class="k">call</span> <span class="k">i32</span><span class="p">*</span> <span class="vg">@get_ptr</span><span class="p">()</span>
+ <span class="nv">%t</span> <span class="p">=</span> <span class="k">load</span> <span class="k">i32</span><span class="p">,</span> <span class="k">i32</span><span class="p">*</span> <span class="nv">%ptr</span> <span class="c">;; handler-pc = label %is_null</span>
+ <span class="k">br</span> <span class="kt">label</span> <span class="nv">%do_something_with_t</span>
+
+<span class="nl">is_null:</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@HFC</span><span class="p">()</span>
+ <span class="k">unreachable</span>
+</pre></div>
+</div>
+<p>This transform happens at the <tt class="docutils literal"><span class="pre">MachineInstr</span></tt> level, not the LLVM IR
+level (so the above example is only representative, not literal). The
+<tt class="docutils literal"><span class="pre">ImplicitNullChecks</span></tt> pass runs during codegen, if
+<tt class="docutils literal"><span class="pre">-enable-implicit-null-checks</span></tt> is passed to <tt class="docutils literal"><span class="pre">llc</span></tt>.</p>
+<p>The <tt class="docutils literal"><span class="pre">ImplicitNullChecks</span></tt> pass adds entries to the
+<tt class="docutils literal"><span class="pre">__llvm_faultmaps</span></tt> section described above as needed.</p>
+<div class="section" id="make-implicit-metadata">
+<h3><a class="toc-backref" href="#id4"><tt class="docutils literal"><span class="pre">make.implicit</span></tt> metadata</a><a class="headerlink" href="#make-implicit-metadata" title="Permalink to this headline">¶</a></h3>
+<p>Making null checks implicit is an aggressive optimization, and it can
+be a net performance pessimization if too many memory operations end
+up faulting because of it. A language runtime typically needs to
+ensure that only a negligible number of implicit null checks actually
+fault once the application has reached a steady state. A standard way
+of doing this is by healing failed implicit null checks into explicit
+null checks via code patching or recompilation. It follows that there
+are two requirements an explicit null check needs to satisfy for it to
+be profitable to convert it to an implicit null check:</p>
+<blockquote>
+<div><ol class="arabic simple">
+<li>The case where the pointer is actually null (i.e. the “failing”
+case) is extremely rare.</li>
+<li>The failing path heals the implicit null check into an explicit
+null check so that the application does not repeatedly page
+fault.</li>
+</ol>
+</div></blockquote>
+<p>The frontend is expected to mark branches that satisfy (1) and (2)
+using a <tt class="docutils literal"><span class="pre">!make.implicit</span></tt> metadata node (the actual content of the
+metadata node is ignored). Only branches that are marked with
+<tt class="docutils literal"><span class="pre">!make.implicit</span></tt> metadata are considered as candidates for
+conversion into implicit null checks.</p>
+<p>(Note that while we could deal with (1) using profiling data, dealing
+with (2) requires some information not present in branch profiles.)</p>
+</div>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="MIRLangRef.html" title="Machine IR (MIR) Format Reference Manual"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="TypeMetadata.html" title="Type Metadata"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/Frontend/PerformanceTips.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/Frontend/PerformanceTips.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/Frontend/PerformanceTips.html (added)
+++ www-releases/trunk/4.0.0/docs/Frontend/PerformanceTips.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,373 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>Performance Tips for Frontend Authors — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="../_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="../_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: '../',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="../_static/jquery.js"></script>
+ <script type="text/javascript" src="../_static/underscore.js"></script>
+ <script type="text/javascript" src="../_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="../index.html" />
+ <link rel="next" title="MCJIT Design and Implementation" href="../MCJITDesignAndImplementation.html" />
+ <link rel="prev" title="The Often Misunderstood GEP Instruction" href="../GetElementPtr.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="../index.html">
+ <img src="../_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="../MCJITDesignAndImplementation.html" title="MCJIT Design and Implementation"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="../GetElementPtr.html" title="The Often Misunderstood GEP Instruction"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="performance-tips-for-frontend-authors">
+<h1>Performance Tips for Frontend Authors<a class="headerlink" href="#performance-tips-for-frontend-authors" title="Permalink to this headline">¶</a></h1>
+<div class="contents local topic" id="contents">
+<ul class="simple">
+<li><a class="reference internal" href="#abstract" id="id2">Abstract</a></li>
+<li><a class="reference internal" href="#ir-best-practices" id="id3">IR Best Practices</a><ul>
+<li><a class="reference internal" href="#the-basics" id="id4">The Basics</a></li>
+<li><a class="reference internal" href="#use-of-allocas" id="id5">Use of allocas</a></li>
+<li><a class="reference internal" href="#avoid-loads-and-stores-of-large-aggregate-type" id="id6">Avoid loads and stores of large aggregate type</a></li>
+<li><a class="reference internal" href="#prefer-zext-over-sext-when-legal" id="id7">Prefer zext over sext when legal</a></li>
+<li><a class="reference internal" href="#zext-gep-indices-to-machine-register-width" id="id8">Zext GEP indices to machine register width</a></li>
+<li><a class="reference internal" href="#when-to-specify-alignment" id="id9">When to specify alignment</a></li>
+<li><a class="reference internal" href="#other-things-to-consider" id="id10">Other Things to Consider</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#describing-language-specific-properties" id="id11">Describing Language Specific Properties</a><ul>
+<li><a class="reference internal" href="#restricted-operation-semantics" id="id12">Restricted Operation Semantics</a></li>
+<li><a class="reference internal" href="#describing-aliasing-properties" id="id13">Describing Aliasing Properties</a></li>
+<li><a class="reference internal" href="#modeling-memory-effects" id="id14">Modeling Memory Effects</a></li>
+<li><a class="reference internal" href="#pass-ordering" id="id15">Pass Ordering</a></li>
+<li><a class="reference internal" href="#i-still-can-t-find-what-i-m-looking-for" id="id16">I Still Can’t Find What I’m Looking For</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#adding-to-this-document" id="id17">Adding to this document</a></li>
+</ul>
+</div>
+<div class="section" id="abstract">
+<h2><a class="toc-backref" href="#id2">Abstract</a><a class="headerlink" href="#abstract" title="Permalink to this headline">¶</a></h2>
+<p>The intended audience of this document is developers of language frontends
+targeting LLVM IR. This document is home to a collection of tips on how to
+generate IR that optimizes well.</p>
+</div>
+<div class="section" id="ir-best-practices">
+<h2><a class="toc-backref" href="#id3">IR Best Practices</a><a class="headerlink" href="#ir-best-practices" title="Permalink to this headline">¶</a></h2>
+<p>As with any optimizer, LLVM has its strengths and weaknesses. In some cases,
+surprisingly small changes in the source IR can have a large effect on the
+generated code.</p>
+<p>Beyond the specific items on the list below, it’s worth noting that the most
+mature frontend for LLVM is Clang. As a result, the further your IR gets from what Clang might emit, the less likely it is to be effectively optimized. It
+can often be useful to write a quick C program with the semantics you’re trying
+to model and see what decisions Clang’s IRGen makes about what IR to emit.
+Studying Clang’s CodeGen directory can also be a good source of ideas. Note
+that Clang and LLVM are explicitly version locked so you’ll need to make sure
+you’re using a Clang built from the same svn revision or release as the LLVM
+library you’re using. As always, it’s <em>strongly</em> recommended that you track
+tip of tree development, particularly during bring up of a new project.</p>
+<div class="section" id="the-basics">
+<h3><a class="toc-backref" href="#id4">The Basics</a><a class="headerlink" href="#the-basics" title="Permalink to this headline">¶</a></h3>
+<ol class="arabic simple">
+<li>Make sure that your Modules contain both a data layout specification and
+target triple. Without these pieces, non of the target specific optimization
+will be enabled. This can have a major effect on the generated code quality.</li>
+<li>For each function or global emitted, use the most private linkage type
+possible (private, internal or linkonce_odr preferably). Doing so will
+make LLVM’s inter-procedural optimizations much more effective.</li>
+<li>Avoid high in-degree basic blocks (e.g. basic blocks with dozens or hundreds
+of predecessors). Among other issues, the register allocator is known to
+perform badly with confronted with such structures. The only exception to
+this guidance is that a unified return block with high in-degree is fine.</li>
+</ol>
+</div>
+<div class="section" id="use-of-allocas">
+<h3><a class="toc-backref" href="#id5">Use of allocas</a><a class="headerlink" href="#use-of-allocas" title="Permalink to this headline">¶</a></h3>
+<p>An alloca instruction can be used to represent a function scoped stack slot,
+but can also represent dynamic frame expansion. When representing function
+scoped variables or locations, placing alloca instructions at the beginning of
+the entry block should be preferred. In particular, place them before any
+call instructions. Call instructions might get inlined and replaced with
+multiple basic blocks. The end result is that a following alloca instruction
+would no longer be in the entry basic block afterward.</p>
+<p>The SROA (Scalar Replacement Of Aggregates) and Mem2Reg passes only attempt
+to eliminate alloca instructions that are in the entry basic block. Given
+SSA is the canonical form expected by much of the optimizer; if allocas can
+not be eliminated by Mem2Reg or SROA, the optimizer is likely to be less
+effective than it could be.</p>
+</div>
+<div class="section" id="avoid-loads-and-stores-of-large-aggregate-type">
+<h3><a class="toc-backref" href="#id6">Avoid loads and stores of large aggregate type</a><a class="headerlink" href="#avoid-loads-and-stores-of-large-aggregate-type" title="Permalink to this headline">¶</a></h3>
+<p>LLVM currently does not optimize well loads and stores of large <a class="reference internal" href="../LangRef.html#t-aggregate"><em>aggregate
+types</em></a> (i.e. structs and arrays). As an alternative, consider
+loading individual fields from memory.</p>
+<p>Aggregates that are smaller than the largest (performant) load or store
+instruction supported by the targeted hardware are well supported. These can
+be an effective way to represent collections of small packed fields.</p>
+</div>
+<div class="section" id="prefer-zext-over-sext-when-legal">
+<h3><a class="toc-backref" href="#id7">Prefer zext over sext when legal</a><a class="headerlink" href="#prefer-zext-over-sext-when-legal" title="Permalink to this headline">¶</a></h3>
+<p>On some architectures (X86_64 is one), sign extension can involve an extra
+instruction whereas zero extension can be folded into a load. LLVM will try to
+replace a sext with a zext when it can be proven safe, but if you have
+information in your source language about the range of a integer value, it can
+be profitable to use a zext rather than a sext.</p>
+<p>Alternatively, you can <a class="reference internal" href="../LangRef.html#range-metadata"><em>specify the range of the value using metadata</em></a> and LLVM can do the sext to zext conversion for you.</p>
+</div>
+<div class="section" id="zext-gep-indices-to-machine-register-width">
+<h3><a class="toc-backref" href="#id8">Zext GEP indices to machine register width</a><a class="headerlink" href="#zext-gep-indices-to-machine-register-width" title="Permalink to this headline">¶</a></h3>
+<p>Internally, LLVM often promotes the width of GEP indices to machine register
+width. When it does so, it will default to using sign extension (sext)
+operations for safety. If your source language provides information about
+the range of the index, you may wish to manually extend indices to machine
+register width using a zext instruction.</p>
+</div>
+<div class="section" id="when-to-specify-alignment">
+<h3><a class="toc-backref" href="#id9">When to specify alignment</a><a class="headerlink" href="#when-to-specify-alignment" title="Permalink to this headline">¶</a></h3>
+<p>LLVM will always generate correct code if you donât specify alignment, but may
+generate inefficient code. For example, if you are targeting MIPS (or older
+ARM ISAs) then the hardware does not handle unaligned loads and stores, and
+so you will enter a trap-and-emulate path if you do a load or store with
+lower-than-natural alignment. To avoid this, LLVM will emit a slower
+sequence of loads, shifts and masks (or load-right + load-left on MIPS) for
+all cases where the load / store does not have a sufficiently high alignment
+in the IR.</p>
+<p>The alignment is used to guarantee the alignment on allocas and globals,
+though in most cases this is unnecessary (most targets have a sufficiently
+high default alignment that theyâll be fine). It is also used to provide a
+contract to the back end saying âeither this load/store has this alignment, or
+it is undefined behaviorâ. This means that the back end is free to emit
+instructions that rely on that alignment (and mid-level optimizers are free to
+perform transforms that require that alignment). For x86, it doesnât make
+much difference, as almost all instructions are alignment-independent. For
+MIPS, it can make a big difference.</p>
+<p>Note that if your loads and stores are atomic, the backend will be unable to
+lower an under aligned access into a sequence of natively aligned accesses.
