Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / clang / e74dc199e95cc523c8806e0de41d8814e8bc5c42 / . / docs / UsersManual.html
blob: e4f44044c2faec2a17c65feee49adfb439154867 [file] [log] [blame]
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]3<html>
4<head>
5<title>Clang Compiler User's Manual</title>
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]6<link type="text/css" rel="stylesheet" href="../menu.css">
7<link type="text/css" rel="stylesheet" href="../content.css">
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]8<style type="text/css">
9td {
10 vertical-align: top;
11}
12</style>
13</head>
14<body>
15
16<!--#include virtual="../menu.html.incl"-->
17
18<div id="content">
19
20<h1>Clang Compiler User's Manual</h1>
21
22<ul>
23<li><a href="#intro">Introduction</a>
24 <ul>
25 <li><a href="#terminology">Terminology</a></li>
26 <li><a href="#basicusage">Basic Usage</a></li>
27 </ul>
28</li>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]29<li><a href="#commandline">Command Line Options</a>
30 <ul>
31 <li><a href="#cl_diagnostics">Options to Control Error and Warning
32 Messages</a></li>
Chad Rosierb211a4b2012-02-22 19:14:46 +0000[diff] [blame]33 <li><a href="#cl_crash_diagnostics">Options to Control Clang Crash
34 Diagnostics</a></li>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]35 </ul>
36</li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]37<li><a href="#general_features">Language and Target-Independent Features</a>
38 <ul>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]39 <li><a href="#diagnostics">Controlling Errors and Warnings</a>
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]40 <ul>
41 <li><a href="#diagnostics_display">Controlling How Clang Displays Diagnostics</a></li>
42 <li><a href="#diagnostics_mappings">Diagnostic Mappings</a></li>
Chris Lattner67db8cd2010-05-30 23:42:51 +0000[diff] [blame]43 <li><a href="#diagnostics_categories">Diagnostic Categories</a></li>
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]44 <li><a href="#diagnostics_commandline">Controlling Diagnostics via Command Line Flags</a></li>
45 <li><a href="#diagnostics_pragmas">Controlling Diagnostics via Pragmas</a></li>
Richard Smithf122a132012-06-13 20:27:03 +0000[diff] [blame]46 <li><a href="#diagnostics_systemheader">Controlling Diagnostics in System Headers</a></li>
Ted Kremenekf391fa72011-08-18 01:17:05 +0000[diff] [blame]47 <li><a href="#diagnostics_enable_everything">Enabling All Warnings</a></li>
Ted Kremenek2fb11eb2010-08-24 18:12:35 +0000[diff] [blame]48 <li><a href="#analyzer_diagnositics">Controlling Static Analyzer Diagnostics</a></li>
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]49 </ul>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]50 </li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]51 <li><a href="#precompiledheaders">Precompiled Headers</a></li>
Mike Stump53664ca2009-12-14 23:53:10 +0000[diff] [blame]52 <li><a href="#codegen">Controlling Code Generation</a></li>
Alexey Samsonov5ab52492012-05-30 06:55:10 +0000[diff] [blame]53 <li><a href="#debuginfosize">Controlling Size of Debug Information</a></li>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]54 </ul>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]55</li>
56<li><a href="#c">C Language Features</a>
57 <ul>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]58 <li><a href="#c_ext">Extensions supported by clang</a></li>
59 <li><a href="#c_modes">Differences between various standard modes</a></li>
60 <li><a href="#c_unimpl_gcc">GCC extensions not implemented yet</a></li>
61 <li><a href="#c_unsupp_gcc">Intentionally unsupported GCC extensions</a></li>
62 <li><a href="#c_ms">Microsoft extensions</a></li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]63 </ul>
64</li>
Richard Smithbdb10182011-11-21 20:54:59 +0000[diff] [blame]65<li><a href="#cxx">C++ Language Features</a>
66 <ul>
67 <li><a href="#cxx_implimits">Controlling implementation limits</a></li>
68 </ul>
69</li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]70<li><a href="#target_features">Target-Specific Features and Limitations</a>
71 <ul>
72 <li><a href="#target_arch">CPU Architectures Features and Limitations</a>
73 <ul>
74 <li><a href="#target_arch_x86">X86</a></li>
Eli Friedman3b658d32009-06-08 05:12:39 +0000[diff] [blame]75 <li><a href="#target_arch_arm">ARM</a></li>
76 <li><a href="#target_arch_other">Other platforms</a></li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]77 </ul>
78 </li>
79 <li><a href="#target_os">Operating System Features and Limitations</a>
80 <ul>
81 <li><a href="#target_os_darwin">Darwin (Mac OS/X)</a></li>
82 <li>Linux, etc.</li>
NAKAMURA Takumia75fdb22011-04-04 15:02:41 +0000[diff] [blame]83 <li><a href="#target_os_win32">Windows</a></li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]84 </ul>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]85 </li>
86 </ul>
87</li>
88</ul>
89
90
91<!-- ======================================================================= -->
92<h2 id="intro">Introduction</h2>
93<!-- ======================================================================= -->
94
95<p>The Clang Compiler is an open-source compiler for the C family of programming
96languages, aiming to be the best in class implementation of these languages.
97Clang builds on the LLVM optimizer and code generator, allowing it to provide
98high-quality optimization and code generation support for many targets. For
99more general information, please see the <a href="http://clang.llvm.org">Clang
100Web Site</a> or the <a href="http://llvm.org">LLVM Web Site</a>.</p>
101
102<p>This document describes important notes about using Clang as a compiler for
103an end-user, documenting the supported features, command line options, etc. If
104you are interested in using Clang to build a tool that processes code, please
105see <a href="InternalsManual.html">the Clang Internals Manual</a>. If you are
Michael J. Spencer8275fc02012-02-12 09:11:39 +0000[diff] [blame]106interested in the <a href="http://clang-analyzer.llvm.org">Clang
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]107Static Analyzer</a>, please see its web page.</p>
108
109<p>Clang is designed to support the C family of programming languages, which
110includes <a href="#c">C</a>, <a href="#objc">Objective-C</a>, <a
111href="#cxx">C++</a>, and <a href="#objcxx">Objective-C++</a> as well as many
112dialects of those. For language-specific information, please see the
113corresponding language specific section:</p>
114
115<ul>
116<li><a href="#c">C Language</a>: K&amp;R C, ANSI C89, ISO C90, ISO C94
117 (C89+AMD1), ISO C99 (+TC1, TC2, TC3). </li>
118<li><a href="#objc">Objective-C Language</a>: ObjC 1, ObjC 2, ObjC 2.1, plus
119 variants depending on base language.</li>
Richard Smithbdb10182011-11-21 20:54:59 +0000[diff] [blame]120<li><a href="#cxx">C++ Language</a></li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]121<li><a href="#objcxx">Objective C++ Language</a></li>
122</ul>
123
124<p>In addition to these base languages and their dialects, Clang supports a
125broad variety of language extensions, which are documented in the corresponding
126language section. These extensions are provided to be compatible with the GCC,
127Microsoft, and other popular compilers as well as to improve functionality
128through Clang-specific features. The Clang driver and language features are
129intentionally designed to be as compatible with the GNU GCC compiler as
130reasonably possible, easing migration from GCC to Clang. In most cases, code
131"just works".</p>
132
133<p>In addition to language specific features, Clang has a variety of features
134that depend on what CPU architecture or operating system is being compiled for.
Douglas Gregorcd5a5052009-11-09 15:15:41 +0000[diff] [blame]135Please see the <a href="#target_features">Target-Specific Features and
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]136Limitations</a> section for more details.</p>
137
138<p>The rest of the introduction introduces some basic <a
139href="#terminology">compiler terminology</a> that is used throughout this manual
140and contains a basic <a href="#basicusage">introduction to using Clang</a>
141as a command line compiler.</p>
142
143<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
144<h3 id="terminology">Terminology</h3>
145<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
146
147<p>Front end, parser, backend, preprocessor, undefined behavior, diagnostic,
148 optimizer</p>
149
150<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
151<h3 id="basicusage">Basic Usage</h3>
152<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
153
154<p>Intro to how to use a C compiler for newbies.</p>
155<p>
156compile + link
157
158compile then link
159
160debug info
161
162enabling optimizations
163
164picking a language to use, defaults to C99 by default. Autosenses based on
165extension.
166
167using a makefile
168</p>
169
170
171<!-- ======================================================================= -->
172<h2 id="commandline">Command Line Options</h2>
173<!-- ======================================================================= -->
174
175<p>
176This section is generally an index into other sections. It does not go into
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]177depth on the ones that are covered by other sections. However, the first part
178introduces the language selection and other high level options like -c, -g, etc.
