clang/docs/UsersManual.html

<html>
<head>
<title>Clang Compiler User's Manual</title>
<link type="text/css" rel="stylesheet" href="../menu.css" />
<link type="text/css" rel="stylesheet" href="../content.css" />
<style type="text/css">
td {
	vertical-align: top;
}
</style>
</head>
<body>

<!--#include virtual="../menu.html.incl"-->

<div id="content">

<h1>Clang Compiler User's Manual</h1>

<ul>
<li><a href="#intro">Introduction</a>
  <ul>
  <li><a href="#terminology">Terminology</a></li>
  <li><a href="#basicusage">Basic Usage</a></li>
  </ul>
</li>
<li><a href="#commandline">Command Line Options</a>
  <ul>
  <li><a href="#cl_diagnostics">Options to Control Error and Warning
      Messages</a></li>
  </ul>
</li>
<li><a href="#general_features">Language and Target-Independent Features</a>
 <ul>
  <li><a href="#diagnostics">Controlling Errors and Warnings</a>
   <ul>
   <li><a href="#diagnostics_display">Controlling How Clang Displays Diagnostics</a></li>
   <li><a href="#diagnostics_mappings">Diagnostic Mappings</a></li>
   <li><a href="#diagnostics_categories">Diagnostic Categories</a></li>
   <li><a href="#diagnostics_commandline">Controlling Diagnostics via Command Line Flags</a></li>
   <li><a href="#diagnostics_pragmas">Controlling Diagnostics via Pragmas</a></li>
   <li><a href="#analyzer_diagnositics">Controlling Static Analyzer Diagnostics</a></li>
   </ul>
  </li>
  <li><a href="#precompiledheaders">Precompiled Headers</a></li>
  <li><a href="#codegen">Controlling Code Generation</a></li>
 </ul>
</li>
<li><a href="#c">C Language Features</a>
  <ul>
  <li><a href="#c_ext">Extensions supported by clang</a></li>
  <li><a href="#c_modes">Differences between various standard modes</a></li>
  <li><a href="#c_unimpl_gcc">GCC extensions not implemented yet</a></li>
  <li><a href="#c_unsupp_gcc">Intentionally unsupported GCC extensions</a></li>
  <li><a href="#c_ms">Microsoft extensions</a></li>
  </ul>
</li>
<li><a href="#target_features">Target-Specific Features and Limitations</a>
  <ul>
  <li><a href="#target_arch">CPU Architectures Features and Limitations</a>
    <ul>
    <li><a href="#target_arch_x86">X86</a></li>
    <li><a href="#target_arch_arm">ARM</a></li>
    <li><a href="#target_arch_other">Other platforms</a></li>
    </ul>
  </li>
  <li><a href="#target_os">Operating System Features and Limitations</a>
    <ul>
    <li><a href="#target_os_darwin">Darwin (Mac OS/X)</a></li>
    <li>Linux, etc.</li>
    <li><a href="#target_os_win32">Windows</a></li>
    </ul>
  </li>
  </ul>
</li>
</ul>


<!-- ======================================================================= -->
<h2 id="intro">Introduction</h2>
<!-- ======================================================================= -->

<p>The Clang Compiler is an open-source compiler for the C family of programming
languages, aiming to be the best in class implementation of these languages.
Clang builds on the LLVM optimizer and code generator, allowing it to provide
high-quality optimization and code generation support for many targets.  For
more general information, please see the <a href="http://clang.llvm.org">Clang
Web Site</a> or the <a href="http://llvm.org">LLVM Web Site</a>.</p>

<p>This document describes important notes about using Clang as a compiler for
an end-user, documenting the supported features, command line options, etc.  If
you are interested in using Clang to build a tool that processes code, please
see <a href="InternalsManual.html">the Clang Internals Manual</a>.  If you are
interested in the <a href="http://clang.llvm.org/StaticAnalysis.html">Clang
Static Analyzer</a>, please see its web page.</p>

<p>Clang is designed to support the C family of programming languages, which
includes <a href="#c">C</a>, <a href="#objc">Objective-C</a>, <a
href="#cxx">C++</a>, and <a href="#objcxx">Objective-C++</a> as well as many
dialects of those.  For language-specific information, please see the
corresponding language specific section:</p>

<ul>
<li><a href="#c">C Language</a>: K&amp;R C, ANSI C89, ISO C90, ISO C94
    (C89+AMD1), ISO C99 (+TC1, TC2, TC3). </li>
<li><a href="#objc">Objective-C Language</a>: ObjC 1, ObjC 2, ObjC 2.1, plus
    variants depending on base language.</li>
<li><a href="#cxx">C++ Language Features</a></li>
<li><a href="#objcxx">Objective C++ Language</a></li>
</ul>

<p>In addition to these base languages and their dialects, Clang supports a
broad variety of language extensions, which are documented in the corresponding
language section.  These extensions are provided to be compatible with the GCC,
Microsoft, and other popular compilers as well as to improve functionality
through Clang-specific features.  The Clang driver and language features are
intentionally designed to be as compatible with the GNU GCC compiler as
reasonably possible, easing migration from GCC to Clang.  In most cases, code
"just works".</p>