+As a result, alignment is mandatory for atomic loads and stores.</p>
+</div>
+<div class="section" id="other-things-to-consider">
+<h3><a class="toc-backref" href="#id10">Other Things to Consider</a><a class="headerlink" href="#other-things-to-consider" title="Permalink to this headline">¶</a></h3>
+<ol class="arabic simple">
+<li>Use ptrtoint/inttoptr sparingly (they interfere with pointer aliasing
+analysis), prefer GEPs</li>
+<li>Prefer globals over inttoptr of a constant address - this gives you
+dereferencability information. In MCJIT, use getSymbolAddress to provide
+actual address.</li>
+<li>Be wary of ordered and atomic memory operations. They are hard to optimize
+and may not be well optimized by the current optimizer. Depending on your
+source language, you may consider using fences instead.</li>
+<li>If calling a function which is known to throw an exception (unwind), use
+an invoke with a normal destination which contains an unreachable
+instruction. This form conveys to the optimizer that the call returns
+abnormally. For an invoke which neither returns normally or requires unwind
+code in the current function, you can use a noreturn call instruction if
+desired. This is generally not required because the optimizer will convert
+an invoke with an unreachable unwind destination to a call instruction.</li>
+<li>Use profile metadata to indicate statically known cold paths, even if
+dynamic profiling information is not available. This can make a large
+difference in code placement and thus the performance of tight loops.</li>
+<li>When generating code for loops, try to avoid terminating the header block of
+the loop earlier than necessary. If the terminator of the loop header
+block is a loop exiting conditional branch, the effectiveness of LICM will
+be limited for loads not in the header. (This is due to the fact that LLVM
+may not know such a load is safe to speculatively execute and thus can’t
+lift an otherwise loop invariant load unless it can prove the exiting
+condition is not taken.) It can be profitable, in some cases, to emit such
+instructions into the header even if they are not used along a rarely
+executed path that exits the loop. This guidance specifically does not
+apply if the condition which terminates the loop header is itself invariant,
+or can be easily discharged by inspecting the loop index variables.</li>
+<li>In hot loops, consider duplicating instructions from small basic blocks
+which end in highly predictable terminators into their successor blocks.
+If a hot successor block contains instructions which can be vectorized
+with the duplicated ones, this can provide a noticeable throughput
+improvement. Note that this is not always profitable and does involve a
+potentially large increase in code size.</li>
+<li>When checking a value against a constant, emit the check using a consistent
+comparison type. The GVN pass <em>will</em> optimize redundant equalities even if
+the type of comparison is inverted, but GVN only runs late in the pipeline.
+As a result, you may miss the opportunity to run other important
+optimizations. Improvements to EarlyCSE to remove this issue are tracked in
+Bug 23333.</li>
+<li>Avoid using arithmetic intrinsics unless you are <em>required</em> by your source
+language specification to emit a particular code sequence. The optimizer
+is quite good at reasoning about general control flow and arithmetic, it is
+not anywhere near as strong at reasoning about the various intrinsics. If
+profitable for code generation purposes, the optimizer will likely form the
+intrinsics itself late in the optimization pipeline. It is <em>very</em> rarely
+profitable to emit these directly in the language frontend. This item
+explicitly includes the use of the <a class="reference internal" href="../LangRef.html#int-overflow"><em>overflow intrinsics</em></a>.</li>
+<li>Avoid using the <a class="reference internal" href="../LangRef.html#int-assume"><em>assume intrinsic</em></a> until you’ve
+established that a) there’s no other way to express the given fact and b)
+that fact is critical for optimization purposes. Assumes are a great
+prototyping mechanism, but they can have negative effects on both compile
+time and optimization effectiveness. The former is fixable with enough
+effort, but the later is fairly fundamental to their designed purpose.</li>
+</ol>
+</div>
+</div>
+<div class="section" id="describing-language-specific-properties">
+<h2><a class="toc-backref" href="#id11">Describing Language Specific Properties</a><a class="headerlink" href="#describing-language-specific-properties" title="Permalink to this headline">¶</a></h2>
+<p>When translating a source language to LLVM, finding ways to express concepts
+and guarantees available in your source language which are not natively
+provided by LLVM IR will greatly improve LLVM’s ability to optimize your code.
+As an example, C/C++’s ability to mark every add as “no signed wrap (nsw)” goes
+a long way to assisting the optimizer in reasoning about loop induction
+variables and thus generating more optimal code for loops.</p>
+<p>The LLVM LangRef includes a number of mechanisms for annotating the IR with
+additional semantic information. It is <em>strongly</em> recommended that you become
+highly familiar with this document. The list below is intended to highlight a
+couple of items of particular interest, but is by no means exhaustive.</p>
+<div class="section" id="restricted-operation-semantics">
+<h3><a class="toc-backref" href="#id12">Restricted Operation Semantics</a><a class="headerlink" href="#restricted-operation-semantics" title="Permalink to this headline">¶</a></h3>
+<ol class="arabic simple">
+<li>Add nsw/nuw flags as appropriate. Reasoning about overflow is
+generally hard for an optimizer so providing these facts from the frontend
+can be very impactful.</li>
+<li>Use fast-math flags on floating point operations if legal. If you don’t
+need strict IEEE floating point semantics, there are a number of additional
+optimizations that can be performed. This can be highly impactful for
+floating point intensive computations.</li>
+</ol>
+</div>
+<div class="section" id="describing-aliasing-properties">
+<h3><a class="toc-backref" href="#id13">Describing Aliasing Properties</a><a class="headerlink" href="#describing-aliasing-properties" title="Permalink to this headline">¶</a></h3>
+<ol class="arabic simple">
+<li>Add noalias/align/dereferenceable/nonnull to function arguments and return
+values as appropriate</li>
+<li>Use pointer aliasing metadata, especially tbaa metadata, to communicate
+otherwise-non-deducible pointer aliasing facts</li>
+<li>Use inbounds on geps. This can help to disambiguate some aliasing queries.</li>
+</ol>
+</div>
+<div class="section" id="modeling-memory-effects">
+<h3><a class="toc-backref" href="#id14">Modeling Memory Effects</a><a class="headerlink" href="#modeling-memory-effects" title="Permalink to this headline">¶</a></h3>
+<ol class="arabic simple">
+<li>Mark functions as readnone/readonly/argmemonly or noreturn/nounwind when
+known. The optimizer will try to infer these flags, but may not always be
+able to. Manual annotations are particularly important for external
+functions that the optimizer can not analyze.</li>
+<li>Use the lifetime.start/lifetime.end and invariant.start/invariant.end
+intrinsics where possible. Common profitable uses are for stack like data
+structures (thus allowing dead store elimination) and for describing
+life times of allocas (thus allowing smaller stack sizes).</li>
+<li>Mark invariant locations using !invariant.load and TBAA’s constant flags</li>
+</ol>
+</div>
+<div class="section" id="pass-ordering">
+<h3><a class="toc-backref" href="#id15">Pass Ordering</a><a class="headerlink" href="#pass-ordering" title="Permalink to this headline">¶</a></h3>
+<p>One of the most common mistakes made by new language frontend projects is to
+use the existing -O2 or -O3 pass pipelines as is. These pass pipelines make a
+good starting point for an optimizing compiler for any language, but they have
+been carefully tuned for C and C++, not your target language. You will almost
+certainly need to use a custom pass order to achieve optimal performance. A
+couple specific suggestions:</p>
+<ol class="arabic simple">
+<li>For languages with numerous rarely executed guard conditions (e.g. null
+checks, type checks, range checks) consider adding an extra execution or
+two of LoopUnswith and LICM to your pass order. The standard pass order,
+which is tuned for C and C++ applications, may not be sufficient to remove
+all dischargeable checks from loops.</li>
+<li>If you language uses range checks, consider using the IRCE pass. It is not
+currently part of the standard pass order.</li>
+<li>A useful sanity check to run is to run your optimized IR back through the
+-O2 pipeline again. If you see noticeable improvement in the resulting IR,
+you likely need to adjust your pass order.</li>
+</ol>
+</div>
+<div class="section" id="i-still-can-t-find-what-i-m-looking-for">
+<h3><a class="toc-backref" href="#id16">I Still Can’t Find What I’m Looking For</a><a class="headerlink" href="#i-still-can-t-find-what-i-m-looking-for" title="Permalink to this headline">¶</a></h3>
+<p>If you didn’t find what you were looking for above, consider proposing an piece
+of metadata which provides the optimization hint you need. Such extensions are
+relatively common and are generally well received by the community. You will
+need to ensure that your proposal is sufficiently general so that it benefits
+others if you wish to contribute it upstream.</p>
+<p>You should also consider describing the problem you’re facing on <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/llvm-dev">llvm-dev</a> and asking for advice.
+It’s entirely possible someone has encountered your problem before and can
+give good advice. If there are multiple interested parties, that also
+increases the chances that a metadata extension would be well received by the
+community as a whole.</p>
+</div>
+</div>
+<div class="section" id="adding-to-this-document">
+<h2><a class="toc-backref" href="#id17">Adding to this document</a><a class="headerlink" href="#adding-to-this-document" title="Permalink to this headline">¶</a></h2>
+<p>If you run across a case that you feel deserves to be covered here, please send
+a patch to <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/llvm-commits">llvm-commits</a> for review.</p>
+<p>If you have questions on these items, please direct them to <a class="reference external" href="http://lists.llvm.org/mailman/listinfo/llvm-dev">llvm-dev</a>. The more relevant
+context you are able to give to your question, the more likely it is to be
+answered.</p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="../genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="../MCJITDesignAndImplementation.html" title="MCJIT Design and Implementation"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="../GetElementPtr.html" title="The Often Misunderstood GEP Instruction"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="../index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
Added: www-releases/trunk/4.0.0/docs/GarbageCollection.html
URL: http://llvm.org/viewvc/llvm-project/www-releases/trunk/4.0.0/docs/GarbageCollection.html?rev=297634&view=auto
==============================================================================
--- www-releases/trunk/4.0.0/docs/GarbageCollection.html (added)
+++ www-releases/trunk/4.0.0/docs/GarbageCollection.html Mon Mar 13 11:30:12 2017
@@ -0,0 +1,1227 @@
+
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+
+
+<html xmlns="http://www.w3.org/1999/xhtml">
+ <head>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
+
+ <title>Garbage Collection with LLVM — LLVM 4 documentation</title>
+
+ <link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
+ <link rel="stylesheet" href="_static/pygments.css" type="text/css" />
+
+ <script type="text/javascript">
+ var DOCUMENTATION_OPTIONS = {
+ URL_ROOT: './',
+ VERSION: '4',
+ COLLAPSE_INDEX: false,
+ FILE_SUFFIX: '.html',
+ HAS_SOURCE: true
+ };
+ </script>
+ <script type="text/javascript" src="_static/jquery.js"></script>
+ <script type="text/javascript" src="_static/underscore.js"></script>
+ <script type="text/javascript" src="_static/doctools.js"></script>
+ <link rel="top" title="LLVM 4 documentation" href="index.html" />
+ <link rel="next" title="Writing an LLVM Pass" href="WritingAnLLVMPass.html" />
+ <link rel="prev" title="How To Use Instruction Mappings" href="HowToUseInstrMappings.html" />
+<style type="text/css">
+ table.right { float: right; margin-left: 20px; }
+ table.right td { border: 1px solid #ccc; }
+</style>
+
+ </head>
+ <body>
+<div class="logo">
+ <a href="index.html">
+ <img src="_static/logo.png"
+ alt="LLVM Logo" width="250" height="88"/></a>
+</div>
+
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ accesskey="I">index</a></li>
+ <li class="right" >
+ <a href="WritingAnLLVMPass.html" title="Writing an LLVM Pass"
+ accesskey="N">next</a> |</li>
+ <li class="right" >
+ <a href="HowToUseInstrMappings.html" title="How To Use Instruction Mappings"
+ accesskey="P">previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+
+
+ <div class="document">
+ <div class="documentwrapper">
+ <div class="body">
+
+ <div class="section" id="garbage-collection-with-llvm">
+<h1>Garbage Collection with LLVM<a class="headerlink" href="#garbage-collection-with-llvm" title="Permalink to this headline">¶</a></h1>
+<div class="contents local topic" id="contents">
+<ul class="simple">
+<li><a class="reference internal" href="#abstract" id="id1">Abstract</a></li>
+<li><a class="reference internal" href="#quick-start" id="id2">Quick Start</a></li>
+<li><a class="reference internal" href="#introduction" id="id3">Introduction</a><ul>
+<li><a class="reference internal" href="#what-is-garbage-collection" id="id4">What is Garbage Collection?</a></li>
+<li><a class="reference internal" href="#goals-and-non-goals" id="id5">Goals and non-goals</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#llvm-ir-features" id="id6">LLVM IR Features</a><ul>
+<li><a class="reference internal" href="#specifying-gc-code-generation-gc" id="id7">Specifying GC code generation: <tt class="docutils literal"><span class="pre">gc</span> <span class="pre">"..."</span></tt></a></li>
+<li><a class="reference internal" href="#gcroot" id="id8">Identifying GC roots on the stack</a><ul>
+<li><a class="reference internal" href="#using-gc-statepoint" id="id9">Using <tt class="docutils literal"><span class="pre">gc.statepoint</span></tt></a></li>
+<li><a class="reference internal" href="#using-llvm-gcwrite" id="id10">Using <tt class="docutils literal"><span class="pre">llvm.gcwrite</span></tt></a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#reading-and-writing-references-in-the-heap" id="id11">Reading and writing references in the heap</a><ul>
+<li><a class="reference internal" href="#write-barrier-llvm-gcwrite" id="id12">Write barrier: <tt class="docutils literal"><span class="pre">llvm.gcwrite</span></tt></a></li>
+<li><a class="reference internal" href="#read-barrier-llvm-gcread" id="id13">Read barrier: <tt class="docutils literal"><span class="pre">llvm.gcread</span></tt></a></li>
+</ul>
+</li>
+</ul>
+</li>
+<li><a class="reference internal" href="#built-in-gc-strategies" id="id14">Built In GC Strategies</a><ul>
+<li><a class="reference internal" href="#the-shadow-stack-gc" id="id15">The Shadow Stack GC</a></li>
+<li><a class="reference internal" href="#the-erlang-and-ocaml-gcs" id="id16">The ‘Erlang’ and ‘Ocaml’ GCs</a></li>
+<li><a class="reference internal" href="#the-statepoint-example-gc" id="id17">The Statepoint Example GC</a></li>
+<li><a class="reference internal" href="#the-coreclr-gc" id="id18">The CoreCLR GC</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#custom-gc-strategies" id="id19">Custom GC Strategies</a><ul>
+<li><a class="reference internal" href="#collector-requirements" id="id20">Collector Requirements</a></li>
+<li><a class="reference internal" href="#implementing-a-collector-plugin" id="id21">Implementing a collector plugin</a></li>
+<li><a class="reference internal" href="#overview-of-available-features" id="id22">Overview of available features</a></li>
+<li><a class="reference internal" href="#computing-stack-maps" id="id23">Computing stack maps</a></li>
+<li><a class="reference internal" href="#initializing-roots-to-null-initroots" id="id24">Initializing roots to null: <tt class="docutils literal"><span class="pre">InitRoots</span></tt></a></li>
+<li><a class="reference internal" href="#custom-lowering-of-intrinsics-customroots-customreadbarriers-and-customwritebarriers" id="id25">Custom lowering of intrinsics: <tt class="docutils literal"><span class="pre">CustomRoots</span></tt>, <tt class="docutils literal"><span class="pre">CustomReadBarriers</span></tt>, and <tt class="docutils literal"><span class="pre">CustomWriteBarriers</span></tt></a></li>
+<li><a class="reference internal" href="#generating-safe-points-neededsafepoints" id="id26">Generating safe points: <tt class="docutils literal"><span class="pre">NeededSafePoints</span></tt></a></li>
+<li><a class="reference internal" href="#emitting-assembly-code-gcmetadataprinter" id="id27">Emitting assembly code: <tt class="docutils literal"><span class="pre">GCMetadataPrinter</span></tt></a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#references" id="id28">References</a></li>
+</ul>
+</div>
+<div class="section" id="abstract">
+<h2><a class="toc-backref" href="#id1">Abstract</a><a class="headerlink" href="#abstract" title="Permalink to this headline">¶</a></h2>
+<p>This document covers how to integrate LLVM into a compiler for a language which
+supports garbage collection. <strong>Note that LLVM itself does not provide a
+garbage collector.</strong> You must provide your own.</p>
+</div>
+<div class="section" id="quick-start">
+<h2><a class="toc-backref" href="#id2">Quick Start</a><a class="headerlink" href="#quick-start" title="Permalink to this headline">¶</a></h2>
+<p>First, you should pick a collector strategy. LLVM includes a number of built
+in ones, but you can also implement a loadable plugin with a custom definition.