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]179</p>
180
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]181
182<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
183<h3 id="cl_diagnostics">Options to Control Error and Warning Messages</h3>
184<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
185
186<p><b>-Werror</b>: Turn warnings into errors.</p>
187<p><b>-Werror=foo</b>: Turn warning "foo" into an error.</p>
188<p><b>-Wno-error=foo</b>: Turn warning "foo" into an warning even if -Werror is
189 specified.</p>
Dmitri Gribenko1dfb72f2012-07-17 18:51:47 +0000[diff] [blame]190<p><b>-Wfoo</b>: Enable warning "foo".</p>
191<p><b>-Wno-foo</b>: Disable warning "foo".</p>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]192<p><b>-w</b>: Disable all warnings.</p>
Dmitri Gribenko1dfb72f2012-07-17 18:51:47 +0000[diff] [blame]193<p><b>-Weverything</b>: <a href="#diagnostics_enable_everything">Enable <b>all</b> warnings.</a></p>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]194<p><b>-pedantic</b>: Warn on language extensions.</p>
195<p><b>-pedantic-errors</b>: Error on language extensions.</p>
196<p><b>-Wsystem-headers</b>: Enable warnings from system headers.</p>
197
Chris Lattner0f0c9632010-04-07 20:49:23 +0000[diff] [blame]198<p><b>-ferror-limit=123</b>: Stop emitting diagnostics after 123 errors have
199 been produced. The default is 20, and the error limit can be disabled with
200 -ferror-limit=0.</p>
201
Douglas Gregor575cf372010-04-20 07:18:24 +0000[diff] [blame]202<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>
203
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]204<!-- ================================================= -->
205<h4 id="cl_diag_formatting">Formatting of Diagnostics</h4>
206<!-- ================================================= -->
207
208<p>Clang aims to produce beautiful diagnostics by default, particularly for new
209users that first come to Clang. However, different people have different
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]210preferences, and sometimes Clang is driven by another program that wants to
211parse simple and consistent output, not a person. For these cases, Clang
212provides a wide range of options to control the exact output format of the
213diagnostics that it generates.</p>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]214
215<dl>
216
217<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
218<dt id="opt_fshow-column"><b>-f[no-]show-column</b>: Print column number in
219diagnostic.</dt>
220<dd>This option, which defaults to on, controls whether or not Clang prints the
221column number of a diagnostic. For example, when this is enabled, Clang will
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]222print something like:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]223
224<pre>
225 test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
226 #endif bad
227 ^
228 //
229</pre>
230
231<p>When this is disabled, Clang will print "test.c:28: warning..." with no
232column number.</p>
Jordan Rosebbe01752012-07-20 18:50:51 +0000[diff] [blame]233
234<p>The printed column numbers count bytes from the beginning of the line; take
235care if your source contains multibyte characters.</p>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]236</dd>
237
238<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
239<dt id="opt_fshow-source-location"><b>-f[no-]show-source-location</b>: Print
240source file/line/column information in diagnostic.</dt>
241<dd>This option, which defaults to on, controls whether or not Clang prints the
242filename, line number and column number of a diagnostic. For example,
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]243when this is enabled, Clang will print something like:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]244
245<pre>
246 test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
247 #endif bad
248 ^
249 //
250</pre>
251
252<p>When this is disabled, Clang will not print the "test.c:28:8: " part.</p>
253</dd>
254
255<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
256<dt id="opt_fcaret-diagnostics"><b>-f[no-]caret-diagnostics</b>: Print source
257line and ranges from source code in diagnostic.</dt>
258<dd>This option, which defaults to on, controls whether or not Clang prints the
259source line, source ranges, and caret when emitting a diagnostic. For example,
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]260when this is enabled, Clang will print something like:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]261
262<pre>
263 test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
264 #endif bad
265 ^
266 //
267</pre>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]268</dd>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]269<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
Douglas Gregore8d44dd2010-07-09 16:31:58 +0000[diff] [blame]270<dt id="opt_fcolor_diagnostics"><b>-f[no-]color-diagnostics</b>: </dt>
271<dd>This option, which defaults to on when a color-capable terminal is
272 detected, controls whether or not Clang prints diagnostics in color.
273 When this option is enabled, Clang will use colors to highlight
274 specific parts of the diagnostic, e.g.,
Chris Lattner4e1c53d2011-01-24 03:47:34 +0000[diff] [blame]275 <pre>
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]276 <b><span style="color:black">test.c:28:8: <span style="color:magenta">warning</span>: extra tokens at end of #endif directive [-Wextra-tokens]</span></b>
Chris Lattner4e1c53d2011-01-24 03:47:34 +0000[diff] [blame]277 #endif bad
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]278 <span style="color:green">^</span>
279 <span style="color:green">//</span>
Douglas Gregore8d44dd2010-07-09 16:31:58 +0000[diff] [blame]280</pre>
Chris Lattner4e1c53d2011-01-24 03:47:34 +0000[diff] [blame]281
282<p>When this is disabled, Clang will just print:</p>
283
284<pre>
285 test.c:2:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
286 #endif bad
287 ^
288 //
289</pre>
290</dd>
Douglas Gregore8d44dd2010-07-09 16:31:58 +0000[diff] [blame]291<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
Douglas Gregorc9471b02011-05-21 17:07:29 +0000[diff] [blame]292<dt id="opt_fdiagnostics-format"><b>-fdiagnostics-format=clang/msvc/vi</b>:
293Changes diagnostic output format to better match IDEs and command line tools.</dt>
294<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:
295
296 <dl>
297 <dt><b>clang</b> (default)</dt>
298 <dd>
299 <pre>t.c:3:11: warning: conversion specifies type 'char *' but the argument has type 'int'</pre>
300 </dd>
301
302 <dt><b>msvc</b></dt>
303 <dd>
304 <pre>t.c(3,11) : warning: conversion specifies type 'char *' but the argument has type 'int'</pre>
305 </dd>
306
307 <dt><b>vi</b></dt>
308 <dd>
309 <pre>t.c +3:11: warning: conversion specifies type 'char *' but the argument has type 'int'</pre>
310 </dd>
311 </dl>
312</dd>
313
314<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
Argyrios Kyrtzidis477aab62011-05-25 05:05:01 +0000[diff] [blame]315<dt id="opt_fdiagnostics-show-name"><b>-f[no-]diagnostics-show-name</b>:
316Enable the display of the diagnostic name.</dt>
317<dd>This option, which defaults to off, controls whether or not
Benjamin Kramer665a8dc2012-01-15 15:26:07 +0000[diff] [blame]318Clang prints the associated name.<p></p></dd>
Argyrios Kyrtzidis477aab62011-05-25 05:05:01 +0000[diff] [blame]319<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]320<dt id="opt_fdiagnostics-show-option"><b>-f[no-]diagnostics-show-option</b>:
321Enable <tt>[-Woption]</tt> information in diagnostic line.</dt>
322<dd>This option, which defaults to on,
323controls whether or not Clang prints the associated <A
324href="#cl_diag_warning_groups">warning group</a> option name when outputting
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]325a warning diagnostic. For example, in this output:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]326
327<pre>
328 test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
329 #endif bad
330 ^
331 //
332</pre>
333
334<p>Passing <b>-fno-diagnostics-show-option</b> will prevent Clang from printing
335the [<a href="#opt_Wextra-tokens">-Wextra-tokens</a>] information in the
336diagnostic. This information tells you the flag needed to enable or disable the
337diagnostic, either from the command line or through <a
338href="#pragma_GCC_diagnostic">#pragma GCC diagnostic</a>.</dd>
339
Chris Lattner28a43a42010-05-05 01:35:28 +0000[diff] [blame]340<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
341<dt id="opt_fdiagnostics-show-category"><b>-fdiagnostics-show-category=none/id/name</b>:
342Enable printing category information in diagnostic line.</dt>
343<dd>This option, which defaults to "none",
344controls whether or not Clang prints the category associated with a diagnostic
345when emitting it. Each diagnostic may or many not have an associated category,
346if it has one, it is listed in the diagnostic categorization field of the
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]347diagnostic line (in the []'s).
Chris Lattner28a43a42010-05-05 01:35:28 +0000[diff] [blame]348
349<p>For example, a format string warning will produce these three renditions
350based on the setting of this option:</p>
351
352<pre>
353 t.c:3:11: warning: conversion specifies type 'char *' but the argument has type 'int' [-Wformat]
354 t.c:3:11: warning: conversion specifies type 'char *' but the argument has type 'int' [-Wformat<b>,1</b>]
355 t.c:3:11: warning: conversion specifies type 'char *' but the argument has type 'int' [-Wformat<b>,Format String</b>]
356</pre>
357
358<p>This category can be used by clients that want to group diagnostics by
359category, so it should be a high level category. We want dozens of these, not
360hundreds or thousands of them.</p>
361</dd>
362
363
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]364
365<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
366<dt id="opt_fdiagnostics-fixit-info"><b>-f[no-]diagnostics-fixit-info</b>:
367Enable "FixIt" information in the diagnostics output.</dt>
368<dd>This option, which defaults to on, controls whether or not Clang prints the
369information on how to fix a specific diagnostic underneath it when it knows.