<p>In addition to language specific features, Clang has a variety of features
that depend on what CPU architecture or operating system is being compiled for.
Please see the <a href="#target_features">Target-Specific Features and
Limitations</a> section for more details.</p>

<p>The rest of the introduction introduces some basic <a
href="#terminology">compiler terminology</a> that is used throughout this manual
and contains a basic <a href="#basicusage">introduction to using Clang</a>
as a command line compiler.</p>

<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="terminology">Terminology</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<p>Front end, parser, backend, preprocessor, undefined behavior, diagnostic,
 optimizer</p>

<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="basicusage">Basic Usage</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<p>Intro to how to use a C compiler for newbies.</p>
<p>
compile + link

compile then link

debug info

enabling optimizations

picking a language to use, defaults to C99 by default.  Autosenses based on
extension.

using a makefile
</p>


<!-- ======================================================================= -->
<h2 id="commandline">Command Line Options</h2>
<!-- ======================================================================= -->

<p>
This section is generally an index into other sections.  It does not go into
depth on the ones that are covered by other sections.  However, the first part
introduces the language selection and other high level options like -c, -g, etc.
</p>


<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="cl_diagnostics">Options to Control Error and Warning Messages</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<p><b>-Werror</b>: Turn warnings into errors.</p>
<p><b>-Werror=foo</b>: Turn warning "foo" into an error.</p>
<p><b>-Wno-error=foo</b>: Turn warning "foo" into an warning even if -Werror is
   specified.</p>
<p><b>-Wfoo</b>: Enable warning foo</p>
<p><b>-Wno-foo</b>: Disable warning foo</p>
<p><b>-w</b>: Disable all warnings.</p>
<p><b>-pedantic</b>: Warn on language extensions.</p>
<p><b>-pedantic-errors</b>: Error on language extensions.</p>
<p><b>-Wsystem-headers</b>: Enable warnings from system headers.</p>

<p><b>-ferror-limit=123</b>: Stop emitting diagnostics after 123 errors have
   been produced.  The default is 20, and the error limit can be disabled with
   -ferror-limit=0.</p>

<p><b>-ftemplate-backtrace-limit=123</b>: Only emit up to 123 template instantiation notes within the template instantiation backtrace for a single warning or error. The default is 10, and the limit can be disabled with -ftemplate-backtrace-limit=0.</p>

<!-- ================================================= -->
<h4 id="cl_diag_formatting">Formatting of Diagnostics</h4>
<!-- ================================================= -->

<p>Clang aims to produce beautiful diagnostics by default, particularly for new
users that first come to Clang.  However, different people have different
preferences, and sometimes Clang is driven by another program that wants to
parse simple and consistent output, not a person. For these cases, Clang
provides a wide range of options to control the exact output format of the
diagnostics that it generates.</p>

<dl>

<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_fshow-column"><b>-f[no-]show-column</b>: Print column number in
diagnostic.</dt>
<dd>This option, which defaults to on, controls whether or not Clang prints the
column number of a diagnostic.  For example, when this is enabled, Clang will
print something like:

<pre>
  test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
  #endif bad
         ^
         //
</pre>

<p>When this is disabled, Clang will print "test.c:28: warning..." with no
column number.</p>
</dd>

<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_fshow-source-location"><b>-f[no-]show-source-location</b>: Print
source file/line/column information in diagnostic.</dt>
<dd>This option, which defaults to on, controls whether or not Clang prints the
filename, line number and column number of a diagnostic.  For example,
when this is enabled, Clang will print something like:

<pre>
  test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
  #endif bad
         ^
         //
</pre>

<p>When this is disabled, Clang will not print the "test.c:28:8: " part.</p>
</dd>

<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_fcaret-diagnostics"><b>-f[no-]caret-diagnostics</b>: Print source
line and ranges from source code in diagnostic.</dt>
<dd>This option, which defaults to on, controls whether or not Clang prints the
source line, source ranges, and caret when emitting a diagnostic.  For example,
when this is enabled, Clang will print something like:

<pre>
  test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
  #endif bad
         ^
         //
</pre>
</dd>
<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_fcolor_diagnostics"><b>-f[no-]color-diagnostics</b>: </dt>
<dd>This option, which defaults to on when a color-capable terminal is
  detected, controls whether or not Clang prints diagnostics in color.
  When this option is enabled, Clang will use colors to highlight
  specific parts of the diagnostic, e.g.,
 <pre>
  <b><font color="black">test.c:28:8: <font color="magenta">warning</font>: extra tokens at end of #endif directive [-Wextra-tokens]</font></b>
  #endif bad
         <font color="green">^</font>
         <font color="green">//</font>
</pre>

<p>When this is disabled, Clang will just print:</p>

<pre>
  test.c:2:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
  #endif bad
         ^
         //
</pre>
</dd>
<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_fdiagnostics-format"><b>-fdiagnostics-format=clang/msvc/vi</b>:
Changes diagnostic output format to better match IDEs and command line tools.</dt>
<dd>This option controls the output format of the filename, line number, and column printed in diagnostic messages. The options, and their affect on formatting a simple conversion diagnostic, follow:

  <dl>
    <dt><b>clang</b> (default)</dt>
    <dd>
      <pre>t.c:3:11: warning: conversion specifies type 'char *' but the argument has type 'int'</pre>
    </dd>