+Note that the collector strategy is a description of how LLVM should generate
+code such that it interacts with your collector and runtime, not a description
+of the collector itself.</p>
+<p>Next, mark your generated functions as using your chosen collector strategy.
+From c++, you can call:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">F</span><span class="p">.</span><span class="n">setGC</span><span class="p">(</span><span class="o"><</span><span class="n">collector</span> <span class="n">description</span> <span class="n">name</span><span class="o">></span><span class="p">);</span>
+</pre></div>
+</div>
+<p>This will produce IR like the following fragment:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">define</span> <span class="kt">void</span> <span class="vg">@foo</span><span class="p">()</span> <span class="k">gc</span> <span class="s">"<collector description name>"</span> <span class="p">{</span> <span class="p">...</span> <span class="p">}</span>
+</pre></div>
+</div>
+<p>When generating LLVM IR for your functions, you will need to:</p>
+<ul class="simple">
+<li>Use <tt class="docutils literal"><span class="pre">@llvm.gcread</span></tt> and/or <tt class="docutils literal"><span class="pre">@llvm.gcwrite</span></tt> in place of standard load and
+store instructions. These intrinsics are used to represent load and store
+barriers. If you collector does not require such barriers, you can skip
+this step.</li>
+<li>Use the memory allocation routines provided by your garbage collector’s
+runtime library.</li>
+<li>If your collector requires them, generate type maps according to your
+runtime’s binary interface. LLVM is not involved in the process. In
+particular, the LLVM type system is not suitable for conveying such
+information though the compiler.</li>
+<li>Insert any coordination code required for interacting with your collector.
+Many collectors require running application code to periodically check a
+flag and conditionally call a runtime function. This is often referred to
+as a safepoint poll.</li>
+</ul>
+<p>You will need to identify roots (i.e. references to heap objects your collector
+needs to know about) in your generated IR, so that LLVM can encode them into
+your final stack maps. Depending on the collector strategy chosen, this is
+accomplished by using either the <tt class="docutils literal"><span class="pre">@llvm.gcroot</span></tt> intrinsics or an
+<tt class="docutils literal"><span class="pre">gc.statepoint</span></tt> relocation sequence.</p>
+<p>Don’t forget to create a root for each intermediate value that is generated when
+evaluating an expression. In <tt class="docutils literal"><span class="pre">h(f(),</span> <span class="pre">g())</span></tt>, the result of <tt class="docutils literal"><span class="pre">f()</span></tt> could
+easily be collected if evaluating <tt class="docutils literal"><span class="pre">g()</span></tt> triggers a collection.</p>
+<p>Finally, you need to link your runtime library with the generated program
+executable (for a static compiler) or ensure the appropriate symbols are
+available for the runtime linker (for a JIT compiler).</p>
+</div>
+<div class="section" id="introduction">
+<h2><a class="toc-backref" href="#id3">Introduction</a><a class="headerlink" href="#introduction" title="Permalink to this headline">¶</a></h2>
+<div class="section" id="what-is-garbage-collection">
+<h3><a class="toc-backref" href="#id4">What is Garbage Collection?</a><a class="headerlink" href="#what-is-garbage-collection" title="Permalink to this headline">¶</a></h3>
+<p>Garbage collection is a widely used technique that frees the programmer from
+having to know the lifetimes of heap objects, making software easier to produce
+and maintain. Many programming languages rely on garbage collection for
+automatic memory management. There are two primary forms of garbage collection:
+conservative and accurate.</p>
+<p>Conservative garbage collection often does not require any special support from
+either the language or the compiler: it can handle non-type-safe programming
+languages (such as C/C++) and does not require any special information from the
+compiler. The <a class="reference external" href="http://www.hpl.hp.com/personal/Hans_Boehm/gc/">Boehm collector</a> is an example of a
+state-of-the-art conservative collector.</p>
+<p>Accurate garbage collection requires the ability to identify all pointers in the
+program at run-time (which requires that the source-language be type-safe in
+most cases). Identifying pointers at run-time requires compiler support to
+locate all places that hold live pointer variables at run-time, including the
+<a class="reference internal" href="#gcroot"><em>processor stack and registers</em></a>.</p>
+<p>Conservative garbage collection is attractive because it does not require any
+special compiler support, but it does have problems. In particular, because the
+conservative garbage collector cannot <em>know</em> that a particular word in the
+machine is a pointer, it cannot move live objects in the heap (preventing the
+use of compacting and generational GC algorithms) and it can occasionally suffer
+from memory leaks due to integer values that happen to point to objects in the
+program. In addition, some aggressive compiler transformations can break
+conservative garbage collectors (though these seem rare in practice).</p>
+<p>Accurate garbage collectors do not suffer from any of these problems, but they
+can suffer from degraded scalar optimization of the program. In particular,
+because the runtime must be able to identify and update all pointers active in
+the program, some optimizations are less effective. In practice, however, the
+locality and performance benefits of using aggressive garbage collection
+techniques dominates any low-level losses.</p>
+<p>This document describes the mechanisms and interfaces provided by LLVM to
+support accurate garbage collection.</p>
+</div>
+<div class="section" id="goals-and-non-goals">
+<h3><a class="toc-backref" href="#id5">Goals and non-goals</a><a class="headerlink" href="#goals-and-non-goals" title="Permalink to this headline">¶</a></h3>
+<p>LLVM’s intermediate representation provides <a class="reference internal" href="#gc-intrinsics"><em>garbage collection intrinsics</em></a> that offer support for a broad class of collector models. For
+instance, the intrinsics permit:</p>
+<ul class="simple">
+<li>semi-space collectors</li>
+<li>mark-sweep collectors</li>
+<li>generational collectors</li>
+<li>incremental collectors</li>
+<li>concurrent collectors</li>
+<li>cooperative collectors</li>
+<li>reference counting</li>
+</ul>
+<p>We hope that the support built into the LLVM IR is sufficient to support a
+broad class of garbage collected languages including Scheme, ML, Java, C#,
+Perl, Python, Lua, Ruby, other scripting languages, and more.</p>
+<p>Note that LLVM <strong>does not itself provide a garbage collector</strong> — this should
+be part of your language’s runtime library. LLVM provides a framework for
+describing the garbage collectors requirements to the compiler. In particular,
+LLVM provides support for generating stack maps at call sites, polling for a
+safepoint, and emitting load and store barriers. You can also extend LLVM -
+possibly through a loadable <a class="reference internal" href="#plugin"><em>code generation plugins</em></a> - to
+generate code and data structures which conforms to the <em>binary interface</em>
+specified by the <em>runtime library</em>. This is similar to the relationship between
+LLVM and DWARF debugging info, for example. The difference primarily lies in
+the lack of an established standard in the domain of garbage collection — thus
+the need for a flexible extension mechanism.</p>
+<p>The aspects of the binary interface with which LLVM’s GC support is
+concerned are:</p>
+<ul class="simple">
+<li>Creation of GC safepoints within code where collection is allowed to execute
+safely.</li>
+<li>Computation of the stack map. For each safe point in the code, object
+references within the stack frame must be identified so that the collector may
+traverse and perhaps update them.</li>
+<li>Write barriers when storing object references to the heap. These are commonly
+used to optimize incremental scans in generational collectors.</li>
+<li>Emission of read barriers when loading object references. These are useful
+for interoperating with concurrent collectors.</li>
+</ul>
+<p>There are additional areas that LLVM does not directly address:</p>
+<ul class="simple">
+<li>Registration of global roots with the runtime.</li>
+<li>Registration of stack map entries with the runtime.</li>
+<li>The functions used by the program to allocate memory, trigger a collection,
+etc.</li>
+<li>Computation or compilation of type maps, or registration of them with the
+runtime. These are used to crawl the heap for object references.</li>
+</ul>
+<p>In general, LLVM’s support for GC does not include features which can be
+adequately addressed with other features of the IR and does not specify a
+particular binary interface. On the plus side, this means that you should be
+able to integrate LLVM with an existing runtime. On the other hand, it can
+have the effect of leaving a lot of work for the developer of a novel
+language. We try to mitigate this by providing built in collector strategy
+descriptions that can work with many common collector designs and easy
+extension points. If you don’t already have a specific binary interface
+you need to support, we recommend trying to use one of these built in collector
+strategies.</p>
+</div>
+</div>
+<div class="section" id="llvm-ir-features">
+<span id="gc-intrinsics"></span><h2><a class="toc-backref" href="#id6">LLVM IR Features</a><a class="headerlink" href="#llvm-ir-features" title="Permalink to this headline">¶</a></h2>
+<p>This section describes the garbage collection facilities provided by the
+<a class="reference internal" href="LangRef.html"><em>LLVM intermediate representation</em></a>. The exact behavior of these
+IR features is specified by the selected <a class="reference internal" href="#plugin"><em>GC strategy description</em></a>.</p>
+<div class="section" id="specifying-gc-code-generation-gc">
+<h3><a class="toc-backref" href="#id7">Specifying GC code generation: <tt class="docutils literal"><span class="pre">gc</span> <span class="pre">"..."</span></tt></a><a class="headerlink" href="#specifying-gc-code-generation-gc" title="Permalink to this headline">¶</a></h3>
+<div class="highlight-text"><div class="highlight"><pre>define <returntype> @name(...) gc "name" { ... }
+</pre></div>
+</div>
+<p>The <tt class="docutils literal"><span class="pre">gc</span></tt> function attribute is used to specify the desired GC strategy to the
+compiler. Its programmatic equivalent is the <tt class="docutils literal"><span class="pre">setGC</span></tt> method of <tt class="docutils literal"><span class="pre">Function</span></tt>.</p>
+<p>Setting <tt class="docutils literal"><span class="pre">gc</span> <span class="pre">"name"</span></tt> on a function triggers a search for a matching subclass
+of GCStrategy. Some collector strategies are built in. You can add others
+using either the loadable plugin mechanism, or by patching your copy of LLVM.