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]370For example, in this output:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]371
372<pre>
373 test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
374 #endif bad
375 ^
376 //
377</pre>
378
379<p>Passing <b>-fno-diagnostics-fixit-info</b> will prevent Clang from printing
380the "//" line at the end of the message. This information is useful for users
381who may not understand what is wrong, but can be confusing for machine
382parsing.</p>
383</dd>
384
385<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
Chris Lattner2a9cc232009-04-21 05:35:32 +0000[diff] [blame]386<dt id="opt_fdiagnostics-print-source-range-info">
387<b>-f[no-]diagnostics-print-source-range-info</b>:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]388Print machine parsable information about source ranges.</dt>
389<dd>This option, which defaults to off, controls whether or not Clang prints
390information about source ranges in a machine parsable format after the
391file/line/column number information. The information is a simple sequence of
392brace enclosed ranges, where each range lists the start and end line/column
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]393locations. For example, in this output:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]394
395<pre>
396exprs.c:47:15:{47:8-47:14}{47:17-47:24}: error: invalid operands to binary expression ('int *' and '_Complex float')
397 P = (P-42) + Gamma*4;
398 ~~~~~~ ^ ~~~~~~~
399</pre>
400
Chris Lattner2a9cc232009-04-21 05:35:32 +0000[diff] [blame]401<p>The {}'s are generated by -fdiagnostics-print-source-range-info.</p>
Jordan Rosebbe01752012-07-20 18:50:51 +0000[diff] [blame]402
403<p>The printed column numbers count bytes from the beginning of the line; take
404care if your source contains multibyte characters.</p>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]405</dd>
406
Douglas Gregor4786c152010-08-19 20:24:43 +0000[diff] [blame]407<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
408<dt id="opt_fdiagnostics-parseable-fixits">
409<b>-fdiagnostics-parseable-fixits</b>:
410Print Fix-Its in a machine parseable form.</dt>
411<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>
412
413<pre>
Douglas Gregorbf5e09d2010-08-20 03:17:33 +0000[diff] [blame]414 fix-it:"t.cpp":{7:25-7:29}:"Gamma"
Douglas Gregor4786c152010-08-19 20:24:43 +0000[diff] [blame]415</pre>
416
NAKAMURA Takumi44626362011-04-05 00:57:02 +0000[diff] [blame]417<p>The range printed is a half-open range, so in this example the characters at
418column 25 up to but not including column 29 on line 7 in t.cpp should be
419replaced with the string &quot;Gamma&quot;. Either the range or the replacement
420string may be empty (representing strict insertions and strict erasures,
421respectively). Both the file name and the insertion string escape backslash (as
422&quot;\\&quot;), tabs (as &quot;\t&quot;), newlines (as &quot;\n&quot;), double
423quotes(as &quot;\&quot;&quot;) and non-printable characters (as octal
424&quot;\xxx&quot;).</p>
Jordan Rosebbe01752012-07-20 18:50:51 +0000[diff] [blame]425
426<p>The printed column numbers count bytes from the beginning of the line; take
427care if your source contains multibyte characters.</p>
Douglas Gregor4786c152010-08-19 20:24:43 +0000[diff] [blame]428</dd>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]429
Richard Trieue59331a2012-06-27 02:00:20 +0000[diff] [blame]430<dt id="opt_fno-elide-type">
431<b>-fno-elide-type</b>:
Dmitri Gribenko1dfb72f2012-07-17 18:51:47 +0000[diff] [blame]432Turns off elision in template type printing.</dt>
Richard Trieue59331a2012-06-27 02:00:20 +0000[diff] [blame]433<dd><p>The default for template type printing is to elide as many template
434arguments as possible, removing those which are the same in both template types,
435leaving only the differences. Adding this flag will print all the template
436arguments. If supported by the terminal, highlighting will still appear on
437differing arguments.</p>
438
439Default:
440<pre>
441t.cc:4:5: <span class="note">note</span>: candidate function not viable: no known conversion from 'vector&lt;map&lt;[...], map&lt;<span class="template-highlight">float</span>, [...]&gt;&gt;&gt;' to 'vector&lt;map&lt;[...], map&lt;<span class="template-highlight">double</span>, [...]&gt;&gt;&gt;' for 1st argument;
442</pre>
443-fno-elide-type:
444<pre>
445t.cc:4:5: <span class="note">note</span>: candidate function not viable: no known conversion from 'vector&lt;map&lt;int, map&lt;<span class="template-highlight">float</span>, int&gt;&gt;&gt;' to 'vector&lt;map&lt;int, map&lt;<span class="template-highlight">double</span>, int&gt;&gt;&gt;' for 1st argument;
446</pre>
447</dd>
448
449<dt id="opt_fdiagnostics-show-template-tree">
450<b>-fdiagnostics-show-template-tree</b>:
Dmitri Gribenko1dfb72f2012-07-17 18:51:47 +0000[diff] [blame]451Template type diffing prints a text tree.</dt>
Richard Trieue59331a2012-06-27 02:00:20 +0000[diff] [blame]452<dd><p>For diffing large templated types, this option will cause Clang to
453display the templates as an indented text tree, one argument per line, with
454differences marked inline. This is compatible with -fno-elide-type.</p>
455
456Default:
457<pre>
458t.cc:4:5: <span class="note">note</span>: candidate function not viable: no known conversion from 'vector&lt;map&lt;[...], map&lt;<span class="template-highlight">float</span>, [...]&gt;&gt;&gt;' to 'vector&lt;map&lt;[...], map&lt;<span class="template-highlight">double</span>, [...]&gt;&gt;&gt;' for 1st argument;
459</pre>
460-fdiagnostics-show-template-tree
461<pre>
462t.cc:4:5: <span class="note">note</span>: candidate function not viable: no known conversion for 1st argument;
463 vector&lt;
464 map&lt;
465 [...],
466 map&lt;
467 [<span class="template-highlight">float</span> != <span class="template-highlight">float</span>],
468 [...]&gt;&gt;&gt;
469</pre>
470</dd>
471
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]472</dl>
473
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]474
475
476<!-- ===================================================== -->
477<h4 id="cl_diag_warning_groups">Individual Warning Groups</h4>
478<!-- ===================================================== -->
479
480<p>TODO: Generate this from tblgen. Define one anchor per warning group.</p>
481
482
483<dl>
484
485
486<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
487<dt id="opt_Wextra-tokens"><b>-Wextra-tokens</b>: Warn about excess tokens at
488 the end of a preprocessor directive.</dt>
489<dd>This option, which defaults to on, enables warnings about extra tokens at
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]490the end of preprocessor directives. For example:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]491
492<pre>
493 test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
494 #endif bad
495 ^
496</pre>
497
498<p>These extra tokens are not strictly conforming, and are usually best handled
499by commenting them out.</p>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]500</dd>
501
Jeffrey Yasskin21d07e42010-06-05 01:39:57 +0000[diff] [blame]502<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
503<dt id="opt_Wambiguous-member-template"><b>-Wambiguous-member-template</b>:
504Warn about unqualified uses of a member template whose name resolves
505to another template at the location of the use.</dt>
506<dd>This option, which defaults to on, enables a warning in the
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]507following code:
Jeffrey Yasskin21d07e42010-06-05 01:39:57 +0000[diff] [blame]508
509<pre>
510template&lt;typename T> struct set{};
511template&lt;typename T> struct trait { typedef const T& type; };
512struct Value {
513 template&lt;typename T> void set(typename trait&lt;T>::type value) {}
514};
515void foo() {
516 Value v;
517 v.set&lt;double>(3.2);
518}
519</pre>
520
521<p>C++ [basic.lookup.classref] requires this to be an error, but,
522because it's hard to work around, Clang downgrades it to a warning as
523an extension.</p>
524</dd>
525
Jeffrey Yasskin57d12fd2010-06-07 15:58:05 +0000[diff] [blame]526<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
527<dt id="opt_Wbind-to-temporary-copy"><b>-Wbind-to-temporary-copy</b>: Warn about
528an unusable copy constructor when binding a reference to a temporary.</dt>
529<dd>This option, which defaults to on, enables warnings about binding a
530reference to a temporary when the temporary doesn't have a usable copy
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]531constructor. For example:
Jeffrey Yasskin57d12fd2010-06-07 15:58:05 +0000[diff] [blame]532
533<pre>
534 struct NonCopyable {
535 NonCopyable();
536 private:
537 NonCopyable(const NonCopyable&);
538 };
539 void foo(const NonCopyable&);
540 void bar() {
David Blaikie5090e9f2011-10-18 05:49:30 +0000[diff] [blame]541 foo(NonCopyable()); // Disallowed in C++98; allowed in C++11.
Jeffrey Yasskin57d12fd2010-06-07 15:58:05 +0000[diff] [blame]542 }
543</pre>
544<pre>
545 struct NonCopyable2 {
546 NonCopyable2();
547 NonCopyable2(NonCopyable2&);
548 };
549 void foo(const NonCopyable2&);
550 void bar() {
David Blaikie5090e9f2011-10-18 05:49:30 +0000[diff] [blame]551 foo(NonCopyable2()); // Disallowed in C++98; allowed in C++11.