    <dt><b>msvc</b></dt>
    <dd>
      <pre>t.c(3,11) : warning: conversion specifies type 'char *' but the argument has type 'int'</pre>
    </dd>

    <dt><b>vi</b></dt>
    <dd>
      <pre>t.c +3:11: warning: conversion specifies type 'char *' but the argument has type 'int'</pre>
    </dd>
  </dl>
</dd>

<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_fdiagnostics-show-name"><b>-f[no-]diagnostics-show-name</b>:
Enable the display of the diagnostic name.</dt>
<dd>This option, which defaults to off, controls whether or not
Clang prints the associated name.</dd>
<br>
<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_fdiagnostics-show-option"><b>-f[no-]diagnostics-show-option</b>:
Enable <tt>[-Woption]</tt> information in diagnostic line.</dt>
<dd>This option, which defaults to on,
controls whether or not Clang prints the associated <A
href="#cl_diag_warning_groups">warning group</a> option name when outputting
a warning diagnostic.  For example, in this output:

<pre>
  test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
  #endif bad
         ^
         //
</pre>

<p>Passing <b>-fno-diagnostics-show-option</b> will prevent Clang from printing
the [<a href="#opt_Wextra-tokens">-Wextra-tokens</a>] information in the
diagnostic.  This information tells you the flag needed to enable or disable the
diagnostic, either from the command line or through <a
href="#pragma_GCC_diagnostic">#pragma GCC diagnostic</a>.</dd>

<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_fdiagnostics-show-category"><b>-fdiagnostics-show-category=none/id/name</b>:
Enable printing category information in diagnostic line.</dt>
<dd>This option, which defaults to "none",
controls whether or not Clang prints the category associated with a diagnostic
when emitting it.  Each diagnostic may or many not have an associated category,
if it has one, it is listed in the diagnostic categorization field of the
diagnostic line (in the []'s).

<p>For example, a format string warning will produce these three renditions
based on the setting of this option:</p>

<pre>
  t.c:3:11: warning: conversion specifies type 'char *' but the argument has type 'int' [-Wformat]
  t.c:3:11: warning: conversion specifies type 'char *' but the argument has type 'int' [-Wformat<b>,1</b>]
  t.c:3:11: warning: conversion specifies type 'char *' but the argument has type 'int' [-Wformat<b>,Format String</b>]
</pre>

<p>This category can be used by clients that want to group diagnostics by
category, so it should be a high level category.  We want dozens of these, not
hundreds or thousands of them.</p>
</dd>


<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_fdiagnostics-fixit-info"><b>-f[no-]diagnostics-fixit-info</b>:
Enable "FixIt" information in the diagnostics output.</dt>
<dd>This option, which defaults to on, controls whether or not Clang prints the
information on how to fix a specific diagnostic underneath it when it knows.
For example, in this output:

<pre>
  test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
  #endif bad
         ^
         //
</pre>

<p>Passing <b>-fno-diagnostics-fixit-info</b> will prevent Clang from printing
the "//" line at the end of the message.  This information is useful for users
who may not understand what is wrong, but can be confusing for machine
parsing.</p>
</dd>

<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_fdiagnostics-print-source-range-info">
<b>-f[no-]diagnostics-print-source-range-info</b>:
Print machine parsable information about source ranges.</dt>
<dd>This option, which defaults to off, controls whether or not Clang prints
information about source ranges in a machine parsable format after the
file/line/column number information.  The information is a simple sequence of
brace enclosed ranges, where each range lists the start and end line/column
locations.  For example, in this output:

<pre>
exprs.c:47:15:{47:8-47:14}{47:17-47:24}: error: invalid operands to binary expression ('int *' and '_Complex float')
   P = (P-42) + Gamma*4;
       ~~~~~~ ^ ~~~~~~~
</pre>

<p>The {}'s are generated by -fdiagnostics-print-source-range-info.</p>
</dd>

<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_fdiagnostics-parseable-fixits">
<b>-fdiagnostics-parseable-fixits</b>:
Print Fix-Its in a machine parseable form.</dt>
<dd><p>This option makes Clang print available Fix-Its in a machine parseable format at the end of diagnostics. The following example illustrates the format:</p>

<pre>
 fix-it:"t.cpp":{7:25-7:29}:"Gamma"
</pre>

<p>The range printed is a half-open range, so in this example the characters at
column 25 up to but not including column 29 on line 7 in t.cpp should be
replaced with the string &quot;Gamma&quot;. Either the range or the replacement
string may be empty (representing strict insertions and strict erasures,
respectively). Both the file name and the insertion string escape backslash (as
&quot;\\&quot;), tabs (as &quot;\t&quot;), newlines (as &quot;\n&quot;), double
quotes(as &quot;\&quot;&quot;) and non-printable characters (as octal
&quot;\xxx&quot;).</p>
</dd>

</dl>


<!-- ===================================================== -->
<h4 id="cl_diag_warning_groups">Individual Warning Groups</h4>
<!-- ===================================================== -->

<p>TODO: Generate this from tblgen.  Define one anchor per warning group.</p>


<dl>


<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_Wextra-tokens"><b>-Wextra-tokens</b>: Warn about excess tokens at
    the end of a preprocessor directive.</dt>
<dd>This option, which defaults to on, enables warnings about extra tokens at
the end of preprocessor directives.  For example:

<pre>
  test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
  #endif bad
         ^
</pre>

<p>These extra tokens are not strictly conforming, and are usually best handled
by commenting them out.</p>

<p>This option is also enabled by <a href="">-Wfoo</a>, <a href="">-Wbar</a>,
 and <a href="">-Wbaz</a>.</p>
</dd>

<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_Wambiguous-member-template"><b>-Wambiguous-member-template</b>:
Warn about unqualified uses of a member template whose name resolves
to another template at the location of the use.</dt>
<dd>This option, which defaults to on, enables a warning in the
following code:

<pre>
template&lt;typename T> struct set{};
template&lt;typename T> struct trait { typedef const T& type; };
struct Value {
  template&lt;typename T> void set(typename trait&lt;T>::type value) {}
};
void foo() {
  Value v;
  v.set&lt;double>(3.2);
}
</pre>

<p>C++ [basic.lookup.classref] requires this to be an error, but,
because it's hard to work around, Clang downgrades it to a warning as
an extension.</p>
</dd>

<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dt id="opt_Wbind-to-temporary-copy"><b>-Wbind-to-temporary-copy</b>: Warn about
an unusable copy constructor when binding a reference to a temporary.</dt>
<dd>This option, which defaults to on, enables warnings about binding a
reference to a temporary when the temporary doesn't have a usable copy
constructor.  For example:

<pre>
  struct NonCopyable {
    NonCopyable();
  private:
    NonCopyable(const NonCopyable&);
  };
  void foo(const NonCopyable&);
  void bar() {
    foo(NonCopyable());  // Disallowed in C++98; allowed in C++0x.
  }
</pre>
<pre>
  struct NonCopyable2 {
    NonCopyable2();
    NonCopyable2(NonCopyable2&);
  };
  void foo(const NonCopyable2&);
  void bar() {
    foo(NonCopyable2());  // Disallowed in C++98; allowed in C++0x.
  }
</pre>

<p>Note that if <tt>NonCopyable2::NonCopyable2()</tt> has a default
argument whose instantiation produces a compile error, that error will
still be a hard error in C++98 mode even if this warning is turned
off.</p>

</dd>

</dl>

<!-- ======================================================================= -->
<h2 id="general_features">Language and Target-Independent Features</h2>
<!-- ======================================================================= -->


<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="diagnostics">Controlling Errors and Warnings</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<p>Clang provides a number of ways to control which code constructs cause it to
emit errors and warning messages, and how they are displayed to the console.</p>

<h4 id="diagnostics_display">Controlling How Clang Displays Diagnostics</h4>

<p>When Clang emits a diagnostic, it includes rich information in the output,
and gives you fine-grain control over which information is printed.  Clang has
the ability to print this information, and these are the options that control
it:</p>

<ol>
<li>A file/line/column indicator that shows exactly where the diagnostic occurs
    in your code [<a href="#opt_fshow-column">-fshow-column</a>, <a
    href="#opt_fshow-source-location">-fshow-source-location</a>].</li>
<li>A categorization of the diagnostic as a note, warning, error, or fatal
    error.</li>
<li>A text string that describes what the problem is.</li>
<li>An option that indicates whether to print the diagnostic name [<a
    href="#opt_fdiagnostics-show-name">-fdiagnostics-show-name</a>].</li>
<li>An option that indicates how to control the diagnostic (for diagnostics that
    support it) [<a
   href="#opt_fdiagnostics-show-option">-fdiagnostics-show-option</a>].</li>
<li>A <a href="#diagnostics_categories">high-level category</a> for the
    diagnostic for clients that want to group diagnostics by class (for
    diagnostics that support it) [<a
   href="#opt_fdiagnostics-show-category">-fdiagnostics-show-category</a>].</li>
<li>The line of source code that the issue occurs on, along with a caret and
    ranges that indicate the important locations [<a
    href="opt_fcaret-diagnostics">-fcaret-diagnostics</a>].</li>
<li>"FixIt" information, which is a concise explanation of how to fix the
    problem (when Clang is certain it knows) [<a
    href="opt_fdiagnostics-fixit-info">-fdiagnostics-fixit-info</a>].</li>
<li>A machine-parsable representation of the ranges involved (off by
    default) [<a
      href="opt_fdiagnostics-print-source-range-info">-fdiagnostics-print-source-range-info</a>].</li>
</ol>

<p>For more information please see <a href="#cl_diag_formatting">Formatting of
Diagnostics</a>.</p>


<h4 id="diagnostics_mappings">Diagnostic Mappings</h4>

<p>All diagnostics are mapped into one of these 5 classes:</p>

<ul>
<li>Ignored</li>
<li>Note</li>
<li>Warning</li>
<li>Error</li>
<li>Fatal</li>
</ul>

<h4 id="diagnostics_categories">Diagnostic Categories</h4>

<p>Though not shown by default, diagnostics may each be associated with a
   high-level category.  This category is intended to make it possible to triage
   builds that produce a large number of errors or warnings in a grouped way.
</p>