+It is the selected GC strategy which defines the exact nature of the code
+generated to support GC. If none is found, the compiler will raise an error.</p>
+<p>Specifying the GC style on a per-function basis allows LLVM to link together
+programs that use different garbage collection algorithms (or none at all).</p>
+</div>
+<div class="section" id="gcroot">
+<span id="identifying-gc-roots-on-the-stack"></span><h3><a class="toc-backref" href="#id8">Identifying GC roots on the stack</a><a class="headerlink" href="#gcroot" title="Permalink to this headline">¶</a></h3>
+<p>LLVM currently supports two different mechanisms for describing references in
+compiled code at safepoints. <tt class="docutils literal"><span class="pre">llvm.gcroot</span></tt> is the older mechanism;
+<tt class="docutils literal"><span class="pre">gc.statepoint</span></tt> has been added more recently. At the moment, you can choose
+either implementation (on a per <a class="reference internal" href="#plugin"><em>GC strategy</em></a> basis). Longer
+term, we will probably either migrate away from <tt class="docutils literal"><span class="pre">llvm.gcroot</span></tt> entirely, or
+substantially merge their implementations. Note that most new development
+work is focused on <tt class="docutils literal"><span class="pre">gc.statepoint</span></tt>.</p>
+<div class="section" id="using-gc-statepoint">
+<h4><a class="toc-backref" href="#id9">Using <tt class="docutils literal"><span class="pre">gc.statepoint</span></tt></a><a class="headerlink" href="#using-gc-statepoint" title="Permalink to this headline">¶</a></h4>
+<p><a class="reference internal" href="Statepoints.html"><em>This page</em></a> contains detailed documentation for
+<tt class="docutils literal"><span class="pre">gc.statepoint</span></tt>.</p>
+</div>
+<div class="section" id="using-llvm-gcwrite">
+<h4><a class="toc-backref" href="#id10">Using <tt class="docutils literal"><span class="pre">llvm.gcwrite</span></tt></a><a class="headerlink" href="#using-llvm-gcwrite" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="kt">void</span> <span class="vg">@llvm.gcroot</span><span class="p">(</span><span class="k">i8</span><span class="p">**</span> <span class="nv">%ptrloc</span><span class="p">,</span> <span class="k">i8</span><span class="p">*</span> <span class="nv">%metadata</span><span class="p">)</span>
+</pre></div>
+</div>
+<p>The <tt class="docutils literal"><span class="pre">llvm.gcroot</span></tt> intrinsic is used to inform LLVM that a stack variable
+references an object on the heap and is to be tracked for garbage collection.
+The exact impact on generated code is specified by the Function’s selected
+<a class="reference internal" href="#plugin"><em>GC strategy</em></a>. All calls to <tt class="docutils literal"><span class="pre">llvm.gcroot</span></tt> <strong>must</strong> reside
+inside the first basic block.</p>
+<p>The first argument <strong>must</strong> be a value referring to an alloca instruction or a
+bitcast of an alloca. The second contains a pointer to metadata that should be
+associated with the pointer, and <strong>must</strong> be a constant or global value
+address. If your target collector uses tags, use a null pointer for metadata.</p>
+<p>A compiler which performs manual SSA construction <strong>must</strong> ensure that SSA
+values representing GC references are stored in to the alloca passed to the
+respective <tt class="docutils literal"><span class="pre">gcroot</span></tt> before every call site and reloaded after every call.
+A compiler which uses mem2reg to raise imperative code using <tt class="docutils literal"><span class="pre">alloca</span></tt> into
+SSA form need only add a call to <tt class="docutils literal"><span class="pre">@llvm.gcroot</span></tt> for those variables which
+are pointers into the GC heap.</p>
+<p>It is also important to mark intermediate values with <tt class="docutils literal"><span class="pre">llvm.gcroot</span></tt>. For
+example, consider <tt class="docutils literal"><span class="pre">h(f(),</span> <span class="pre">g())</span></tt>. Beware leaking the result of <tt class="docutils literal"><span class="pre">f()</span></tt> in the
+case that <tt class="docutils literal"><span class="pre">g()</span></tt> triggers a collection. Note, that stack variables must be
+initialized and marked with <tt class="docutils literal"><span class="pre">llvm.gcroot</span></tt> in function’s prologue.</p>
+<p>The <tt class="docutils literal"><span class="pre">%metadata</span></tt> argument can be used to avoid requiring heap objects to have
+‘isa’ pointers or tag bits. [<a class="reference internal" href="#appel89">Appel89</a>, <a class="reference internal" href="#goldberg91">Goldberg91</a>, <a class="reference internal" href="#tolmach94">Tolmach94</a>] If specified,
+its value will be tracked along with the location of the pointer in the stack
+frame.</p>
+<p>Consider the following fragment of Java code:</p>
+<div class="highlight-java"><div class="highlight"><pre><span class="o">{</span>
+ <span class="n">Object</span> <span class="n">X</span><span class="o">;</span> <span class="c1">// A null-initialized reference to an object</span>
+ <span class="o">...</span>
+<span class="o">}</span>
+</pre></div>
+</div>
+<p>This block (which may be located in the middle of a function or in a loop nest),
+could be compiled to this LLVM code:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="nl">Entry:</span>
+ <span class="c">;; In the entry block for the function, allocate the</span>
+ <span class="c">;; stack space for X, which is an LLVM pointer.</span>
+ <span class="nv">%X</span> <span class="p">=</span> <span class="k">alloca</span> <span class="nv">%Object</span><span class="p">*</span>
+
+ <span class="c">;; Tell LLVM that the stack space is a stack root.</span>
+ <span class="c">;; Java has type-tags on objects, so we pass null as metadata.</span>
+ <span class="nv">%tmp</span> <span class="p">=</span> <span class="k">bitcast</span> <span class="nv">%Object</span><span class="p">**</span> <span class="nv">%X</span> <span class="k">to</span> <span class="k">i8</span><span class="p">**</span>
+ <span class="k">call</span> <span class="kt">void</span> <span class="vg">@llvm.gcroot</span><span class="p">(</span><span class="k">i8</span><span class="p">**</span> <span class="nv">%tmp</span><span class="p">,</span> <span class="k">i8</span><span class="p">*</span> <span class="k">null</span><span class="p">)</span>
+ <span class="p">...</span>
+
+ <span class="c">;; "CodeBlock" is the block corresponding to the start</span>
+ <span class="c">;; of the scope above.</span>
+<span class="nl">CodeBlock:</span>
+ <span class="c">;; Java null-initializes pointers.</span>
+ <span class="k">store</span> <span class="nv">%Object</span><span class="p">*</span> <span class="k">null</span><span class="p">,</span> <span class="nv">%Object</span><span class="p">**</span> <span class="nv">%X</span>
+
+ <span class="p">...</span>
+
+ <span class="c">;; As the pointer goes out of scope, store a null value into</span>
+ <span class="c">;; it, to indicate that the value is no longer live.</span>
+ <span class="k">store</span> <span class="nv">%Object</span><span class="p">*</span> <span class="k">null</span><span class="p">,</span> <span class="nv">%Object</span><span class="p">**</span> <span class="nv">%X</span>
+ <span class="p">...</span>
+</pre></div>
+</div>
+</div>
+</div>
+<div class="section" id="reading-and-writing-references-in-the-heap">
+<h3><a class="toc-backref" href="#id11">Reading and writing references in the heap</a><a class="headerlink" href="#reading-and-writing-references-in-the-heap" title="Permalink to this headline">¶</a></h3>
+<p>Some collectors need to be informed when the mutator (the program that needs
+garbage collection) either reads a pointer from or writes a pointer to a field
+of a heap object. The code fragments inserted at these points are called <em>read
+barriers</em> and <em>write barriers</em>, respectively. The amount of code that needs to
+be executed is usually quite small and not on the critical path of any
+computation, so the overall performance impact of the barrier is tolerable.</p>
+<p>Barriers often require access to the <em>object pointer</em> rather than the <em>derived
+pointer</em> (which is a pointer to the field within the object). Accordingly,
+these intrinsics take both pointers as separate arguments for completeness. In
+this snippet, <tt class="docutils literal"><span class="pre">%object</span></tt> is the object pointer, and <tt class="docutils literal"><span class="pre">%derived</span></tt> is the derived
+pointer:</p>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="c">;; An array type.</span>
+<span class="nv">%class.Array</span> <span class="p">=</span> <span class="k">type</span> <span class="p">{</span> <span class="nv">%class.Object</span><span class="p">,</span> <span class="k">i32</span><span class="p">,</span> <span class="p">[</span><span class="m">0</span> <span class="k">x</span> <span class="nv">%class.Object</span><span class="p">*]</span> <span class="p">}</span>
+<span class="p">...</span>
+
+<span class="c">;; Load the object pointer from a gcroot.</span>
+<span class="nv">%object</span> <span class="p">=</span> <span class="k">load</span> <span class="nv">%class.Array</span><span class="p">**</span> <span class="nv">%object_addr</span>
+
+<span class="c">;; Compute the derived pointer.</span>
+<span class="nv">%derived</span> <span class="p">=</span> <span class="k">getelementptr</span> <span class="nv">%object</span><span class="p">,</span> <span class="k">i32</span> <span class="m">0</span><span class="p">,</span> <span class="k">i32</span> <span class="m">2</span><span class="p">,</span> <span class="k">i32</span> <span class="nv">%n</span>
+</pre></div>
+</div>
+<p>LLVM does not enforce this relationship between the object and derived pointer
+(although a particular <a class="reference internal" href="#plugin"><em>collector strategy</em></a> might). However, it
+would be an unusual collector that violated it.</p>
+<p>The use of these intrinsics is naturally optional if the target GC does not
+require the corresponding barrier. The GC strategy used with such a collector
+should replace the intrinsic calls with the corresponding <tt class="docutils literal"><span class="pre">load</span></tt> or
+<tt class="docutils literal"><span class="pre">store</span></tt> instruction if they are used.</p>
+<p>One known deficiency with the current design is that the barrier intrinsics do
+not include the size or alignment of the underlying operation performed. It is
+currently assumed that the operation is of pointer size and the alignment is
+assumed to be the target machine’s default alignment.</p>
+<div class="section" id="write-barrier-llvm-gcwrite">
+<h4><a class="toc-backref" href="#id12">Write barrier: <tt class="docutils literal"><span class="pre">llvm.gcwrite</span></tt></a><a class="headerlink" href="#write-barrier-llvm-gcwrite" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="kt">void</span> <span class="vg">@llvm.gcwrite</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%value</span><span class="p">,</span> <span class="k">i8</span><span class="p">*</span> <span class="nv">%object</span><span class="p">,</span> <span class="k">i8</span><span class="p">**</span> <span class="nv">%derived</span><span class="p">)</span>
+</pre></div>
+</div>
+<p>For write barriers, LLVM provides the <tt class="docutils literal"><span class="pre">llvm.gcwrite</span></tt> intrinsic function. It
+has exactly the same semantics as a non-volatile <tt class="docutils literal"><span class="pre">store</span></tt> to the derived
+pointer (the third argument). The exact code generated is specified by the
+Function’s selected <a class="reference internal" href="#plugin"><em>GC strategy</em></a>.</p>
+<p>Many important algorithms require write barriers, including generational and
+concurrent collectors. Additionally, write barriers could be used to implement
+reference counting.</p>
+</div>
+<div class="section" id="read-barrier-llvm-gcread">
+<h4><a class="toc-backref" href="#id13">Read barrier: <tt class="docutils literal"><span class="pre">llvm.gcread</span></tt></a><a class="headerlink" href="#read-barrier-llvm-gcread" title="Permalink to this headline">¶</a></h4>
+<div class="highlight-llvm"><div class="highlight"><pre><span class="k">i8</span><span class="p">*</span> <span class="vg">@llvm.gcread</span><span class="p">(</span><span class="k">i8</span><span class="p">*</span> <span class="nv">%object</span><span class="p">,</span> <span class="k">i8</span><span class="p">**</span> <span class="nv">%derived</span><span class="p">)</span>
+</pre></div>
+</div>
+<p>For read barriers, LLVM provides the <tt class="docutils literal"><span class="pre">llvm.gcread</span></tt> intrinsic function. It has
+exactly the same semantics as a non-volatile <tt class="docutils literal"><span class="pre">load</span></tt> from the derived pointer
+(the second argument). The exact code generated is specified by the Function’s