Jeffrey Yasskin57d12fd2010-06-07 15:58:05 +0000[diff] [blame]552 }
553</pre>
554
555<p>Note that if <tt>NonCopyable2::NonCopyable2()</tt> has a default
556argument whose instantiation produces a compile error, that error will
557still be a hard error in C++98 mode even if this warning is turned
558off.</p>
559
560</dd>
561
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]562</dl>
563
Chad Rosiercd376a12012-02-22 20:26:10 +0000[diff] [blame]564<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
565<h3 id="cl_crash_diagnostics">Options to Control Clang Crash Diagnostics</h3>
566<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
567
568<p>As unbelievable as it may sound, Clang does crash from time to time.
569Generally, this only occurs to those living on the
570<a href="http://llvm.org/releases/download.html#svn">bleeding edge</a>. Clang
571goes to great lengths to assist you in filing a bug report. Specifically, Clang
572generates preprocessed source file(s) and associated run script(s) upon a
573crash. These files should be attached to a bug report to ease reproducibility
574of the failure. Below are the command line options to control the crash
575diagnostics.
576</p>
577
578<p><b>-fno-crash-diagnostics</b>: Disable auto-generation of preprocessed
579source files during a clang crash.</p>
580
581<p>The -fno-crash-diagnostics flag can be helpful for speeding the process of
582generating a delta reduced test case.</p>
583
584
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]585<!-- ======================================================================= -->
586<h2 id="general_features">Language and Target-Independent Features</h2>
587<!-- ======================================================================= -->
588
589
590<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
591<h3 id="diagnostics">Controlling Errors and Warnings</h3>
592<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
593
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]594<p>Clang provides a number of ways to control which code constructs cause it to
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]595emit errors and warning messages, and how they are displayed to the console.</p>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]596
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]597<h4 id="diagnostics_display">Controlling How Clang Displays Diagnostics</h4>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]598
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]599<p>When Clang emits a diagnostic, it includes rich information in the output,
600and gives you fine-grain control over which information is printed. Clang has
601the ability to print this information, and these are the options that control
602it:</p>
603
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]604<ol>
605<li>A file/line/column indicator that shows exactly where the diagnostic occurs
606 in your code [<a href="#opt_fshow-column">-fshow-column</a>, <a
607 href="#opt_fshow-source-location">-fshow-source-location</a>].</li>
608<li>A categorization of the diagnostic as a note, warning, error, or fatal
609 error.</li>
610<li>A text string that describes what the problem is.</li>
611<li>An option that indicates how to control the diagnostic (for diagnostics that
612 support it) [<a
613 href="#opt_fdiagnostics-show-option">-fdiagnostics-show-option</a>].</li>
Chris Lattner3f145382010-05-24 21:35:18 +0000[diff] [blame]614<li>A <a href="#diagnostics_categories">high-level category</a> for the
615 diagnostic for clients that want to group diagnostics by class (for
616 diagnostics that support it) [<a
Chris Lattner28a43a42010-05-05 01:35:28 +0000[diff] [blame]617 href="#opt_fdiagnostics-show-category">-fdiagnostics-show-category</a>].</li>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]618<li>The line of source code that the issue occurs on, along with a caret and
619 ranges that indicate the important locations [<a
620 href="opt_fcaret-diagnostics">-fcaret-diagnostics</a>].</li>
621<li>"FixIt" information, which is a concise explanation of how to fix the
622 problem (when Clang is certain it knows) [<a
623 href="opt_fdiagnostics-fixit-info">-fdiagnostics-fixit-info</a>].</li>
624<li>A machine-parsable representation of the ranges involved (off by
625 default) [<a
Chris Lattner2a9cc232009-04-21 05:35:32 +0000[diff] [blame]626 href="opt_fdiagnostics-print-source-range-info">-fdiagnostics-print-source-range-info</a>].</li>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]627</ol>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]628
629<p>For more information please see <a href="#cl_diag_formatting">Formatting of
630Diagnostics</a>.</p>
631
Chris Lattner3f145382010-05-24 21:35:18 +0000[diff] [blame]632
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]633<h4 id="diagnostics_mappings">Diagnostic Mappings</h4>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]634
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]635<p>All diagnostics are mapped into one of these 5 classes:</p>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]636
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]637<ul>
638<li>Ignored</li>
639<li>Note</li>
640<li>Warning</li>
641<li>Error</li>
642<li>Fatal</li>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]643</ul>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]644
Chris Lattner3f145382010-05-24 21:35:18 +0000[diff] [blame]645<h4 id="diagnostics_categories">Diagnostic Categories</h4>
646
647<p>Though not shown by default, diagnostics may each be associated with a
648 high-level category. This category is intended to make it possible to triage
649 builds that produce a large number of errors or warnings in a grouped way.
650</p>
651
652<p>Categories are not shown by default, but they can be turned on with the
653<a href="#opt_fdiagnostics-show-category">-fdiagnostics-show-category</a> option.
654When set to "<tt>name</tt>", the category is printed textually in the diagnostic
655output. When it is set to "<tt>id</tt>", a category number is printed. The
656mapping of category names to category id's can be obtained by running '<tt>clang
657 --print-diagnostic-categories</tt>'.
658</p>
659
660<h4 id="diagnostics_commandline">Controlling Diagnostics via Command Line
661 Flags</h4>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]662
Dmitri Gribenko1dfb72f2012-07-17 18:51:47 +0000[diff] [blame]663<p>TODO: -W flags, -pedantic, etc</p>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]664
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]665<h4 id="diagnostics_pragmas">Controlling Diagnostics via Pragmas</h4>
666
667<p>Clang can also control what diagnostics are enabled through the use of
668pragmas in the source code. This is useful for turning off specific warnings
669in a section of source code. Clang supports GCC's pragma for compatibility
670with existing source code, as well as several extensions. </p>
671
672<p>The pragma may control any warning that can be used from the command line.
673Warnings may be set to ignored, warning, error, or fatal. The following
674example code will tell Clang or GCC to ignore the -Wall warnings:</p>
675
676<pre>
677#pragma GCC diagnostic ignored "-Wall"
678</pre>
679
Chris Lattnerd462b6a2011-01-24 03:47:59 +0000[diff] [blame]680<p>In addition to all of the functionality provided by GCC's pragma, Clang
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]681also allows you to push and pop the current warning state. This is particularly
682useful when writing a header file that will be compiled by other people, because
683you don't know what warning flags they build with.</p>
684
685<p>In the below example
686-Wmultichar is ignored for only a single line of code, after which the
687diagnostics return to whatever state had previously existed.</p>
688
689<pre>
690#pragma clang diagnostic push
691#pragma clang diagnostic ignored "-Wmultichar"
692
693char b = 'df'; // no warning.
694
695#pragma clang diagnostic pop
696</pre>
697
698<p>The push and pop pragmas will save and restore the full diagnostic state of
699the compiler, regardless of how it was set. That means that it is possible to
700use push and pop around GCC compatible diagnostics and Clang will push and pop
701them appropriately, while GCC will ignore the pushes and pops as unknown
702pragmas. It should be noted that while Clang supports the GCC pragma, Clang and
703GCC do not support the exact same set of warnings, so even when using GCC
704compatible #pragmas there is no guarantee that they will have identical behaviour
705on both compilers. </p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]706
Richard Smithf122a132012-06-13 20:27:03 +0000[diff] [blame]707<h4 id="diagnostics_systemheader">Controlling Diagnostics in System Headers</h4>
708
709<p>Warnings are suppressed when they occur in system headers. By default, an
710included file is treated as a system header if it is found in an include path
711specified by <tt>-isystem</tt>, but this can be overridden in several ways.</p>
712
713<p>The <tt>system_header</tt> pragma can be used to mark the current file as
714being a system header. No warnings will be produced from the location of the
715pragma onwards within the same file.</p>
716
717<pre>
718char a = 'xy'; // warning
719
720#pragma clang system_header
721
722char b = 'ab'; // no warning
723</pre>
724
725<p>The <tt>-isystem-prefix</tt> and <tt>-ino-system-prefix</tt> command-line
726arguments can be used to override whether subsets of an include path are treated
727as system headers. When the name in a <tt>#include</tt> directive is found
728within a header search path and starts with a system prefix, the header is
729treated as a system header. The last prefix on the command-line which matches
730the specified header name takes precedence. For instance:</p>
731
732<pre>
733clang -Ifoo -isystem bar -isystem-prefix x/ -ino-system-prefix x/y/
734</pre>
735
736<p>Here, <tt>#include "x/a.h"</tt> is treated as including a system header, even
737if the header is found in <tt>foo</tt>, and <tt>#include "x/y/b.h"</tt> is
738treated as not including a system header, even if the header is found in
739<tt>bar</tt>.
740</p>
741
742<p>A <tt>#include</tt> directive which finds a file relative to the current
743directory is treated as including a system header if the including file is
744treated as a system header.</p>
745
Ted Kremenekf391fa72011-08-18 01:17:05 +0000[diff] [blame]746<h4 id="diagnostics_enable_everything">Enabling All Warnings</h4>
747
748<p>In addition to the traditional <tt>-W</tt> flags, one can enable <b>all</b>
749 warnings by passing <tt>-Weverything</tt>.