<p>Categories are not shown by default, but they can be turned on with the
<a href="#opt_fdiagnostics-show-category">-fdiagnostics-show-category</a> option.
When set to "<tt>name</tt>", the category is printed textually in the diagnostic
output.  When it is set to "<tt>id</tt>", a category number is printed.  The
mapping of category names to category id's can be obtained by running '<tt>clang
  --print-diagnostic-categories</tt>'.
</p>

<h4 id="diagnostics_commandline">Controlling Diagnostics via Command Line
 Flags</h4>

<p>-W flags, -pedantic, etc</p>

<h4 id="diagnostics_pragmas">Controlling Diagnostics via Pragmas</h4>

<p>Clang can also control what diagnostics are enabled through the use of
pragmas in the source code. This is useful for turning off specific warnings
in a section of source code. Clang supports GCC's pragma for compatibility
with existing source code, as well as several extensions. </p>

<p>The pragma may control any warning that can be used from the command line.
Warnings may be set to ignored, warning, error, or fatal. The following
example code will tell Clang or GCC to ignore the -Wall warnings:</p>

<pre>
#pragma GCC diagnostic ignored "-Wall"
</pre>

<p>In addition to all of the functionality provided by GCC's pragma, Clang
also allows you to push and pop the current warning state.  This is particularly
useful when writing a header file that will be compiled by other people, because
you don't know what warning flags they build with.</p>

<p>In the below example
-Wmultichar is ignored for only a single line of code, after which the
diagnostics return to whatever state had previously existed.</p>

<pre>
#pragma clang diagnostic push
#pragma clang diagnostic ignored "-Wmultichar"

char b = 'df'; // no warning.

#pragma clang diagnostic pop
</pre>

<p>The push and pop pragmas will save and restore the full diagnostic state of
the compiler, regardless of how it was set. That means that it is possible to
use push and pop around GCC compatible diagnostics and Clang will push and pop
them appropriately, while GCC will ignore the pushes and pops as unknown
pragmas. It should be noted that while Clang supports the GCC pragma, Clang and
GCC do not support the exact same set of warnings, so even when using GCC
compatible #pragmas there is no guarantee that they will have identical behaviour
on both compilers. </p>

<h4 id="analyzer_diagnositics">Controlling Static Analyzer Diagnostics</h4>

<p>While not strictly part of the compiler, the diagnostics from Clang's <a
href="http://clang-analyzer.llvm.org">static analyzer</a> can also be influenced
by the user via changes to the source code.  This can be done in two ways:

<ul>

<li id="analyzer_annotations"><b>Annotations</b>: The static analyzer recognizes various GCC-style
attributes (e.g., <tt>__attribute__((nonnull)))</tt>) that can either suppress
static analyzer warnings or teach the analyzer about code invariants which
enable it to find more bugs. While many of these attributes are standard GCC
attributes, additional ones have been added to Clang to specifically support the
static analyzer. Detailed information on these annotations can be found in the
<a href="http://clang-analyzer.llvm.org/annotations.html">analyzer's
documentation</a>.</li>

<li><b><tt>__clang_analyzer__</tt></b>: When the static analyzer is using Clang
to parse source files, it implicitly defines the preprocessor macro
<tt>__clang_analyzer__</tt>. While discouraged, code can use this macro to
selectively exclude code the analyzer examines.  Here is an example:

<pre>
#ifndef __clang_analyzer__
// Code not to be analyzed
#endif
</pre>

In general, this usage is discouraged. Instead, we prefer that users file bugs
against the analyzer when it flags false positives. There is also active
discussion of allowing users in the future to selectively silence specific
analyzer warnings (some of which can already be done using <a
href="analyzer_annotations">annotations</a>).</li>

</ul>

<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="precompiledheaders">Precompiled Headers</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<p><a href="http://en.wikipedia.org/wiki/Precompiled_header">Precompiled
headers</a> are a general approach employed by many compilers to reduce
compilation time. The underlying motivation of the approach is that it is
common for the same (and often large) header files to be included by
multiple source files. Consequently, compile times can often be greatly improved
by caching some of the (redundant) work done by a compiler to process headers.
Precompiled header files, which represent one of many ways to implement
this optimization, are literally files that represent an on-disk cache that
contains the vital information necessary to reduce some of the work
needed to process a corresponding header file. While details of precompiled
headers vary between compilers, precompiled headers have been shown to be
highly effective at speeding up program compilation on systems with very large
system headers (e.g., Mac OS/X).</p>

<h4>Generating a PCH File</h4>

<p>To generate a PCH file using Clang, one invokes Clang with
the <b><tt>-x <i>&lt;language&gt;</i>-header</tt></b> option. This mirrors the
interface in GCC for generating PCH files:</p>

<pre>
  $ gcc -x c-header test.h -o test.h.gch
  $ clang -x c-header test.h -o test.h.pch
</pre>

<h4>Using a PCH File</h4>

<p>A PCH file can then be used as a prefix header when a
<b><tt>-include</tt></b> option is passed to <tt>clang</tt>:</p>

<pre>
  $ clang -include test.h test.c -o test
</pre>

<p>The <tt>clang</tt> driver will first check if a PCH file for <tt>test.h</tt>
is available; if so, the contents of <tt>test.h</tt> (and the files it includes)
will be processed from the PCH file. Otherwise, Clang falls back to
directly processing the content of <tt>test.h</tt>. This mirrors the behavior of
GCC.</p>