+selected <a class="reference internal" href="#plugin"><em>GC strategy</em></a>.</p>
+<p>Read barriers are needed by fewer algorithms than write barriers, and may have a
+greater performance impact since pointer reads are more frequent than writes.</p>
+</div>
+</div>
+</div>
+<div class="section" id="built-in-gc-strategies">
+<span id="builtin-gc-strategies"></span><span id="plugin"></span><h2><a class="toc-backref" href="#id14">Built In GC Strategies</a><a class="headerlink" href="#built-in-gc-strategies" title="Permalink to this headline">¶</a></h2>
+<p>LLVM includes built in support for several varieties of garbage collectors.</p>
+<div class="section" id="the-shadow-stack-gc">
+<h3><a class="toc-backref" href="#id15">The Shadow Stack GC</a><a class="headerlink" href="#the-shadow-stack-gc" title="Permalink to this headline">¶</a></h3>
+<p>To use this collector strategy, mark your functions with:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">F</span><span class="p">.</span><span class="n">setGC</span><span class="p">(</span><span class="s">"shadow-stack"</span><span class="p">);</span>
+</pre></div>
+</div>
+<p>Unlike many GC algorithms which rely on a cooperative code generator to compile
+stack maps, this algorithm carefully maintains a linked list of stack roots
+[<a class="reference internal" href="#henderson02"><em>Henderson2002</em></a>]. This so-called “shadow stack” mirrors the
+machine stack. Maintaining this data structure is slower than using a stack map
+compiled into the executable as constant data, but has a significant portability
+advantage because it requires no special support from the target code generator,
+and does not require tricky platform-specific code to crawl the machine stack.</p>
+<p>The tradeoff for this simplicity and portability is:</p>
+<ul class="simple">
+<li>High overhead per function call.</li>
+<li>Not thread-safe.</li>
+</ul>
+<p>Still, it’s an easy way to get started. After your compiler and runtime are up
+and running, writing a <a class="reference internal" href="#plugin"><em>plugin</em></a> will allow you to take advantage
+of <a class="reference internal" href="#collector-algos"><em>more advanced GC features</em></a> of LLVM in order to
+improve performance.</p>
+<p>The shadow stack doesn’t imply a memory allocation algorithm. A semispace
+collector or building atop <tt class="docutils literal"><span class="pre">malloc</span></tt> are great places to start, and can be
+implemented with very little code.</p>
+<p>When it comes time to collect, however, your runtime needs to traverse the stack
+roots, and for this it needs to integrate with the shadow stack. Luckily, doing
+so is very simple. (This code is heavily commented to help you understand the
+data structure, but there are only 20 lines of meaningful code.)</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="c1">/// @brief The map for a single function's stack frame. One of these is</span>
+<span class="c1">/// compiled as constant data into the executable for each function.</span>
+<span class="c1">///</span>
+<span class="c1">/// Storage of metadata values is elided if the %metadata parameter to</span>
+<span class="c1">/// @llvm.gcroot is null.</span>
+<span class="k">struct</span> <span class="n">FrameMap</span> <span class="p">{</span>
+ <span class="kt">int32_t</span> <span class="n">NumRoots</span><span class="p">;</span> <span class="c1">//< Number of roots in stack frame.</span>
+ <span class="kt">int32_t</span> <span class="n">NumMeta</span><span class="p">;</span> <span class="c1">//< Number of metadata entries. May be < NumRoots.</span>
+ <span class="k">const</span> <span class="kt">void</span> <span class="o">*</span><span class="n">Meta</span><span class="p">[</span><span class="mi">0</span><span class="p">];</span> <span class="c1">//< Metadata for each root.</span>
+<span class="p">};</span>
+
+<span class="c1">/// @brief A link in the dynamic shadow stack. One of these is embedded in</span>
+<span class="c1">/// the stack frame of each function on the call stack.</span>
+<span class="k">struct</span> <span class="n">StackEntry</span> <span class="p">{</span>
+ <span class="n">StackEntry</span> <span class="o">*</span><span class="n">Next</span><span class="p">;</span> <span class="c1">//< Link to next stack entry (the caller's).</span>
+ <span class="k">const</span> <span class="n">FrameMap</span> <span class="o">*</span><span class="n">Map</span><span class="p">;</span> <span class="c1">//< Pointer to constant FrameMap.</span>
+ <span class="kt">void</span> <span class="o">*</span><span class="n">Roots</span><span class="p">[</span><span class="mi">0</span><span class="p">];</span> <span class="c1">//< Stack roots (in-place array).</span>
+<span class="p">};</span>
+
+<span class="c1">/// @brief The head of the singly-linked list of StackEntries. Functions push</span>
+<span class="c1">/// and pop onto this in their prologue and epilogue.</span>
+<span class="c1">///</span>
+<span class="c1">/// Since there is only a global list, this technique is not threadsafe.</span>
+<span class="n">StackEntry</span> <span class="o">*</span><span class="n">llvm_gc_root_chain</span><span class="p">;</span>
+
+<span class="c1">/// @brief Calls Visitor(root, meta) for each GC root on the stack.</span>
+<span class="c1">/// root and meta are exactly the values passed to</span>
+<span class="c1">/// @llvm.gcroot.</span>
+<span class="c1">///</span>
+<span class="c1">/// Visitor could be a function to recursively mark live objects. Or it</span>
+<span class="c1">/// might copy them to another heap or generation.</span>
+<span class="c1">///</span>
+<span class="c1">/// @param Visitor A function to invoke for every GC root on the stack.</span>
+<span class="kt">void</span> <span class="nf">visitGCRoots</span><span class="p">(</span><span class="kt">void</span> <span class="p">(</span><span class="o">*</span><span class="n">Visitor</span><span class="p">)(</span><span class="kt">void</span> <span class="o">**</span><span class="n">Root</span><span class="p">,</span> <span class="k">const</span> <span class="kt">void</span> <span class="o">*</span><span class="n">Meta</span><span class="p">))</span> <span class="p">{</span>
+ <span class="k">for</span> <span class="p">(</span><span class="n">StackEntry</span> <span class="o">*</span><span class="n">R</span> <span class="o">=</span> <span class="n">llvm_gc_root_chain</span><span class="p">;</span> <span class="n">R</span><span class="p">;</span> <span class="n">R</span> <span class="o">=</span> <span class="n">R</span><span class="o">-></span><span class="n">Next</span><span class="p">)</span> <span class="p">{</span>
+ <span class="kt">unsigned</span> <span class="n">i</span> <span class="o">=</span> <span class="mi">0</span><span class="p">;</span>
+
+ <span class="c1">// For roots [0, NumMeta), the metadata pointer is in the FrameMap.</span>
+ <span class="k">for</span> <span class="p">(</span><span class="kt">unsigned</span> <span class="n">e</span> <span class="o">=</span> <span class="n">R</span><span class="o">-></span><span class="n">Map</span><span class="o">-></span><span class="n">NumMeta</span><span class="p">;</span> <span class="n">i</span> <span class="o">!=</span> <span class="n">e</span><span class="p">;</span> <span class="o">++</span><span class="n">i</span><span class="p">)</span>
+ <span class="n">Visitor</span><span class="p">(</span><span class="o">&</span><span class="n">R</span><span class="o">-></span><span class="n">Roots</span><span class="p">[</span><span class="n">i</span><span class="p">],</span> <span class="n">R</span><span class="o">-></span><span class="n">Map</span><span class="o">-></span><span class="n">Meta</span><span class="p">[</span><span class="n">i</span><span class="p">]);</span>
+
+ <span class="c1">// For roots [NumMeta, NumRoots), the metadata pointer is null.</span>
+ <span class="k">for</span> <span class="p">(</span><span class="kt">unsigned</span> <span class="n">e</span> <span class="o">=</span> <span class="n">R</span><span class="o">-></span><span class="n">Map</span><span class="o">-></span><span class="n">NumRoots</span><span class="p">;</span> <span class="n">i</span> <span class="o">!=</span> <span class="n">e</span><span class="p">;</span> <span class="o">++</span><span class="n">i</span><span class="p">)</span>
+ <span class="n">Visitor</span><span class="p">(</span><span class="o">&</span><span class="n">R</span><span class="o">-></span><span class="n">Roots</span><span class="p">[</span><span class="n">i</span><span class="p">],</span> <span class="nb">NULL</span><span class="p">);</span>
+ <span class="p">}</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+</div>
+<div class="section" id="the-erlang-and-ocaml-gcs">
+<h3><a class="toc-backref" href="#id16">The ‘Erlang’ and ‘Ocaml’ GCs</a><a class="headerlink" href="#the-erlang-and-ocaml-gcs" title="Permalink to this headline">¶</a></h3>
+<p>LLVM ships with two example collectors which leverage the <tt class="docutils literal"><span class="pre">gcroot</span></tt>
+mechanisms. To our knowledge, these are not actually used by any language
+runtime, but they do provide a reasonable starting point for someone interested
+in writing an <tt class="docutils literal"><span class="pre">gcroot</span></tt> compatible GC plugin. In particular, these are the
+only in tree examples of how to produce a custom binary stack map format using
+a <tt class="docutils literal"><span class="pre">gcroot</span></tt> strategy.</p>
+<p>As there names imply, the binary format produced is intended to model that
+used by the Erlang and OCaml compilers respectively.</p>
+</div>
+<div class="section" id="the-statepoint-example-gc">
+<span id="statepoint-example-gc"></span><h3><a class="toc-backref" href="#id17">The Statepoint Example GC</a><a class="headerlink" href="#the-statepoint-example-gc" title="Permalink to this headline">¶</a></h3>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">F</span><span class="p">.</span><span class="n">setGC</span><span class="p">(</span><span class="s">"statepoint-example"</span><span class="p">);</span>
+</pre></div>
+</div>
+<p>This GC provides an example of how one might use the infrastructure provided
+by <tt class="docutils literal"><span class="pre">gc.statepoint</span></tt>. This example GC is compatible with the
+<a class="reference internal" href="Statepoints.html#placesafepoints"><em>PlaceSafepoints</em></a> and <a class="reference internal" href="Statepoints.html#rewritestatepointsforgc"><em>RewriteStatepointsForGC</em></a> utility passes
+which simplify <tt class="docutils literal"><span class="pre">gc.statepoint</span></tt> sequence insertion. If you need to build a
+custom GC strategy around the <tt class="docutils literal"><span class="pre">gc.statepoints</span></tt> mechanisms, it is recommended
+that you use this one as a starting point.</p>
+<p>This GC strategy does not support read or write barriers. As a result, these
+intrinsics are lowered to normal loads and stores.</p>
+<p>The stack map format generated by this GC strategy can be found in the
+<a class="reference internal" href="StackMaps.html#stackmap-section"><em>Stack Map Section</em></a> using a format documented <a class="reference internal" href="Statepoints.html#statepoint-stackmap-format"><em>here</em></a>. This format is intended to be the standard
+format supported by LLVM going forward.</p>
+</div>
+<div class="section" id="the-coreclr-gc">
+<h3><a class="toc-backref" href="#id18">The CoreCLR GC</a><a class="headerlink" href="#the-coreclr-gc" title="Permalink to this headline">¶</a></h3>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">F</span><span class="p">.</span><span class="n">setGC</span><span class="p">(</span><span class="s">"coreclr"</span><span class="p">);</span>
+</pre></div>
+</div>
+<p>This GC leverages the <tt class="docutils literal"><span class="pre">gc.statepoint</span></tt> mechanism to support the
+<a class="reference external" href="https://github.com/dotnet/coreclr">CoreCLR</a> runtime.</p>
+<p>Support for this GC strategy is a work in progress. This strategy will
+differ from
+<a class="reference internal" href="#statepoint-example-gc"><em>statepoint-example GC</em></a> strategy in
+certain aspects like:</p>
+<ul class="simple">
+<li>Base-pointers of interior pointers are not explicitly
+tracked and reported.</li>
+<li>A different format is used for encoding stack maps.</li>
+<li>Safe-point polls are only needed before loop-back edges
+and before tail-calls (not needed at function-entry).</li>
+</ul>
+</div>
+</div>
+<div class="section" id="custom-gc-strategies">
+<h2><a class="toc-backref" href="#id19">Custom GC Strategies</a><a class="headerlink" href="#custom-gc-strategies" title="Permalink to this headline">¶</a></h2>
+<p>If none of the built in GC strategy descriptions met your needs above, you will
+need to define a custom GCStrategy and possibly, a custom LLVM pass to perform
+lowering. Your best example of where to start defining a custom GCStrategy
+would be to look at one of the built in strategies.</p>
+<p>You may be able to structure this additional code as a loadable plugin library.