750 This works as expected with <tt>-Werror</tt>,
751 and also includes the warnings from <tt>-pedantic</tt>.</p>
752
753<p>Note that when combined with <tt>-w</tt> (which disables all warnings), that
754 flag wins.</p>
755
Ted Kremenek2fb11eb2010-08-24 18:12:35 +0000[diff] [blame]756<h4 id="analyzer_diagnositics">Controlling Static Analyzer Diagnostics</h4>
757
758<p>While not strictly part of the compiler, the diagnostics from Clang's <a
759href="http://clang-analyzer.llvm.org">static analyzer</a> can also be influenced
Sylvestre Ledru43e3dee2012-07-31 06:56:50 +0000[diff] [blame]760by the user via changes to the source code. See the available
Anna Zaks843f0e82012-06-16 00:30:21 +0000[diff] [blame]761<a href = "http://clang-analyzer.llvm.org/annotations.html" >annotations</a> and
762the analyzer's
763<a href= "http://clang-analyzer.llvm.org/faq.html#exclude_code" >FAQ page</a> for
764more information.
Ted Kremenek2fb11eb2010-08-24 18:12:35 +0000[diff] [blame]765
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]766<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
767<h3 id="precompiledheaders">Precompiled Headers</h3>
768<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
769
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]770<p><a href="http://en.wikipedia.org/wiki/Precompiled_header">Precompiled
771headers</a> are a general approach employed by many compilers to reduce
772compilation time. The underlying motivation of the approach is that it is
773common for the same (and often large) header files to be included by
774multiple source files. Consequently, compile times can often be greatly improved
775by caching some of the (redundant) work done by a compiler to process headers.
776Precompiled header files, which represent one of many ways to implement
777this optimization, are literally files that represent an on-disk cache that
778contains the vital information necessary to reduce some of the work
779needed to process a corresponding header file. While details of precompiled
Chris Lattnerd462b6a2011-01-24 03:47:59 +0000[diff] [blame]780headers vary between compilers, precompiled headers have been shown to be
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]781highly effective at speeding up program compilation on systems with very large
782system headers (e.g., Mac OS/X).</p>
783
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]784<h4>Generating a PCH File</h4>
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]785
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]786<p>To generate a PCH file using Clang, one invokes Clang with
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]787the <b><tt>-x <i>&lt;language&gt;</i>-header</tt></b> option. This mirrors the
788interface in GCC for generating PCH files:</p>
789
790<pre>
791 $ gcc -x c-header test.h -o test.h.gch
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]792 $ clang -x c-header test.h -o test.h.pch
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]793</pre>
794
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]795<h4>Using a PCH File</h4>
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]796
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]797<p>A PCH file can then be used as a prefix header when a
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]798<b><tt>-include</tt></b> option is passed to <tt>clang</tt>:</p>
799
800<pre>
801 $ clang -include test.h test.c -o test
802</pre>
803
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]804<p>The <tt>clang</tt> driver will first check if a PCH file for <tt>test.h</tt>
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]805is available; if so, the contents of <tt>test.h</tt> (and the files it includes)
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]806will be processed from the PCH file. Otherwise, Clang falls back to
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]807directly processing the content of <tt>test.h</tt>. This mirrors the behavior of
808GCC.</p>
809
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]810<p><b>NOTE:</b> Clang does <em>not</em> automatically use PCH files
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]811for headers that are directly included within a source file. For example:</p>
812
813<pre>
Chris Lattnere42ec542009-06-13 20:35:58 +0000[diff] [blame]814 $ clang -x c-header test.h -o test.h.pch
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]815 $ cat test.c
816 #include "test.h"
817 $ clang test.c -o test
818</pre>
819
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]820<p>In this example, <tt>clang</tt> will not automatically use the PCH file for
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]821<tt>test.h</tt> since <tt>test.h</tt> was included directly in the source file
822and not specified on the command line using <tt>-include</tt>.</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]823
Douglas Gregore650c8c2009-07-07 00:12:59 +0000[diff] [blame]824<h4>Relocatable PCH Files</h4>
825<p>It is sometimes necessary to build a precompiled header from headers that
826are not yet in their final, installed locations. For example, one might build a
827precompiled header within the build tree that is then meant to be installed
828alongside the headers. Clang permits the creation of "relocatable" precompiled
829headers, which are built with a given path (into the build directory) and can
830later be used from an installed location.</p>
831
832<p>To build a relocatable precompiled header, place your headers into a
833subdirectory whose structure mimics the installed location. For example, if you
834want to build a precompiled header for the header <code>mylib.h</code> that
835will be installed into <code>/usr/include</code>, create a subdirectory
836<code>build/usr/include</code> and place the header <code>mylib.h</code> into
837that subdirectory. If <code>mylib.h</code> depends on other headers, then
838they can be stored within <code>build/usr/include</code> in a way that mimics
839the installed location.</p>
840
841<p>Building a relocatable precompiled header requires two additional arguments.
842First, pass the <code>--relocatable-pch</code> flag to indicate that the
843resulting PCH file should be relocatable. Second, pass
844<code>-isysroot /path/to/build</code>, which makes all includes for your
845library relative to the build directory. For example:</p>
846
847<pre>
848 # clang -x c-header --relocatable-pch -isysroot /path/to/build /path/to/build/mylib.h mylib.h.pch
849</pre>
850
851<p>When loading the relocatable PCH file, the various headers used in the PCH
852file are found from the system header root. For example, <code>mylib.h</code>
853can be found in <code>/usr/include/mylib.h</code>. If the headers are installed
854in some other system root, the <code>-isysroot</code> option can be used provide
855a different system root from which the headers will be based. For example,
856<code>-isysroot /Developer/SDKs/MacOSX10.4u.sdk</code> will look for
857<code>mylib.h</code> in
858<code>/Developer/SDKs/MacOSX10.4u.sdk/usr/include/mylib.h</code>.</p>
859
860<p>Relocatable precompiled headers are intended to be used in a limited number
861of cases where the compilation environment is tightly controlled and the
862precompiled header cannot be generated after headers have been installed.
863Relocatable precompiled headers also have some performance impact, because
864the difference in location between the header locations at PCH build time vs.
865at the time of PCH use requires one of the PCH optimizations,
866<code>stat()</code> caching, to be disabled. However, this change is only
867likely to affect PCH files that reference a large number of headers.</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]868
Mike Stump53664ca2009-12-14 23:53:10 +0000[diff] [blame]869<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
870<h3 id="codegen">Controlling Code Generation</h3>
871<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
872
873<p>Clang provides a number of ways to control code generation. The options are listed below.</p>
874
875<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]876<dl>
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]877<dt id="opt_fsanitize"><b>-fsanitize=check1,check2</b>: Turn on runtime checks
Will Dietzb8540362012-11-27 15:01:55 +0000[diff] [blame]878for various forms of undefined or suspicious behavior.</dt>
Mike Stump53664ca2009-12-14 23:53:10 +0000[diff] [blame]879
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]880<dd>This option controls whether Clang adds runtime checks for various forms of
Will Dietzb8540362012-11-27 15:01:55 +0000[diff] [blame]881undefined or suspicious behavior, and is disabled by default. If a check
882fails, a diagnostic message is produced at runtime explaining the problem. The
883main checks are:
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]884
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]885<ul>
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]886<li id="opt_fsanitize_address"><tt>-fsanitize=address</tt>:
887 <a href="AddressSanitizer.html">AddressSanitizer</a>, a memory error
888 detector.</li>
Alexey Samsonov4d1a6e42012-11-29 22:36:21 +0000[diff] [blame]889<li id="opt_fsanitize_address_full"><tt>-fsanitize=address-full</tt>:
890 AddressSanitizer with all the experimental features listed below.
Will Dietzb8540362012-11-27 15:01:55 +0000[diff] [blame]891<li id="opt_fsanitize_integer"><tt>-fsanitize=integer</tt>:
892 Enables checks for undefined or suspicious integer behavior.</li>
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]893<li id="opt_fsanitize_thread"><tt>-fsanitize=thread</tt>:
894 <a href="ThreadSanitizer.html">ThreadSanitizer</a>, an <em>experimental</em>
895 data race detector. Not ready for widespread use.</li>
896<li id="opt_fsanitize_undefined"><tt>-fsanitize=undefined</tt>:
Will Dietzb8540362012-11-27 15:01:55 +0000[diff] [blame]897 Fast and compatible undefined behavior checker. Enables the undefined behavior
898 checks that have small runtime cost and no impact on address space layout
899 or ABI. This includes all of the checks listed below other than unsigned
900 integer overflow.</li>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]901</ul>
Richard Smith23fee342012-08-25 00:44:02 +0000[diff] [blame]902
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]903The following more fine-grained checks are also available:
Richard Smith23fee342012-08-25 00:44:02 +0000[diff] [blame]904
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]905<ul>
906<li id="opt_fsanitize_alignment"><tt>-fsanitize=alignment</tt>:
907 Use of a misaligned pointer or creation of a misaligned reference.</li>
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]908<li id="opt_fsanitize_float-cast-overflow"><tt>-fsanitize=float-cast-overflow</tt>:
909 Conversion to, from, or between floating-point types which would overflow
910 the destination.</li>
Will Dietzb8540362012-11-27 15:01:55 +0000[diff] [blame]911<li id="opt_fsanitize_float-divide-by-zero"><tt>-fsanitize=float-divide-by-zero</tt>:
912 Floating point division by zero.</li>
913<li id="opt_fsanitize_integer-divide-by-zero"><tt>-fsanitize=integer-divide-by-zero</tt>:
914 Integer division by zero.</li>
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]915<li id="opt_fsanitize_null"><tt>-fsanitize=null</tt>:
916 Use of a null pointer or creation of a null reference.</li>
917<li id="opt_fsanitize_object-size"><tt>-fsanitize=object-size</tt>:
918 An attempt to use bytes which the optimizer can determine are not part of
919 the object being accessed.