<p><b>NOTE:</b> Clang does <em>not</em> automatically use PCH files
for headers that are directly included within a source file. For example:</p>

<pre>
  $ clang -x c-header test.h -o test.h.pch
  $ cat test.c
  #include "test.h"
  $ clang test.c -o test
</pre>

<p>In this example, <tt>clang</tt> will not automatically use the PCH file for
<tt>test.h</tt> since <tt>test.h</tt> was included directly in the source file
and not specified on the command line using <tt>-include</tt>.</p>

<h4>Relocatable PCH Files</h4>
<p>It is sometimes necessary to build a precompiled header from headers that
are not yet in their final, installed locations. For example, one might build a
precompiled header within the build tree that is then meant to be installed
alongside the headers. Clang permits the creation of "relocatable" precompiled
headers, which are built with a given path (into the build directory) and can
later be used from an installed location.</p>

<p>To build a relocatable precompiled header, place your headers into a
subdirectory whose structure mimics the installed location. For example, if you
want to build a precompiled header for the header <code>mylib.h</code> that
will be installed into <code>/usr/include</code>, create a subdirectory
<code>build/usr/include</code> and place the header <code>mylib.h</code> into
that subdirectory. If <code>mylib.h</code> depends on other headers, then
they can be stored within <code>build/usr/include</code> in a way that mimics
the installed location.</p>

<p>Building a relocatable precompiled header requires two additional arguments.
First, pass the <code>--relocatable-pch</code> flag to indicate that the
resulting PCH file should be relocatable. Second, pass
<code>-isysroot /path/to/build</code>, which makes all includes for your
library relative to the build directory. For example:</p>

<pre>
  # clang -x c-header --relocatable-pch -isysroot /path/to/build /path/to/build/mylib.h mylib.h.pch
</pre>

<p>When loading the relocatable PCH file, the various headers used in the PCH
file are found from the system header root. For example, <code>mylib.h</code>
can be found in <code>/usr/include/mylib.h</code>. If the headers are installed
in some other system root, the <code>-isysroot</code> option can be used provide
a different system root from which the headers will be based. For example,
<code>-isysroot /Developer/SDKs/MacOSX10.4u.sdk</code> will look for
<code>mylib.h</code> in
<code>/Developer/SDKs/MacOSX10.4u.sdk/usr/include/mylib.h</code>.</p>

<p>Relocatable precompiled headers are intended to be used in a limited number
of cases where the compilation environment is tightly controlled and the
precompiled header cannot be generated after headers have been installed.
Relocatable precompiled headers also have some performance impact, because
the difference in location between the header locations at PCH build time vs.
at the time of PCH use requires one of the PCH optimizations,
<code>stat()</code> caching, to be disabled. However, this change is only
likely to affect PCH files that reference a large number of headers.</p>

<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="codegen">Controlling Code Generation</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<p>Clang provides a number of ways to control code generation.  The options are listed below.</p>

<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
<dl>
<dt id="opt_fcatch-undefined-behavior"><b>-fcatch-undefined-behavior</b>: Turn
on runtime code generation to check for undefined behavior.</dt>

<dd>This option, which defaults to off, controls whether or not Clang
adds runtime checks for undefined runtime behavior.  If a check fails,
<tt>__builtin_trap()</tt> is used to indicate failure.
The checks are:
<ul>
<li>Subscripting where the static type of one operand is a variable
    which is decayed from an array type and the other operand is
    greater than the size of the array or less than zero.</li>
<li>Shift operators where the amount shifted is greater or equal to the
    promoted bit-width of the left-hand-side or less than zero.</li>
<li>If control flow reaches __builtin_unreachable.
<li>When llvm implements more __builtin_object_size support, reads and
    writes for objects that __builtin_object_size indicates we aren't
    accessing valid memory.  Bit-fields and vectors are not yet checked.
</ul>
</dd>

<dt id="opt_fno-assume-sane-operator-new"><b>-fno-assume-sane-operator-new</b>:
Don't assume that the C++'s new operator is sane.</dt>
<dd>This option tells the compiler to do not assume that C++'s global new
operator will always return a pointer that does not
alias any other pointer when the function returns.</dd>

<dt id="opt_ftrap-function"><b>-ftrap-function=[name]</b>: Instruct code
generator to emit a function call to the specified function name for
<tt>__builtin_trap()</tt>.</dt>

<dd>LLVM code generator translates <tt>__builtin_trap()</tt> to a trap
instruction if it is supported by the target ISA. Otherwise, the builtin is
translated into a call to <tt>abort</tt>. If this option is set, then the code
generator will always lower the builtin to a call to the specified function
regardless of whether the target ISA has a trap instruction. This option is
useful for environments (e.g. deeply embedded) where a trap cannot be properly
handled, or when some custom behavior is desired.</dd>
</dl>

<!-- ======================================================================= -->
<h2 id="c">C Language Features</h2>
<!-- ======================================================================= -->

<p>The support for standard C in clang is feature-complete except for the C99
floating-point pragmas.</p>

<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="c_ext">Extensions supported by clang</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<p>See <a href="LanguageExtensions.html">clang language extensions</a>.</p>