+Loadable plugins are sufficient if all you need is to enable a different
+combination of built in functionality, but if you need to provide a custom
+lowering pass, you will need to build a patched version of LLVM. If you think
+you need a patched build, please ask for advice on llvm-dev. There may be an
+easy way we can extend the support to make it work for your use case without
+requiring a custom build.</p>
+<div class="section" id="collector-requirements">
+<h3><a class="toc-backref" href="#id20">Collector Requirements</a><a class="headerlink" href="#collector-requirements" title="Permalink to this headline">¶</a></h3>
+<p>You should be able to leverage any existing collector library that includes the following elements:</p>
+<ol class="arabic simple">
+<li>A memory allocator which exposes an allocation function your compiled
+code can call.</li>
+<li>A binary format for the stack map. A stack map describes the location
+of references at a safepoint and is used by precise collectors to identify
+references within a stack frame on the machine stack. Note that collectors
+which conservatively scan the stack don’t require such a structure.</li>
+<li>A stack crawler to discover functions on the call stack, and enumerate the
+references listed in the stack map for each call site.</li>
+<li>A mechanism for identifying references in global locations (e.g. global
+variables).</li>
+<li>If you collector requires them, an LLVM IR implementation of your collectors
+load and store barriers. Note that since many collectors don’t require
+barriers at all, LLVM defaults to lowering such barriers to normal loads
+and stores unless you arrange otherwise.</li>
+</ol>
+</div>
+<div class="section" id="implementing-a-collector-plugin">
+<h3><a class="toc-backref" href="#id21">Implementing a collector plugin</a><a class="headerlink" href="#implementing-a-collector-plugin" title="Permalink to this headline">¶</a></h3>
+<p>User code specifies which GC code generation to use with the <tt class="docutils literal"><span class="pre">gc</span></tt> function
+attribute or, equivalently, with the <tt class="docutils literal"><span class="pre">setGC</span></tt> method of <tt class="docutils literal"><span class="pre">Function</span></tt>.</p>
+<p>To implement a GC plugin, it is necessary to subclass <tt class="docutils literal"><span class="pre">llvm::GCStrategy</span></tt>,
+which can be accomplished in a few lines of boilerplate code. LLVM’s
+infrastructure provides access to several important algorithms. For an
+uncontroversial collector, all that remains may be to compile LLVM’s computed
+stack map to assembly code (using the binary representation expected by the
+runtime library). This can be accomplished in about 100 lines of code.</p>
+<p>This is not the appropriate place to implement a garbage collected heap or a
+garbage collector itself. That code should exist in the language’s runtime
+library. The compiler plugin is responsible for generating code which conforms
+to the binary interface defined by library, most essentially the <a class="reference internal" href="#stack-map"><em>stack map</em></a>.</p>
+<p>To subclass <tt class="docutils literal"><span class="pre">llvm::GCStrategy</span></tt> and register it with the compiler:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="c1">// lib/MyGC/MyGC.cpp - Example LLVM GC plugin</span>
+
+<span class="cp">#include "llvm/CodeGen/GCStrategy.h"</span>
+<span class="cp">#include "llvm/CodeGen/GCMetadata.h"</span>
+<span class="cp">#include "llvm/Support/Compiler.h"</span>
+
+<span class="k">using</span> <span class="k">namespace</span> <span class="n">llvm</span><span class="p">;</span>
+
+<span class="k">namespace</span> <span class="p">{</span>
+ <span class="k">class</span> <span class="nc">LLVM_LIBRARY_VISIBILITY</span> <span class="n">MyGC</span> <span class="o">:</span> <span class="k">public</span> <span class="n">GCStrategy</span> <span class="p">{</span>
+ <span class="nl">public:</span>
+ <span class="n">MyGC</span><span class="p">()</span> <span class="p">{}</span>
+ <span class="p">};</span>
+
+ <span class="n">GCRegistry</span><span class="o">::</span><span class="n">Add</span><span class="o"><</span><span class="n">MyGC</span><span class="o">></span>
+ <span class="n">X</span><span class="p">(</span><span class="s">"mygc"</span><span class="p">,</span> <span class="s">"My bespoke garbage collector."</span><span class="p">);</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>This boilerplate collector does nothing. More specifically:</p>
+<ul class="simple">
+<li><tt class="docutils literal"><span class="pre">llvm.gcread</span></tt> calls are replaced with the corresponding <tt class="docutils literal"><span class="pre">load</span></tt>
+instruction.</li>
+<li><tt class="docutils literal"><span class="pre">llvm.gcwrite</span></tt> calls are replaced with the corresponding <tt class="docutils literal"><span class="pre">store</span></tt>
+instruction.</li>
+<li>No safe points are added to the code.</li>
+<li>The stack map is not compiled into the executable.</li>
+</ul>
+<p>Using the LLVM makefiles, this code
+can be compiled as a plugin using a simple makefile:</p>
+<div class="highlight-make"><div class="highlight"><pre><span class="c"># lib/MyGC/Makefile</span>
+
+<span class="nv">LEVEL</span> <span class="o">:=</span> ../..
+<span class="nv">LIBRARYNAME</span> <span class="o">=</span> MyGC
+<span class="nv">LOADABLE_MODULE</span> <span class="o">=</span> 1
+
+<span class="cp">include $(LEVEL)/Makefile.common</span>
+</pre></div>
+</div>
+<p>Once the plugin is compiled, code using it may be compiled using <tt class="docutils literal"><span class="pre">llc</span>
+<span class="pre">-load=MyGC.so</span></tt> (though MyGC.so may have some other platform-specific
+extension):</p>
+<div class="highlight-python"><div class="highlight"><pre>$ cat sample.ll
+define void @f() gc "mygc" {
+entry:
+ ret void
+}
+$ llvm-as < sample.ll | llc -load=MyGC.so
+</pre></div>
+</div>
+<p>It is also possible to statically link the collector plugin into tools, such as
+a language-specific compiler front-end.</p>
+</div>
+<div class="section" id="overview-of-available-features">
+<span id="collector-algos"></span><h3><a class="toc-backref" href="#id22">Overview of available features</a><a class="headerlink" href="#overview-of-available-features" title="Permalink to this headline">¶</a></h3>
+<p><tt class="docutils literal"><span class="pre">GCStrategy</span></tt> provides a range of features through which a plugin may do useful
+work. Some of these are callbacks, some are algorithms that can be enabled,
+disabled, or customized. This matrix summarizes the supported (and planned)
+features and correlates them with the collection techniques which typically
+require them.</p>
+<table border="1" class="docutils">
+<colgroup>
+<col width="14%" />
+<col width="7%" />
+<col width="9%" />
+<col width="11%" />
+<col width="8%" />
+<col width="10%" />
+<col width="15%" />
+<col width="11%" />
+<col width="14%" />
+</colgroup>
+<thead valign="bottom">
+<tr class="row-odd"><th class="head">Algorithm</th>
+<th class="head">Done</th>
+<th class="head">Shadow
+stack</th>
+<th class="head">refcount</th>
+<th class="head">mark-
+sweep</th>
+<th class="head">copying</th>
+<th class="head">incremental</th>
+<th class="head">threaded</th>
+<th class="head">concurrent</th>
+</tr>
+</thead>
+<tbody valign="top">
+<tr class="row-even"><td>stack map</td>
+<td>â</td>
+<td> </td>
+<td> </td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+</tr>
+<tr class="row-odd"><td>initialize
+roots</td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+</tr>
+<tr class="row-even"><td>derived
+pointers</td>
+<td>NO</td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td><strong>N</strong>*</td>
+<td><strong>N</strong>*</td>
+</tr>
+<tr class="row-odd"><td><strong>custom
+lowering</strong></td>
+<td>â</td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+</tr>
+<tr class="row-even"><td><em>gcroot</em></td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+</tr>
+<tr class="row-odd"><td><em>gcwrite</em></td>
+<td>â</td>
+<td> </td>
+<td>â</td>
+<td> </td>
+<td> </td>
+<td>â</td>
+<td> </td>
+<td>â</td>
+</tr>
+<tr class="row-even"><td><em>gcread</em></td>
+<td>â</td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td>â</td>
+</tr>
+<tr class="row-odd"><td><strong>safe
+points</strong></td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+</tr>
+<tr class="row-even"><td><em>in
+calls</em></td>
+<td>â</td>
+<td> </td>
+<td> </td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+</tr>
+<tr class="row-odd"><td><em>before
+calls</em></td>
+<td>â</td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td>â</td>
+<td>â</td>
+</tr>
+<tr class="row-even"><td><em>for
+loops</em></td>
+<td>NO</td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td><strong>N</strong></td>
+<td><strong>N</strong></td>
+</tr>
+<tr class="row-odd"><td><em>before
+escape</em></td>
+<td>â</td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td>â</td>
+<td>â</td>
+</tr>
+<tr class="row-even"><td>emit code
+at safe
+points</td>
+<td>NO</td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td><strong>N</strong></td>
+<td><strong>N</strong></td>
+</tr>
+<tr class="row-odd"><td><strong>output</strong></td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+<td> </td>
+</tr>
+<tr class="row-even"><td><em>assembly</em></td>
+<td>â</td>
+<td> </td>
+<td> </td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+<td>â</td>
+</tr>
+<tr class="row-odd"><td><em>JIT</em></td>
+<td>NO</td>
+<td> </td>
+<td> </td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+</tr>
+<tr class="row-even"><td><em>obj</em></td>
+<td>NO</td>
+<td> </td>
+<td> </td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+</tr>
+<tr class="row-odd"><td>live
+analysis</td>
+<td>NO</td>
+<td> </td>
+<td> </td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+</tr>
+<tr class="row-even"><td>register
+map</td>
+<td>NO</td>
+<td> </td>
+<td> </td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+<td><strong>?</strong></td>
+</tr>
+<tr class="row-odd"><td colspan="9">* Derived pointers only pose a hasard to copying collections.</td>
+</tr>
+<tr class="row-even"><td colspan="9"><strong>?</strong> denotes a feature which could be utilized if available.</td>
+</tr>
+</tbody>
+</table>
+<p>To be clear, the collection techniques above are defined as:</p>
+<dl class="docutils">
+<dt>Shadow Stack</dt>
+<dd>The mutator carefully maintains a linked list of stack roots.</dd>
+<dt>Reference Counting</dt>
+<dd>The mutator maintains a reference count for each object and frees an object
+when its count falls to zero.</dd>
+<dt>Mark-Sweep</dt>
+<dd>When the heap is exhausted, the collector marks reachable objects starting
+from the roots, then deallocates unreachable objects in a sweep phase.</dd>
+<dt>Copying</dt>
+<dd>As reachability analysis proceeds, the collector copies objects from one heap
+area to another, compacting them in the process. Copying collectors enable
+highly efficient “bump pointer” allocation and can improve locality of
+reference.</dd>
+<dt>Incremental</dt>
+<dd>(Including generational collectors.) Incremental collectors generally have all
+the properties of a copying collector (regardless of whether the mature heap
+is compacting), but bring the added complexity of requiring write barriers.</dd>
+<dt>Threaded</dt>
+<dd>Denotes a multithreaded mutator; the collector must still stop the mutator
+(“stop the world”) before beginning reachability analysis. Stopping a
+multithreaded mutator is a complicated problem. It generally requires highly
+platform-specific code in the runtime, and the production of carefully
+designed machine code at safe points.</dd>
+<dt>Concurrent</dt>
+<dd>In this technique, the mutator and the collector run concurrently, with the
+goal of eliminating pause times. In a <em>cooperative</em> collector, the mutator
+further aids with collection should a pause occur, allowing collection to take
+advantage of multiprocessor hosts. The “stop the world” problem of threaded
+collectors is generally still present to a limited extent. Sophisticated
+marking algorithms are necessary. Read barriers may be necessary.</dd>
+</dl>
+<p>As the matrix indicates, LLVM’s garbage collection infrastructure is already
+suitable for a wide variety of collectors, but does not currently extend to
+multithreaded programs. This will be added in the future as there is
+interest.</p>
+</div>
+<div class="section" id="computing-stack-maps">
+<span id="stack-map"></span><h3><a class="toc-backref" href="#id23">Computing stack maps</a><a class="headerlink" href="#computing-stack-maps" title="Permalink to this headline">¶</a></h3>
+<p>LLVM automatically computes a stack map. One of the most important features
+of a <tt class="docutils literal"><span class="pre">GCStrategy</span></tt> is to compile this information into the executable in
+the binary representation expected by the runtime library.</p>
+<p>The stack map consists of the location and identity of each GC root in the
+each function in the module. For each root:</p>
+<ul class="simple">
+<li><tt class="docutils literal"><span class="pre">RootNum</span></tt>: The index of the root.</li>
+<li><tt class="docutils literal"><span class="pre">StackOffset</span></tt>: The offset of the object relative to the frame pointer.</li>
+<li><tt class="docutils literal"><span class="pre">RootMetadata</span></tt>: The value passed as the <tt class="docutils literal"><span class="pre">%metadata</span></tt> parameter to the
+<tt class="docutils literal"><span class="pre">@llvm.gcroot</span></tt> intrinsic.</li>