920 The sizes of objects are determined using <tt>__builtin_object_size</tt>, and
921 consequently may be able to detect more problems at higher optimization
922 levels.</li>
923<li id="opt_fsanitize_return"><tt>-fsanitize=return</tt>:
924 In C++, reaching the end of a value-returning function without returning a
925 value.</li>
926<li id="opt_fsanitize_shift"><tt>-fsanitize=shift</tt>:
927 Shift operators where the amount shifted is greater or equal to the
928 promoted bit-width of the left hand side or less than zero, or where
929 the left hand side is negative. For a signed left shift, also checks
930 for signed overflow in C, and for unsigned overflow in C++.</li>
931<li id="opt_fsanitize_signed-integer-overflow"><tt>-fsanitize=signed-integer-overflow</tt>:
932 Signed integer overflow, including all the checks added by <tt>-ftrapv</tt>,
933 and checking for overflow in signed division (<tt>INT_MIN / -1</tt>).</li>
934<li id="opt_fsanitize_unreachable"><tt>-fsanitize=unreachable</tt>:
935 If control flow reaches __builtin_unreachable.</li>
Will Dietzb8540362012-11-27 15:01:55 +0000[diff] [blame]936<li id="opt_fsanitize_unsigned-integer-overflow"><tt>-fsanitize=unsigned-integer-overflow</tt>:
937 Unsigned integer overflows.</li>
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]938<li id="opt_fsanitize_vla-bound"><tt>-fsanitize=vla-bound</tt>:
939 A variable-length array whose bound does not evaluate to a positive value.</li>
940<li id="opt_fsanitize_vptr"><tt>-fsanitize=vptr</tt>:
941 Use of an object whose vptr indicates that it is of the wrong dynamic type,
942 or that its lifetime has not begun or has ended. Incompatible with
943 <tt>-fno-rtti</tt>.</li>
944</ul>
945
Alexey Samsonov4d1a6e42012-11-29 22:36:21 +0000[diff] [blame]946Experimental features of AddressSanitizer (not ready for widespread use,
947require explicit <tt>-fsanitize=address</tt>):
948
949<ul>
950<li id="opt_fsanitize_init_order"><tt>-fsanitize=init-order</tt>:
951Check for dynamic initialization order problems.</li>
952<li id="opt_fsanitize_use_after_return"><tt>-fsanitize=use-after-return</tt>:
953Check for use-after-return errors (accessing local variable after the function
954exit).
955<li id="opt_fsanitize_use_after_scope"><tt>-fsanitize=use-after-scope</tt>:
956Check for use-after-scope errors (accesing local variable after it went out of
957scope).
958</ul>
959
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]960The <tt>-fsanitize=</tt> argument must also be provided when linking, in order
961to link to the appropriate runtime library. It is not possible to combine the
962<tt>-fsanitize=address</tt> and <tt>-fsanitize=thread</tt> checkers in the same
963program.
Mike Stump53664ca2009-12-14 23:53:10 +0000[diff] [blame]964</dd>
965
Kostya Serebryany48a8cd92012-03-01 23:14:19 +0000[diff] [blame]966<dt id="opt_faddress-sanitizer"><b>-f[no-]address-sanitizer</b>:
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]967Deprecated synonym for <a href="#opt_fsanitize_address"><tt>-f[no-]sanitize=address</tt></a>.
Kostya Serebryany48a8cd92012-03-01 23:14:19 +0000[diff] [blame]968
969<dt id="opt_fthread-sanitizer"><b>-f[no-]thread-sanitizer</b>:
Richard Smithfd9d0e12012-11-06 19:23:14 +0000[diff] [blame]970Deprecated synonym for <a href="#opt_fsanitize_address"><tt>-f[no-]sanitize=thread</tt></a>.
971
972<dt id="opt_fcatch-undefined-behavior"><b>-fcatch-undefined-behavior</b>:
973Deprecated synonym for <a href="#opt_fsanitize_undefined"><tt>-fsanitize=undefined</tt></a>.
Kostya Serebryany48a8cd92012-03-01 23:14:19 +0000[diff] [blame]974
Nuno Lopesaa526242009-12-17 10:00:52 +0000[diff] [blame]975<dt id="opt_fno-assume-sane-operator-new"><b>-fno-assume-sane-operator-new</b>:
976Don't assume that the C++'s new operator is sane.</dt>
Nuno Lopesb23f20d2009-12-17 10:15:49 +0000[diff] [blame]977<dd>This option tells the compiler to do not assume that C++'s global new
Chris Lattnerd462b6a2011-01-24 03:47:59 +0000[diff] [blame]978operator will always return a pointer that does not
Nuno Lopesaa526242009-12-17 10:00:52 +0000[diff] [blame]979alias any other pointer when the function returns.</dd>
Evan Chengfda026b2011-04-08 22:18:01 +0000[diff] [blame]980
Evan Cheng4cf4b592011-04-08 22:34:21 +0000[diff] [blame]981<dt id="opt_ftrap-function"><b>-ftrap-function=[name]</b>: Instruct code
Evan Chengfda026b2011-04-08 22:18:01 +0000[diff] [blame]982generator to emit a function call to the specified function name for
983<tt>__builtin_trap()</tt>.</dt>
984
985<dd>LLVM code generator translates <tt>__builtin_trap()</tt> to a trap
986instruction if it is supported by the target ISA. Otherwise, the builtin is
987translated into a call to <tt>abort</tt>. If this option is set, then the code
988generator will always lower the builtin to a call to the specified function
989regardless of whether the target ISA has a trap instruction. This option is
Chris Lattner02e266b2011-04-08 22:58:43 +0000[diff] [blame]990useful for environments (e.g. deeply embedded) where a trap cannot be properly
991handled, or when some custom behavior is desired.</dd>
Hans Wennborg7bf9d9b2012-07-13 12:47:15 +0000[diff] [blame]992
993<dt id="opt_ftls-model"><b>-ftls-model=[model]</b>: Select which TLS model to
994use.</dt>
995<dd>Valid values are: <tt>global-dynamic</tt>, <tt>local-dynamic</tt>,
996<tt>initial-exec</tt> and <tt>local-exec</tt>. The default value is
997<tt>global-dynamic</tt>. The compiler may use a different model if the selected
998model is not supported by the target, or if a more efficient model can be used.
999The TLS model can be overridden per variable using the <tt>tls_model</tt>
1000attribute.
1001</dd>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]1002</dl>
Nuno Lopesaa526242009-12-17 10:00:52 +0000[diff] [blame]1003
Alexey Samsonov5ab52492012-05-30 06:55:10 +0000[diff] [blame]1004<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1005<h3 id="debuginfosize">Controlling Size of Debug Information</h3>
1006<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1007
1008<p>Debug info kind generated by Clang can be set by one of the flags listed
1009below. If multiple flags are present, the last one is used.</p>
1010
1011<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
1012<dl>
1013<dt id="opt_g0"><b>-g0</b>: Don't generate any debug info (default).
1014
1015<dt id="opt_gline-tables-only"><b>-gline-tables-only</b>:
1016Generate line number tables only.
1017<dd>
1018This kind of debug info allows to obtain stack traces with function
1019names, file names and line numbers (by such tools as
1020gdb or addr2line). It doesn't contain any other data (e.g.
1021description of local variables or function parameters).
1022</dd>
1023
1024<dt id="opt_g"><b>-g</b>: Generate complete debug info.
1025</dl>
1026
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1027<!-- ======================================================================= -->
1028<h2 id="c">C Language Features</h2>
1029<!-- ======================================================================= -->
1030
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1031<p>The support for standard C in clang is feature-complete except for the C99
1032floating-point pragmas.</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1033
1034<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1035<h3 id="c_ext">Extensions supported by clang</h3>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1036<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1037
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1038<p>See <a href="LanguageExtensions.html">clang language extensions</a>.</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1039
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1040<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1041<h3 id="c_modes">Differences between various standard modes</h3>
1042<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1043
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1044<p>clang supports the -std option, which changes what language mode clang uses.