<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="c_modes">Differences between various standard modes</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<p>clang supports the -std option, which changes what language mode clang uses.
The supported modes for C are c89, gnu89, c94, c99, gnu99 and various aliases
for those modes.  If no -std option is specified, clang defaults to gnu99 mode.
</p>

<p>Differences between all c* and gnu* modes:</p>
<ul>
<li>c* modes define "__STRICT_ANSI__".</li>
<li>Target-specific defines not prefixed by underscores, like "linux", are
defined in gnu* modes.</li>
<li>Trigraphs default to being off in gnu* modes; they can be enabled by the
-trigraphs option.</li>
<li>The parser recognizes "asm" and "typeof" as keywords in gnu* modes; the
variants "__asm__" and "__typeof__" are recognized in all modes.</li>
<li>The Apple "blocks" extension is recognized by default in gnu* modes
on some platforms; it can be enabled in any mode with the "-fblocks"
option.</li>
<li>Arrays that are VLA's according to the standard, but which can be constant
    folded by the frontend are treated as fixed size arrays.  This occurs for
    things like "int X[(1, 2)];", which is technically a VLA.  c* modes are
    strictly compliant and treat these as VLAs.</li>
</ul>

<p>Differences between *89 and *99 modes:</p>
<ul>
<li>The *99 modes default to implementing "inline" as specified in C99, while
the *89 modes implement the GNU version.  This can be overridden for individual
functions with the __gnu_inline__ attribute.</li>
<li>Digraphs are not recognized in c89 mode.</li>
<li>The scope of names defined inside a "for", "if", "switch", "while", or "do"
statement is different. (example: "if ((struct x {int x;}*)0) {}".)</li>
<li>__STDC_VERSION__ is not defined in *89 modes.</li>
<li>"inline" is not recognized as a keyword in c89 mode.</li>
<li>"restrict" is not recognized as a keyword in *89 modes.</li>
<li>Commas are allowed in integer constant expressions in *99 modes.</li>
<li>Arrays which are not lvalues are not implicitly promoted to pointers in
*89 modes.</li>
<li>Some warnings are different.</li>
</ul>

<p>c94 mode is identical to c89 mode except that digraphs are enabled in
c94 mode (FIXME: And __STDC_VERSION__ should be defined!).</p>

<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="c_unimpl_gcc">GCC extensions not implemented yet</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<p>clang tries to be compatible with gcc as much as possible, but some gcc
extensions are not implemented yet:</p>

<ul>

<li>clang does not support #pragma weak
(<a href="http://llvm.org/bugs/show_bug.cgi?id=3679">bug 3679</a>). Due to
the uses described in the bug, this is likely to be implemented at some
point, at least partially.</li>

<li>clang does not support decimal floating point types (_Decimal32 and
friends) or fixed-point types (_Fract and friends); nobody has expressed
interest in these features yet, so it's hard to say when they will be
implemented.</li>

<li>clang does not support nested functions; this is a complex feature which
is infrequently used, so it is unlikely to be implemented anytime soon.</li>

<li>clang does not support global register variables, this is unlikely
to be implemented soon because it requires additional LLVM backend support.
</li>

<li>clang does not support static initialization of flexible array
members. This appears to be a rarely used extension, but could be
implemented pending user demand.</li>

<li>clang does not support __builtin_va_arg_pack/__builtin_va_arg_pack_len.
This is used rarely, but in some potentially interesting places, like the
glibc headers, so it may be implemented pending user demand.  Note that
because clang pretends to be like GCC 4.2, and this extension was introduced
in 4.3, the glibc headers will not try to use this extension with clang at
the moment.</li>

<li>clang does not support the gcc extension for forward-declaring function
parameters; this has not showed up in any real-world code yet, though, so it
might never be implemented.</li>

</ul>

<p>This is not a complete list; if you find an unsupported extension
missing from this list, please send an e-mail to cfe-dev.  This list
currently excludes C++; see <a href="#cxx">C++ Language Features</a>.
Also, this list does not include bugs in mostly-implemented features; please
see the <a href="http://llvm.org/bugs/buglist.cgi?quicksearch=product%3Aclang+component%3A-New%2BBugs%2CAST%2CBasic%2CDriver%2CHeaders%2CLLVM%2BCodeGen%2Cparser%2Cpreprocessor%2CSemantic%2BAnalyzer">
bug tracker</a> for known existing bugs (FIXME: Is there a section for
bug-reporting guidelines somewhere?).</p>

<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="c_unsupp_gcc">Intentionally unsupported GCC extensions</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<ul>

<li>clang does not support the gcc extension that allows variable-length arrays
in structures.  This is for a few reasons: one, it is tricky
to implement, two, the extension is completely undocumented, and three, the
extension appears to be rarely used.  Note that clang <em>does</em> support
flexible array members (arrays with a zero or unspecified size at the end of
a structure).</li>

<li>clang does not have an equivalent to gcc's "fold"; this means that
clang doesn't accept some constructs gcc might accept in contexts where a
constant expression is required, like "x-x" where x is a variable.</li>

<li>clang does not support multiple alternative constraints in inline asm; this
is an extremely obscure feature which would be complicated to implement
correctly.</li>