+</ul>
+<p>Also, for the function as a whole:</p>
+<ul>
+<li><dl class="first docutils">
+<dt><tt class="docutils literal"><span class="pre">getFrameSize()</span></tt>: The overall size of the function’s initial stack frame,</dt>
+<dd><p class="first last">not accounting for any dynamic allocation.</p>
+</dd>
+</dl>
+</li>
+<li><p class="first"><tt class="docutils literal"><span class="pre">roots_size()</span></tt>: The count of roots in the function.</p>
+</li>
+</ul>
+<p>To access the stack map, use <tt class="docutils literal"><span class="pre">GCFunctionMetadata::roots_begin()</span></tt> and
+-<tt class="docutils literal"><span class="pre">end()</span></tt> from the <a class="reference internal" href="#assembly"><em>GCMetadataPrinter</em></a>:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">for</span> <span class="p">(</span><span class="n">iterator</span> <span class="n">I</span> <span class="o">=</span> <span class="n">begin</span><span class="p">(),</span> <span class="n">E</span> <span class="o">=</span> <span class="n">end</span><span class="p">();</span> <span class="n">I</span> <span class="o">!=</span> <span class="n">E</span><span class="p">;</span> <span class="o">++</span><span class="n">I</span><span class="p">)</span> <span class="p">{</span>
+ <span class="n">GCFunctionInfo</span> <span class="o">*</span><span class="n">FI</span> <span class="o">=</span> <span class="o">*</span><span class="n">I</span><span class="p">;</span>
+ <span class="kt">unsigned</span> <span class="n">FrameSize</span> <span class="o">=</span> <span class="n">FI</span><span class="o">-></span><span class="n">getFrameSize</span><span class="p">();</span>
+ <span class="kt">size_t</span> <span class="n">RootCount</span> <span class="o">=</span> <span class="n">FI</span><span class="o">-></span><span class="n">roots_size</span><span class="p">();</span>
+
+ <span class="k">for</span> <span class="p">(</span><span class="n">GCFunctionInfo</span><span class="o">::</span><span class="n">roots_iterator</span> <span class="n">RI</span> <span class="o">=</span> <span class="n">FI</span><span class="o">-></span><span class="n">roots_begin</span><span class="p">(),</span>
+ <span class="n">RE</span> <span class="o">=</span> <span class="n">FI</span><span class="o">-></span><span class="n">roots_end</span><span class="p">();</span>
+ <span class="n">RI</span> <span class="o">!=</span> <span class="n">RE</span><span class="p">;</span> <span class="o">++</span><span class="n">RI</span><span class="p">)</span> <span class="p">{</span>
+ <span class="kt">int</span> <span class="n">RootNum</span> <span class="o">=</span> <span class="n">RI</span><span class="o">-></span><span class="n">Num</span><span class="p">;</span>
+ <span class="kt">int</span> <span class="n">RootStackOffset</span> <span class="o">=</span> <span class="n">RI</span><span class="o">-></span><span class="n">StackOffset</span><span class="p">;</span>
+ <span class="n">Constant</span> <span class="o">*</span><span class="n">RootMetadata</span> <span class="o">=</span> <span class="n">RI</span><span class="o">-></span><span class="n">Metadata</span><span class="p">;</span>
+ <span class="p">}</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>If the <tt class="docutils literal"><span class="pre">llvm.gcroot</span></tt> intrinsic is eliminated before code generation by a
+custom lowering pass, LLVM will compute an empty stack map. This may be useful
+for collector plugins which implement reference counting or a shadow stack.</p>
+</div>
+<div class="section" id="initializing-roots-to-null-initroots">
+<span id="init-roots"></span><h3><a class="toc-backref" href="#id24">Initializing roots to null: <tt class="docutils literal"><span class="pre">InitRoots</span></tt></a><a class="headerlink" href="#initializing-roots-to-null-initroots" title="Permalink to this headline">¶</a></h3>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">MyGC</span><span class="o">::</span><span class="n">MyGC</span><span class="p">()</span> <span class="p">{</span>
+ <span class="n">InitRoots</span> <span class="o">=</span> <span class="nb">true</span><span class="p">;</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>When set, LLVM will automatically initialize each root to <tt class="docutils literal"><span class="pre">null</span></tt> upon entry to
+the function. This prevents the GC’s sweep phase from visiting uninitialized
+pointers, which will almost certainly cause it to crash. This initialization
+occurs before custom lowering, so the two may be used together.</p>
+<p>Since LLVM does not yet compute liveness information, there is no means of
+distinguishing an uninitialized stack root from an initialized one. Therefore,
+this feature should be used by all GC plugins. It is enabled by default.</p>
+</div>
+<div class="section" id="custom-lowering-of-intrinsics-customroots-customreadbarriers-and-customwritebarriers">
+<h3><a class="toc-backref" href="#id25">Custom lowering of intrinsics: <tt class="docutils literal"><span class="pre">CustomRoots</span></tt>, <tt class="docutils literal"><span class="pre">CustomReadBarriers</span></tt>, and <tt class="docutils literal"><span class="pre">CustomWriteBarriers</span></tt></a><a class="headerlink" href="#custom-lowering-of-intrinsics-customroots-customreadbarriers-and-customwritebarriers" title="Permalink to this headline">¶</a></h3>
+<p>For GCs which use barriers or unusual treatment of stack roots, these
+flags allow the collector to perform arbitrary transformations of the
+LLVM IR:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">class</span> <span class="nc">MyGC</span> <span class="o">:</span> <span class="k">public</span> <span class="n">GCStrategy</span> <span class="p">{</span>
+<span class="nl">public:</span>
+ <span class="n">MyGC</span><span class="p">()</span> <span class="p">{</span>
+ <span class="n">CustomRoots</span> <span class="o">=</span> <span class="nb">true</span><span class="p">;</span>
+ <span class="n">CustomReadBarriers</span> <span class="o">=</span> <span class="nb">true</span><span class="p">;</span>
+ <span class="n">CustomWriteBarriers</span> <span class="o">=</span> <span class="nb">true</span><span class="p">;</span>
+ <span class="p">}</span>
+<span class="p">};</span>
+</pre></div>
+</div>
+<p>If any of these flags are set, LLVM suppresses its default lowering for
+the corresponding intrinsics. Instead, you must provide a custom Pass
+which lowers the intrinsics as desired. If you have opted in to custom
+lowering of a particular intrinsic your pass <strong>must</strong> eliminate all
+instances of the corresponding intrinsic in functions which opt in to
+your GC. The best example of such a pass is the ShadowStackGC and it’s
+ShadowStackGCLowering pass.</p>
+<p>There is currently no way to register such a custom lowering pass
+without building a custom copy of LLVM.</p>
+</div>
+<div class="section" id="generating-safe-points-neededsafepoints">
+<span id="safe-points"></span><h3><a class="toc-backref" href="#id26">Generating safe points: <tt class="docutils literal"><span class="pre">NeededSafePoints</span></tt></a><a class="headerlink" href="#generating-safe-points-neededsafepoints" title="Permalink to this headline">¶</a></h3>
+<p>LLVM can compute four kinds of safe points:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">namespace</span> <span class="n">GC</span> <span class="p">{</span>
+ <span class="c1">/// PointKind - The type of a collector-safe point.</span>
+ <span class="c1">///</span>
+ <span class="k">enum</span> <span class="n">PointKind</span> <span class="p">{</span>
+ <span class="n">Loop</span><span class="p">,</span> <span class="c1">//< Instr is a loop (backwards branch).</span>
+ <span class="n">Return</span><span class="p">,</span> <span class="c1">//< Instr is a return instruction.</span>
+ <span class="n">PreCall</span><span class="p">,</span> <span class="c1">//< Instr is a call instruction.</span>
+ <span class="n">PostCall</span> <span class="c1">//< Instr is the return address of a call.</span>
+ <span class="p">};</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>A collector can request any combination of the four by setting the
+<tt class="docutils literal"><span class="pre">NeededSafePoints</span></tt> mask:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">MyGC</span><span class="o">::</span><span class="n">MyGC</span><span class="p">()</span> <span class="p">{</span>
+ <span class="n">NeededSafePoints</span> <span class="o">=</span> <span class="mi">1</span> <span class="o"><<</span> <span class="n">GC</span><span class="o">::</span><span class="n">Loop</span>
+ <span class="o">|</span> <span class="mi">1</span> <span class="o"><<</span> <span class="n">GC</span><span class="o">::</span><span class="n">Return</span>
+ <span class="o">|</span> <span class="mi">1</span> <span class="o"><<</span> <span class="n">GC</span><span class="o">::</span><span class="n">PreCall</span>
+ <span class="o">|</span> <span class="mi">1</span> <span class="o"><<</span> <span class="n">GC</span><span class="o">::</span><span class="n">PostCall</span><span class="p">;</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>It can then use the following routines to access safe points.</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="k">for</span> <span class="p">(</span><span class="n">iterator</span> <span class="n">I</span> <span class="o">=</span> <span class="n">begin</span><span class="p">(),</span> <span class="n">E</span> <span class="o">=</span> <span class="n">end</span><span class="p">();</span> <span class="n">I</span> <span class="o">!=</span> <span class="n">E</span><span class="p">;</span> <span class="o">++</span><span class="n">I</span><span class="p">)</span> <span class="p">{</span>
+ <span class="n">GCFunctionInfo</span> <span class="o">*</span><span class="n">MD</span> <span class="o">=</span> <span class="o">*</span><span class="n">I</span><span class="p">;</span>
+ <span class="kt">size_t</span> <span class="n">PointCount</span> <span class="o">=</span> <span class="n">MD</span><span class="o">-></span><span class="n">size</span><span class="p">();</span>
+
+ <span class="k">for</span> <span class="p">(</span><span class="n">GCFunctionInfo</span><span class="o">::</span><span class="n">iterator</span> <span class="n">PI</span> <span class="o">=</span> <span class="n">MD</span><span class="o">-></span><span class="n">begin</span><span class="p">(),</span>
+ <span class="n">PE</span> <span class="o">=</span> <span class="n">MD</span><span class="o">-></span><span class="n">end</span><span class="p">();</span> <span class="n">PI</span> <span class="o">!=</span> <span class="n">PE</span><span class="p">;</span> <span class="o">++</span><span class="n">PI</span><span class="p">)</span> <span class="p">{</span>
+ <span class="n">GC</span><span class="o">::</span><span class="n">PointKind</span> <span class="n">PointKind</span> <span class="o">=</span> <span class="n">PI</span><span class="o">-></span><span class="n">Kind</span><span class="p">;</span>
+ <span class="kt">unsigned</span> <span class="n">PointNum</span> <span class="o">=</span> <span class="n">PI</span><span class="o">-></span><span class="n">Num</span><span class="p">;</span>
+ <span class="p">}</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>Almost every collector requires <tt class="docutils literal"><span class="pre">PostCall</span></tt> safe points, since these correspond
+to the moments when the function is suspended during a call to a subroutine.</p>
+<p>Threaded programs generally require <tt class="docutils literal"><span class="pre">Loop</span></tt> safe points to guarantee that the
+application will reach a safe point within a bounded amount of time, even if it
+is executing a long-running loop which contains no function calls.</p>
+<p>Threaded collectors may also require <tt class="docutils literal"><span class="pre">Return</span></tt> and <tt class="docutils literal"><span class="pre">PreCall</span></tt> safe points to
+implement “stop the world” techniques using self-modifying code, where it is
+important that the program not exit the function without reaching a safe point
+(because only the topmost function has been patched).</p>
+</div>
+<div class="section" id="emitting-assembly-code-gcmetadataprinter">
+<span id="assembly"></span><h3><a class="toc-backref" href="#id27">Emitting assembly code: <tt class="docutils literal"><span class="pre">GCMetadataPrinter</span></tt></a><a class="headerlink" href="#emitting-assembly-code-gcmetadataprinter" title="Permalink to this headline">¶</a></h3>
+<p>LLVM allows a plugin to print arbitrary assembly code before and after the rest
+of a module’s assembly code. At the end of the module, the GC can compile the
+LLVM stack map into assembly code. (At the beginning, this information is not
+yet computed.)</p>
+<p>Since AsmWriter and CodeGen are separate components of LLVM, a separate abstract
+base class and registry is provided for printing assembly code, the
+<tt class="docutils literal"><span class="pre">GCMetadaPrinter</span></tt> and <tt class="docutils literal"><span class="pre">GCMetadataPrinterRegistry</span></tt>. The AsmWriter will look
+for such a subclass if the <tt class="docutils literal"><span class="pre">GCStrategy</span></tt> sets <tt class="docutils literal"><span class="pre">UsesMetadata</span></tt>:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="n">MyGC</span><span class="o">::</span><span class="n">MyGC</span><span class="p">()</span> <span class="p">{</span>