1045The supported modes for C are c89, gnu89, c94, c99, gnu99 and various aliases
1046for those modes. If no -std option is specified, clang defaults to gnu99 mode.
1047</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1048
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1049<p>Differences between all c* and gnu* modes:</p>
1050<ul>
1051<li>c* modes define "__STRICT_ANSI__".</li>
Eli Friedman26fa0ed2009-05-27 23:02:57 +0000[diff] [blame]1052<li>Target-specific defines not prefixed by underscores, like "linux", are
1053defined in gnu* modes.</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1054<li>Trigraphs default to being off in gnu* modes; they can be enabled by the
1055-trigraphs option.</li>
1056<li>The parser recognizes "asm" and "typeof" as keywords in gnu* modes; the
1057variants "__asm__" and "__typeof__" are recognized in all modes.</li>
Eli Friedmanb0ac5452009-05-16 23:17:30 +0000[diff] [blame]1058<li>The Apple "blocks" extension is recognized by default in gnu* modes
1059on some platforms; it can be enabled in any mode with the "-fblocks"
1060option.</li>
Chris Lattner741be6a2011-06-15 00:41:40 +0000[diff] [blame]1061<li>Arrays that are VLA's according to the standard, but which can be constant
1062 folded by the frontend are treated as fixed size arrays. This occurs for
1063 things like "int X[(1, 2)];", which is technically a VLA. c* modes are
1064 strictly compliant and treat these as VLAs.</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1065</ul>
1066
1067<p>Differences between *89 and *99 modes:</p>
1068<ul>
1069<li>The *99 modes default to implementing "inline" as specified in C99, while
1070the *89 modes implement the GNU version. This can be overridden for individual
1071functions with the __gnu_inline__ attribute.</li>
Eli Friedman26fa0ed2009-05-27 23:02:57 +0000[diff] [blame]1072<li>Digraphs are not recognized in c89 mode.</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1073<li>The scope of names defined inside a "for", "if", "switch", "while", or "do"
1074statement is different. (example: "if ((struct x {int x;}*)0) {}".)</li>
1075<li>__STDC_VERSION__ is not defined in *89 modes.</li>
Eli Friedman26fa0ed2009-05-27 23:02:57 +0000[diff] [blame]1076<li>"inline" is not recognized as a keyword in c89 mode.</li>
1077<li>"restrict" is not recognized as a keyword in *89 modes.</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1078<li>Commas are allowed in integer constant expressions in *99 modes.</li>
1079<li>Arrays which are not lvalues are not implicitly promoted to pointers in
1080*89 modes.</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1081<li>Some warnings are different.</li>
1082</ul>
1083
1084<p>c94 mode is identical to c89 mode except that digraphs are enabled in
1085c94 mode (FIXME: And __STDC_VERSION__ should be defined!).</p>
1086
1087<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1088<h3 id="c_unimpl_gcc">GCC extensions not implemented yet</h3>
1089<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1090
1091<p>clang tries to be compatible with gcc as much as possible, but some gcc
1092extensions are not implemented yet:</p>
1093
1094<ul>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1095
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1096<li>clang does not support #pragma weak
Eli Friedman4da92552009-06-02 08:21:31 +0000[diff] [blame]1097(<a href="http://llvm.org/bugs/show_bug.cgi?id=3679">bug 3679</a>). Due to
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1098the uses described in the bug, this is likely to be implemented at some
1099point, at least partially.</li>
1100
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1101<li>clang does not support decimal floating point types (_Decimal32 and
1102friends) or fixed-point types (_Fract and friends); nobody has expressed
1103interest in these features yet, so it's hard to say when they will be
1104implemented.</li>
1105
1106<li>clang does not support nested functions; this is a complex feature which
Dmitri Gribenko1dfb72f2012-07-17 18:51:47 +0000[diff] [blame]1107is infrequently used, so it is unlikely to be implemented anytime soon. In C++11
1108it can be emulated by assigning lambda functions to local variables, e.g:
1109<pre>
1110 auto const local_function = [&](int parameter) {
1111 // Do something
1112 };
1113 ...
1114 local_function(1);
1115</pre>
1116</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1117
Dmitri Gribenko1dfb72f2012-07-17 18:51:47 +0000[diff] [blame]1118<li>clang does not support global register variables; this is unlikely
Eli Friedmanca130bd2009-06-12 20:04:25 +0000[diff] [blame]1119to be implemented soon because it requires additional LLVM backend support.
1120</li>
Daniel Dunbar05fa6292009-05-12 23:12:07 +0000[diff] [blame]1121
1122<li>clang does not support static initialization of flexible array
1123members. This appears to be a rarely used extension, but could be
1124implemented pending user demand.</li>
1125
Eli Friedmanca130bd2009-06-12 20:04:25 +0000[diff] [blame]1126<li>clang does not support __builtin_va_arg_pack/__builtin_va_arg_pack_len.
1127This is used rarely, but in some potentially interesting places, like the
1128glibc headers, so it may be implemented pending user demand. Note that
1129because clang pretends to be like GCC 4.2, and this extension was introduced
1130in 4.3, the glibc headers will not try to use this extension with clang at
1131the moment.</li>
1132
1133<li>clang does not support the gcc extension for forward-declaring function
Dmitri Gribenko1dfb72f2012-07-17 18:51:47 +0000[diff] [blame]1134parameters; this has not shown up in any real-world code yet, though, so it
Eli Friedmanca130bd2009-06-12 20:04:25 +0000[diff] [blame]1135might never be implemented.</li>
1136
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1137</ul>
1138
1139<p>This is not a complete list; if you find an unsupported extension
1140missing from this list, please send an e-mail to cfe-dev. This list
1141currently excludes C++; see <a href="#cxx">C++ Language Features</a>.
1142Also, this list does not include bugs in mostly-implemented features; please
1143see 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">
1144bug tracker</a> for known existing bugs (FIXME: Is there a section for
1145bug-reporting guidelines somewhere?).</p>
1146
1147<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1148<h3 id="c_unsupp_gcc">Intentionally unsupported GCC extensions</h3>
1149<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1150
Eli Friedman0b326002009-06-12 20:11:05 +0000[diff] [blame]1151<ul>
1152
Eli Friedman2f3fca02009-06-12 20:08:48 +0000[diff] [blame]1153<li>clang does not support the gcc extension that allows variable-length arrays
Chris Lattnerd462b6a2011-01-24 03:47:59 +0000[diff] [blame]1154in structures. This is for a few reasons: one, it is tricky
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1155to implement, two, the extension is completely undocumented, and three, the
Chris Lattnerb9d511c2010-10-28 02:20:32 +0000[diff] [blame]1156extension appears to be rarely used. Note that clang <em>does</em> support
1157flexible array members (arrays with a zero or unspecified size at the end of
1158a structure).</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1159
Eli Friedmanca130bd2009-06-12 20:04:25 +0000[diff] [blame]1160<li>clang does not have an equivalent to gcc's "fold"; this means that
1161clang doesn't accept some constructs gcc might accept in contexts where a
Eli Friedman248f9732011-06-13 23:12:01 +0000[diff] [blame]1162constant expression is required, like "x-x" where x is a variable.</li>
Eli Friedmanca130bd2009-06-12 20:04:25 +0000[diff] [blame]1163
Eli Friedmanca130bd2009-06-12 20:04:25 +0000[diff] [blame]1164<li>clang does not support __builtin_apply and friends; this extension is
1165extremely obscure and difficult to implement reliably.</li>
1166
Eli Friedman0b326002009-06-12 20:11:05 +0000[diff] [blame]1167</ul>
1168
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]1169<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1170<h3 id="c_ms">Microsoft extensions</h3>
1171<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1172
1173<p>clang has some experimental support for extensions from
1174Microsoft Visual C++; to enable it, use the -fms-extensions command-line
Eli Friedmana1821b52009-06-08 06:21:03 +0000[diff] [blame]1175option. This is the default for Windows targets. Note that the
1176support is incomplete; enabling Microsoft extensions will silently drop
1177certain constructs (including __declspec and Microsoft-style asm statements).