<li>clang does not support __builtin_apply and friends; this extension is
extremely obscure and difficult to implement reliably.</li>

</ul>

<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="c_ms">Microsoft extensions</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<p>clang has some experimental support for extensions from
Microsoft Visual C++; to enable it, use the -fms-extensions command-line
option.  This is the default for Windows targets.  Note that the
support is incomplete; enabling Microsoft extensions will silently drop
certain constructs (including __declspec and Microsoft-style asm statements).
</p>

<ul>
<li>clang allows setting _MSC_VER with -fmsc-version=. It defaults to 1300 which
is the same as Visual C/C++ 2003. Any number is supported and can greatly affect
what Windows SDK and c++stdlib headers clang can compile. This option will be
removed when clang supports the full set of MS extensions required for these
headers.</li>

<li>clang does not support the Microsoft extension where anonymous
record members can be declared using user defined typedefs.</li>

<li>clang supports the Microsoft "#pragma pack" feature for
controlling record layout. GCC also contains support for this feature,
however where MSVC and GCC are incompatible clang follows the MSVC
definition.</li>
</ul>

<!-- ======================================================================= -->
<h2 id="target_features">Target-Specific Features and Limitations</h2>
<!-- ======================================================================= -->


<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="target_arch">CPU Architectures Features and Limitations</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<!-- ======================== -->
<h4 id="target_arch_x86">X86</h4>
<!-- ======================== -->

<p>The support for X86 (both 32-bit and 64-bit) is considered stable on Darwin
(Mac OS/X), Linux, FreeBSD, and Dragonfly BSD: it has been tested to correctly
compile many large C, C++, Objective-C, and Objective-C++ codebases.</p>

<p>On x86_64-mingw32, passing i128(by value) is incompatible to Microsoft x64
calling conversion. You might need to tweak WinX86_64ABIInfo::classify()
in lib/CodeGen/TargetInfo.cpp.</p>

<!-- ======================== -->
<h4 id="target_arch_arm">ARM</h4>
<!-- ======================== -->

<p>The support for ARM (specifically ARMv6 and ARMv7) is considered stable on
Darwin (iOS): it has been tested to correctly compile many large C, C++,
Objective-C, and Objective-C++ codebases.  Clang only supports a limited number
of ARM architectures. It does not yet fully support ARMv5, for example.</p>

<!-- ======================== -->
<h4 id="target_arch_other">Other platforms</h4>
<!-- ======================== -->
clang currently contains some support for PPC and Sparc; however, significant
pieces of code generation are still missing, and they haven't undergone
significant testing.

<p>clang contains limited support for the MSP430 embedded processor, but both
the clang support and the LLVM backend support are highly experimental.

<p>Other platforms are completely unsupported at the moment.  Adding the
minimal support needed for parsing and semantic analysis on a new platform
is quite easy; see lib/Basic/Targets.cpp in the clang source tree. This level
of support is also sufficient for conversion to LLVM IR for simple programs.
Proper support for conversion to LLVM IR requires adding code to
lib/CodeGen/CGCall.cpp at the moment; this is likely to change soon, though.
Generating assembly requires a suitable LLVM backend.

<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
<h3 id="target_os">Operating System Features and Limitations</h3>
<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->

<!-- ======================================= -->
<h4 id="target_os_darwin">Darwin (Mac OS/X)</h4>
<!-- ======================================= -->

<p>No __thread support, 64-bit ObjC support requires SL tools.</p>

<!-- ======================================= -->
<h4 id="target_os_win32">Windows</h4>
<!-- ======================================= -->

<p>Experimental supports are on Cygming.</p>

<h5>Cygwin</h5>

<p>Clang works on Cygwin-1.7.</p>

<h5>MinGW32</h5>

<p>Clang works on some mingw32 distributions.
Clang assumes directories as below;</p>

<ul>
<li><tt>C:/mingw/include</tt></li>
<li><tt>C:/mingw/lib</tt></li>
<li><tt>C:/mingw/lib/gcc/mingw32/4.[3-5].0/include/c++</tt></li>
</ul>

<p>On MSYS, a few tests might fail. It is due to <a href="http://llvm.org/bugs/show_bug.cgi?id=8520">Bug 8520</a> and is fixed in <a href="http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20110314/118106.html">LLVM's r127724</a>.</p>

<h5>MinGW-w64</h5>

<p>For x32(i686-w64-mingw32), it is not supported yet.</p>

<p>For x64(x86_64-w64-mingw32), <a href="http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20110321/118499.html">an essential patch(LLVM's r128206)</a> would be needed. It is incompatible to <a href="http://tdm-gcc.tdragon.net/development">TDM-GCC</a> due to the definition of symbol &quot;<code>___chkstk</code>&quot;. Clang assumes as below;<p>

<ul>
<li><tt>C:/mingw/x86_64-w64-mingw32/include</tt></li>
<li><tt>C:/mingw/x86_64-w64-mingw32/include/c++/4.5.[23]</tt></li>
<li>GCC driver &quot;gcc.exe&quot; to build x86_64-w64-mingw32 binary.</li>
</ul>

<p><a href="http://llvm.org/bugs/show_bug.cgi?id=8833">Some tests might fail</a>
on x64.</p>

</div>
</body>
</html>