+ <span class="n">UsesMetadata</span> <span class="o">=</span> <span class="nb">true</span><span class="p">;</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>This separation allows JIT-only clients to be smaller.</p>
+<p>Note that LLVM does not currently have analogous APIs to support code generation
+in the JIT, nor using the object writers.</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="c1">// lib/MyGC/MyGCPrinter.cpp - Example LLVM GC printer</span>
+
+<span class="cp">#include "llvm/CodeGen/GCMetadataPrinter.h"</span>
+<span class="cp">#include "llvm/Support/Compiler.h"</span>
+
+<span class="k">using</span> <span class="k">namespace</span> <span class="n">llvm</span><span class="p">;</span>
+
+<span class="k">namespace</span> <span class="p">{</span>
+ <span class="k">class</span> <span class="nc">LLVM_LIBRARY_VISIBILITY</span> <span class="n">MyGCPrinter</span> <span class="o">:</span> <span class="k">public</span> <span class="n">GCMetadataPrinter</span> <span class="p">{</span>
+ <span class="nl">public:</span>
+ <span class="k">virtual</span> <span class="kt">void</span> <span class="n">beginAssembly</span><span class="p">(</span><span class="n">AsmPrinter</span> <span class="o">&</span><span class="n">AP</span><span class="p">);</span>
+
+ <span class="k">virtual</span> <span class="kt">void</span> <span class="nf">finishAssembly</span><span class="p">(</span><span class="n">AsmPrinter</span> <span class="o">&</span><span class="n">AP</span><span class="p">);</span>
+ <span class="p">};</span>
+
+ <span class="n">GCMetadataPrinterRegistry</span><span class="o">::</span><span class="n">Add</span><span class="o"><</span><span class="n">MyGCPrinter</span><span class="o">></span>
+ <span class="n">X</span><span class="p">(</span><span class="s">"mygc"</span><span class="p">,</span> <span class="s">"My bespoke garbage collector."</span><span class="p">);</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+<p>The collector should use <tt class="docutils literal"><span class="pre">AsmPrinter</span></tt> to print portable assembly code. The
+collector itself contains the stack map for the entire module, and may access
+the <tt class="docutils literal"><span class="pre">GCFunctionInfo</span></tt> using its own <tt class="docutils literal"><span class="pre">begin()</span></tt> and <tt class="docutils literal"><span class="pre">end()</span></tt> methods. Here’s
+a realistic example:</p>
+<div class="highlight-c++"><div class="highlight"><pre><span class="cp">#include "llvm/CodeGen/AsmPrinter.h"</span>
+<span class="cp">#include "llvm/IR/Function.h"</span>
+<span class="cp">#include "llvm/IR/DataLayout.h"</span>
+<span class="cp">#include "llvm/Target/TargetAsmInfo.h"</span>
+<span class="cp">#include "llvm/Target/TargetMachine.h"</span>
+
+<span class="kt">void</span> <span class="n">MyGCPrinter</span><span class="o">::</span><span class="n">beginAssembly</span><span class="p">(</span><span class="n">AsmPrinter</span> <span class="o">&</span><span class="n">AP</span><span class="p">)</span> <span class="p">{</span>
+ <span class="c1">// Nothing to do.</span>
+<span class="p">}</span>
+
+<span class="kt">void</span> <span class="n">MyGCPrinter</span><span class="o">::</span><span class="n">finishAssembly</span><span class="p">(</span><span class="n">AsmPrinter</span> <span class="o">&</span><span class="n">AP</span><span class="p">)</span> <span class="p">{</span>
+ <span class="n">MCStreamer</span> <span class="o">&</span><span class="n">OS</span> <span class="o">=</span> <span class="n">AP</span><span class="p">.</span><span class="n">OutStreamer</span><span class="p">;</span>
+ <span class="kt">unsigned</span> <span class="n">IntPtrSize</span> <span class="o">=</span> <span class="n">AP</span><span class="p">.</span><span class="n">getPointerSize</span><span class="p">();</span>
+
+ <span class="c1">// Put this in the data section.</span>
+ <span class="n">OS</span><span class="p">.</span><span class="n">SwitchSection</span><span class="p">(</span><span class="n">AP</span><span class="p">.</span><span class="n">getObjFileLowering</span><span class="p">().</span><span class="n">getDataSection</span><span class="p">());</span>
+
+ <span class="c1">// For each function...</span>
+ <span class="k">for</span> <span class="p">(</span><span class="n">iterator</span> <span class="n">FI</span> <span class="o">=</span> <span class="n">begin</span><span class="p">(),</span> <span class="n">FE</span> <span class="o">=</span> <span class="n">end</span><span class="p">();</span> <span class="n">FI</span> <span class="o">!=</span> <span class="n">FE</span><span class="p">;</span> <span class="o">++</span><span class="n">FI</span><span class="p">)</span> <span class="p">{</span>
+ <span class="n">GCFunctionInfo</span> <span class="o">&</span><span class="n">MD</span> <span class="o">=</span> <span class="o">**</span><span class="n">FI</span><span class="p">;</span>
+
+ <span class="c1">// A compact GC layout. Emit this data structure:</span>
+ <span class="c1">//</span>
+ <span class="c1">// struct {</span>
+ <span class="c1">// int32_t PointCount;</span>
+ <span class="c1">// void *SafePointAddress[PointCount];</span>
+ <span class="c1">// int32_t StackFrameSize; // in words</span>
+ <span class="c1">// int32_t StackArity;</span>
+ <span class="c1">// int32_t LiveCount;</span>
+ <span class="c1">// int32_t LiveOffsets[LiveCount];</span>
+ <span class="c1">// } __gcmap_<FUNCTIONNAME>;</span>
+
+ <span class="c1">// Align to address width.</span>
+ <span class="n">AP</span><span class="p">.</span><span class="n">EmitAlignment</span><span class="p">(</span><span class="n">IntPtrSize</span> <span class="o">==</span> <span class="mi">4</span> <span class="o">?</span> <span class="mi">2</span> <span class="o">:</span> <span class="mi">3</span><span class="p">);</span>
+
+ <span class="c1">// Emit PointCount.</span>
+ <span class="n">OS</span><span class="p">.</span><span class="n">AddComment</span><span class="p">(</span><span class="s">"safe point count"</span><span class="p">);</span>
+ <span class="n">AP</span><span class="p">.</span><span class="n">EmitInt32</span><span class="p">(</span><span class="n">MD</span><span class="p">.</span><span class="n">size</span><span class="p">());</span>
+
+ <span class="c1">// And each safe point...</span>
+ <span class="k">for</span> <span class="p">(</span><span class="n">GCFunctionInfo</span><span class="o">::</span><span class="n">iterator</span> <span class="n">PI</span> <span class="o">=</span> <span class="n">MD</span><span class="p">.</span><span class="n">begin</span><span class="p">(),</span>
+ <span class="n">PE</span> <span class="o">=</span> <span class="n">MD</span><span class="p">.</span><span class="n">end</span><span class="p">();</span> <span class="n">PI</span> <span class="o">!=</span> <span class="n">PE</span><span class="p">;</span> <span class="o">++</span><span class="n">PI</span><span class="p">)</span> <span class="p">{</span>
+ <span class="c1">// Emit the address of the safe point.</span>
+ <span class="n">OS</span><span class="p">.</span><span class="n">AddComment</span><span class="p">(</span><span class="s">"safe point address"</span><span class="p">);</span>
+ <span class="n">MCSymbol</span> <span class="o">*</span><span class="n">Label</span> <span class="o">=</span> <span class="n">PI</span><span class="o">-></span><span class="n">Label</span><span class="p">;</span>
+ <span class="n">AP</span><span class="p">.</span><span class="n">EmitLabelPlusOffset</span><span class="p">(</span><span class="n">Label</span><span class="cm">/*Hi*/</span><span class="p">,</span> <span class="mi">0</span><span class="cm">/*Offset*/</span><span class="p">,</span> <span class="mi">4</span><span class="cm">/*Size*/</span><span class="p">);</span>
+ <span class="p">}</span>
+
+ <span class="c1">// Stack information never change in safe points! Only print info from the</span>
+ <span class="c1">// first call-site.</span>
+ <span class="n">GCFunctionInfo</span><span class="o">::</span><span class="n">iterator</span> <span class="n">PI</span> <span class="o">=</span> <span class="n">MD</span><span class="p">.</span><span class="n">begin</span><span class="p">();</span>
+
+ <span class="c1">// Emit the stack frame size.</span>
+ <span class="n">OS</span><span class="p">.</span><span class="n">AddComment</span><span class="p">(</span><span class="s">"stack frame size (in words)"</span><span class="p">);</span>
+ <span class="n">AP</span><span class="p">.</span><span class="n">EmitInt32</span><span class="p">(</span><span class="n">MD</span><span class="p">.</span><span class="n">getFrameSize</span><span class="p">()</span> <span class="o">/</span> <span class="n">IntPtrSize</span><span class="p">);</span>
+
+ <span class="c1">// Emit stack arity, i.e. the number of stacked arguments.</span>
+ <span class="kt">unsigned</span> <span class="n">RegisteredArgs</span> <span class="o">=</span> <span class="n">IntPtrSize</span> <span class="o">==</span> <span class="mi">4</span> <span class="o">?</span> <span class="mi">5</span> <span class="o">:</span> <span class="mi">6</span><span class="p">;</span>
+ <span class="kt">unsigned</span> <span class="n">StackArity</span> <span class="o">=</span> <span class="n">MD</span><span class="p">.</span><span class="n">getFunction</span><span class="p">().</span><span class="n">arg_size</span><span class="p">()</span> <span class="o">></span> <span class="n">RegisteredArgs</span> <span class="o">?</span>
+ <span class="n">MD</span><span class="p">.</span><span class="n">getFunction</span><span class="p">().</span><span class="n">arg_size</span><span class="p">()</span> <span class="o">-</span> <span class="n">RegisteredArgs</span> <span class="o">:</span> <span class="mi">0</span><span class="p">;</span>
+ <span class="n">OS</span><span class="p">.</span><span class="n">AddComment</span><span class="p">(</span><span class="s">"stack arity"</span><span class="p">);</span>
+ <span class="n">AP</span><span class="p">.</span><span class="n">EmitInt32</span><span class="p">(</span><span class="n">StackArity</span><span class="p">);</span>
+
+ <span class="c1">// Emit the number of live roots in the function.</span>
+ <span class="n">OS</span><span class="p">.</span><span class="n">AddComment</span><span class="p">(</span><span class="s">"live root count"</span><span class="p">);</span>
+ <span class="n">AP</span><span class="p">.</span><span class="n">EmitInt32</span><span class="p">(</span><span class="n">MD</span><span class="p">.</span><span class="n">live_size</span><span class="p">(</span><span class="n">PI</span><span class="p">));</span>
+
+ <span class="c1">// And for each live root...</span>
+ <span class="k">for</span> <span class="p">(</span><span class="n">GCFunctionInfo</span><span class="o">::</span><span class="n">live_iterator</span> <span class="n">LI</span> <span class="o">=</span> <span class="n">MD</span><span class="p">.</span><span class="n">live_begin</span><span class="p">(</span><span class="n">PI</span><span class="p">),</span>
+ <span class="n">LE</span> <span class="o">=</span> <span class="n">MD</span><span class="p">.</span><span class="n">live_end</span><span class="p">(</span><span class="n">PI</span><span class="p">);</span>
+ <span class="n">LI</span> <span class="o">!=</span> <span class="n">LE</span><span class="p">;</span> <span class="o">++</span><span class="n">LI</span><span class="p">)</span> <span class="p">{</span>
+ <span class="c1">// Emit live root's offset within the stack frame.</span>
+ <span class="n">OS</span><span class="p">.</span><span class="n">AddComment</span><span class="p">(</span><span class="s">"stack index (offset / wordsize)"</span><span class="p">);</span>
+ <span class="n">AP</span><span class="p">.</span><span class="n">EmitInt32</span><span class="p">(</span><span class="n">LI</span><span class="o">-></span><span class="n">StackOffset</span><span class="p">);</span>
+ <span class="p">}</span>
+ <span class="p">}</span>
+<span class="p">}</span>
+</pre></div>
+</div>
+</div>
+</div>
+<div class="section" id="references">
+<h2><a class="toc-backref" href="#id28">References</a><a class="headerlink" href="#references" title="Permalink to this headline">¶</a></h2>
+<p id="appel89">[Appel89] Runtime Tags Aren’t Necessary. Andrew W. Appel. Lisp and Symbolic
+Computation 19(7):703-705, July 1989.</p>
+<p id="goldberg91">[Goldberg91] Tag-free garbage collection for strongly typed programming
+languages. Benjamin Goldberg. ACM SIGPLAN PLDI‘91.</p>
+<p id="tolmach94">[Tolmach94] Tag-free garbage collection using explicit type parameters. Andrew
+Tolmach. Proceedings of the 1994 ACM conference on LISP and functional
+programming.</p>
+<p id="henderson02">[Henderson2002] <a class="reference external" href="http://citeseer.ist.psu.edu/henderson02accurate.html">Accurate Garbage Collection in an Uncooperative Environment</a></p>
+</div>
+</div>
+
+
+ </div>
+ </div>
+ <div class="clearer"></div>
+ </div>
+ <div class="related">
+ <h3>Navigation</h3>
+ <ul>
+ <li class="right" style="margin-right: 10px">
+ <a href="genindex.html" title="General Index"
+ >index</a></li>
+ <li class="right" >
+ <a href="WritingAnLLVMPass.html" title="Writing an LLVM Pass"
+ >next</a> |</li>
+ <li class="right" >
+ <a href="HowToUseInstrMappings.html" title="How To Use Instruction Mappings"
+ >previous</a> |</li>
+ <li><a href="http://llvm.org/">LLVM Home</a> | </li>
+ <li><a href="index.html">Documentation</a>»</li>
+
+ </ul>
+ </div>
+ <div class="footer">
+ © Copyright 2003-2017, LLVM Project.
+ Last updated on 2017-03-08.
+ Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.2.
+ </div>
+ </body>
+</html>
\ No newline at end of file
More information about the llvm-commits
mailing list