1178</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1179
Nico Weberd4047b32012-07-11 16:56:28 +0000[diff] [blame]1180<p>clang has a -fms-compatibility flag that makes clang accept enough
1181invalid C++ to be able to parse most Microsoft headers. This flag is enabled by
1182default for Windows targets.</p>
1183
1184<p>-fdelayed-template-parsing lets clang delay all template instantiation until
1185the end of a translation unit. This flag is enabled by default for Windows
1186targets.</p>
1187
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]1188<ul>
Douglas Gregord3b227d2010-12-14 16:21:49 +0000[diff] [blame]1189<li>clang allows setting _MSC_VER with -fmsc-version=. It defaults to 1300 which
Michael J. Spencerdae4ac42010-10-21 05:21:48 +0000[diff] [blame]1190is the same as Visual C/C++ 2003. Any number is supported and can greatly affect
1191what Windows SDK and c++stdlib headers clang can compile. This option will be
1192removed when clang supports the full set of MS extensions required for these
1193headers.</li>
1194
Daniel Dunbar05fa6292009-05-12 23:12:07 +0000[diff] [blame]1195<li>clang does not support the Microsoft extension where anonymous
Daniel Dunbar869e0542009-05-13 00:23:53 +0000[diff] [blame]1196record members can be declared using user defined typedefs.</li>
Daniel Dunbar05fa6292009-05-12 23:12:07 +0000[diff] [blame]1197
Daniel Dunbar9375ed12009-05-13 21:40:49 +0000[diff] [blame]1198<li>clang supports the Microsoft "#pragma pack" feature for
1199controlling record layout. GCC also contains support for this feature,
1200however where MSVC and GCC are incompatible clang follows the MSVC
1201definition.</li>
Nico Weberd4047b32012-07-11 16:56:28 +0000[diff] [blame]1202
1203<li>clang defaults to C++11 for Windows targets.</li>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]1204</ul>
Daniel Dunbar9375ed12009-05-13 21:40:49 +0000[diff] [blame]1205
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1206<!-- ======================================================================= -->
Richard Smithbdb10182011-11-21 20:54:59 +0000[diff] [blame]1207<h2 id="cxx">C++ Language Features</h2>
1208<!-- ======================================================================= -->
1209
1210<p>clang fully implements all of standard C++98 except for exported templates
1211(which were removed in C++11), and
1212<a href="http://clang.llvm.org/cxx_status.html">many C++11 features</a> are also
1213implemented.</p>
1214
1215<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1216<h3 id="cxx_implimits">Controlling implementation limits</h3>
1217<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1218
1219<p><b>-fconstexpr-depth=N</b>: Sets the limit for recursive constexpr function
1220invocations to N. The default is 512.</p>
1221
1222<p><b>-ftemplate-depth=N</b>: Sets the limit for recursively nested template
1223instantiations to N. The default is 1024.</p>
1224
1225<!-- ======================================================================= -->
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1226<h2 id="target_features">Target-Specific Features and Limitations</h2>
1227<!-- ======================================================================= -->
1228
1229
1230<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1231<h3 id="target_arch">CPU Architectures Features and Limitations</h3>
1232<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1233
1234<!-- ======================== -->
1235<h4 id="target_arch_x86">X86</h4>
1236<!-- ======================== -->
Daniel Dunbarbcaf7aa2010-09-19 19:26:59 +0000[diff] [blame]1237
1238<p>The support for X86 (both 32-bit and 64-bit) is considered stable on Darwin
1239(Mac OS/X), Linux, FreeBSD, and Dragonfly BSD: it has been tested to correctly
1240compile many large C, C++, Objective-C, and Objective-C++ codebases.</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1241
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]1242<p>On x86_64-mingw32, passing i128(by value) is incompatible to Microsoft x64
1243calling conversion. You might need to tweak WinX86_64ABIInfo::classify()
1244in lib/CodeGen/TargetInfo.cpp.</p>
NAKAMURA Takumia75fdb22011-04-04 15:02:41 +0000[diff] [blame]1245
Eli Friedman3b658d32009-06-08 05:12:39 +0000[diff] [blame]1246<!-- ======================== -->
1247<h4 id="target_arch_arm">ARM</h4>
1248<!-- ======================== -->
Daniel Dunbarbcaf7aa2010-09-19 19:26:59 +0000[diff] [blame]1249
1250<p>The support for ARM (specifically ARMv6 and ARMv7) is considered stable on
1251Darwin (iOS): it has been tested to correctly compile many large C, C++,
Bob Wilson4ea8dfa2011-01-10 17:55:14 +0000[diff] [blame]1252Objective-C, and Objective-C++ codebases. Clang only supports a limited number
1253of ARM architectures. It does not yet fully support ARMv5, for example.</p>
Eli Friedman3b658d32009-06-08 05:12:39 +0000[diff] [blame]1254
1255<!-- ======================== -->
1256<h4 id="target_arch_other">Other platforms</h4>
1257<!-- ======================== -->
1258clang currently contains some support for PPC and Sparc; however, significant
1259pieces of code generation are still missing, and they haven't undergone
1260significant testing.
1261
Eli Friedman3b658d32009-06-08 05:12:39 +0000[diff] [blame]1262<p>clang contains limited support for the MSP430 embedded processor, but both
1263the clang support and the LLVM backend support are highly experimental.
1264
1265<p>Other platforms are completely unsupported at the moment. Adding the
1266minimal support needed for parsing and semantic analysis on a new platform
1267is quite easy; see lib/Basic/Targets.cpp in the clang source tree. This level
1268of support is also sufficient for conversion to LLVM IR for simple programs.
1269Proper support for conversion to LLVM IR requires adding code to
1270lib/CodeGen/CGCall.cpp at the moment; this is likely to change soon, though.
1271Generating assembly requires a suitable LLVM backend.
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1272
1273<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1274<h3 id="target_os">Operating System Features and Limitations</h3>
1275<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1276
1277<!-- ======================================= -->
1278<h4 id="target_os_darwin">Darwin (Mac OS/X)</h4>
1279<!-- ======================================= -->
1280
Chris Lattner396663b2012-05-04 21:11:08 +0000[diff] [blame]1281<p>None</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1282
NAKAMURA Takumia75fdb22011-04-04 15:02:41 +0000[diff] [blame]1283<!-- ======================================= -->
1284<h4 id="target_os_win32">Windows</h4>
1285<!-- ======================================= -->
1286
1287<p>Experimental supports are on Cygming.</p>
1288
Nico Weberd4047b32012-07-11 16:56:28 +0000[diff] [blame]1289<p>See also <a href="#c_ms">Microsoft Extensions</a>.</p>
1290
NAKAMURA Takumia75fdb22011-04-04 15:02:41 +0000[diff] [blame]1291<h5>Cygwin</h5>
1292
1293<p>Clang works on Cygwin-1.7.</p>
1294
1295<h5>MinGW32</h5>
1296
1297<p>Clang works on some mingw32 distributions.
1298Clang assumes directories as below;</p>
1299
1300<ul>
1301<li><tt>C:/mingw/include</tt></li>
1302<li><tt>C:/mingw/lib</tt></li>
1303<li><tt>C:/mingw/lib/gcc/mingw32/4.[3-5].0/include/c++</tt></li>
1304</ul>
1305
NAKAMURA Takumi0e53df62011-11-17 07:54:13 +0000[diff] [blame]1306<p>On MSYS, a few tests might fail.</p>
NAKAMURA Takumia75fdb22011-04-04 15:02:41 +0000[diff] [blame]1307
1308<h5>MinGW-w64</h5>
1309
Douglas Gregordad879a2011-08-30 00:40:12 +0000[diff] [blame]1310<p>For 32-bit (i686-w64-mingw32), and 64-bit (x86_64-w64-mingw32), Clang assumes as below;<p>
NAKAMURA Takumia75fdb22011-04-04 15:02:41 +0000[diff] [blame]1311
1312<ul>
Douglas Gregordad879a2011-08-30 00:40:12 +0000[diff] [blame]1313<li><tt>GCC versions 4.5.0 to 4.5.3, 4.6.0 to 4.6.2, or 4.7.0 (for the C++ header search path)</tt></li>
1314<li><tt>some_directory/bin/gcc.exe</tt></li>
1315<li><tt>some_directory/bin/clang.exe</tt></li>
1316<li><tt>some_directory/bin/clang++.exe</tt></li>
1317<li><tt>some_directory/bin/../include/c++/GCC_version</tt></li>
1318<li><tt>some_directory/bin/../include/c++/GCC_version/x86_64-w64-mingw32</tt></li>
1319<li><tt>some_directory/bin/../include/c++/GCC_version/i686-w64-mingw32</tt></li>
1320<li><tt>some_directory/bin/../include/c++/GCC_version/backward</tt></li>
1321<li><tt>some_directory/bin/../x86_64-w64-mingw32/include</tt></li>
1322<li><tt>some_directory/bin/../i686-w64-mingw32/include</tt></li>
1323<li><tt>some_directory/bin/../include</tt></li>
NAKAMURA Takumia75fdb22011-04-04 15:02:41 +0000[diff] [blame]1324</ul>
1325
David Blaikiea9639622012-04-19 17:21:46 +0000[diff] [blame]1326<p>This directory layout is standard for any toolchain you will find on the official <a href="http://mingw-w64.sourceforge.net">MinGW-w64 website</a>.
Douglas Gregordad879a2011-08-30 00:40:12 +0000[diff] [blame]1327
1328<p>Clang expects the GCC executable &quot;gcc.exe&quot; compiled for i686-w64-mingw32 (or x86_64-w64-mingw32) to be present on PATH.</p>
1329
1330<p><a href="http://llvm.org/bugs/show_bug.cgi?id=9072">Some tests might fail</a>
1331on x86_64-w64-mingw32.</p>
NAKAMURA Takumia75fdb22011-04-04 15:02:41 +0000[diff] [blame]1332
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1333</div>
1334</body>
1335</html>