Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / clang / c9471b0ff1815ed0149dbfcad0f385ed8648eeb0 / . / docs / UsersManual.html
blob: d4f9dfbb017c0eb35c3b21d08bdd04389f7246c3 [file] [log] [blame]
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1<html>
2<head>
3<title>Clang Compiler User's Manual</title>
4<link type="text/css" rel="stylesheet" href="../menu.css" />
5<link type="text/css" rel="stylesheet" href="../content.css" />
6<style type="text/css">
7td {
8 vertical-align: top;
9}
10</style>
11</head>
12<body>
13
14<!--#include virtual="../menu.html.incl"-->
15
16<div id="content">
17
18<h1>Clang Compiler User's Manual</h1>
19
20<ul>
21<li><a href="#intro">Introduction</a>
22 <ul>
23 <li><a href="#terminology">Terminology</a></li>
24 <li><a href="#basicusage">Basic Usage</a></li>
25 </ul>
26</li>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]27<li><a href="#commandline">Command Line Options</a>
28 <ul>
29 <li><a href="#cl_diagnostics">Options to Control Error and Warning
30 Messages</a></li>
31 </ul>
32</li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]33<li><a href="#general_features">Language and Target-Independent Features</a>
34 <ul>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]35 <li><a href="#diagnostics">Controlling Errors and Warnings</a>
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]36 <ul>
37 <li><a href="#diagnostics_display">Controlling How Clang Displays Diagnostics</a></li>
38 <li><a href="#diagnostics_mappings">Diagnostic Mappings</a></li>
Chris Lattner67db8cd2010-05-30 23:42:51 +0000[diff] [blame]39 <li><a href="#diagnostics_categories">Diagnostic Categories</a></li>
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]40 <li><a href="#diagnostics_commandline">Controlling Diagnostics via Command Line Flags</a></li>
41 <li><a href="#diagnostics_pragmas">Controlling Diagnostics via Pragmas</a></li>
Ted Kremenek2fb11eb2010-08-24 18:12:35 +0000[diff] [blame]42 <li><a href="#analyzer_diagnositics">Controlling Static Analyzer Diagnostics</a></li>
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]43 </ul>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]44 </li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]45 <li><a href="#precompiledheaders">Precompiled Headers</a></li>
Mike Stump53664ca2009-12-14 23:53:10 +0000[diff] [blame]46 <li><a href="#codegen">Controlling Code Generation</a></li>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]47 </ul>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]48</li>
49<li><a href="#c">C Language Features</a>
50 <ul>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]51 <li><a href="#c_ext">Extensions supported by clang</a></li>
52 <li><a href="#c_modes">Differences between various standard modes</a></li>
53 <li><a href="#c_unimpl_gcc">GCC extensions not implemented yet</a></li>
54 <li><a href="#c_unsupp_gcc">Intentionally unsupported GCC extensions</a></li>
55 <li><a href="#c_ms">Microsoft extensions</a></li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]56 </ul>
57</li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]58<li><a href="#target_features">Target-Specific Features and Limitations</a>
59 <ul>
60 <li><a href="#target_arch">CPU Architectures Features and Limitations</a>
61 <ul>
62 <li><a href="#target_arch_x86">X86</a></li>
Eli Friedman3b658d32009-06-08 05:12:39 +0000[diff] [blame]63 <li><a href="#target_arch_arm">ARM</a></li>
64 <li><a href="#target_arch_other">Other platforms</a></li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]65 </ul>
66 </li>
67 <li><a href="#target_os">Operating System Features and Limitations</a>
68 <ul>
69 <li><a href="#target_os_darwin">Darwin (Mac OS/X)</a></li>
70 <li>Linux, etc.</li>
NAKAMURA Takumia75fdb22011-04-04 15:02:41 +0000[diff] [blame]71 <li><a href="#target_os_win32">Windows</a></li>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]72 </ul>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]73 </li>
74 </ul>
75</li>
76</ul>
77
78
79<!-- ======================================================================= -->
80<h2 id="intro">Introduction</h2>
81<!-- ======================================================================= -->
82
83<p>The Clang Compiler is an open-source compiler for the C family of programming
84languages, aiming to be the best in class implementation of these languages.
85Clang builds on the LLVM optimizer and code generator, allowing it to provide
86high-quality optimization and code generation support for many targets. For
87more general information, please see the <a href="http://clang.llvm.org">Clang
88Web Site</a> or the <a href="http://llvm.org">LLVM Web Site</a>.</p>
89
90<p>This document describes important notes about using Clang as a compiler for
91an end-user, documenting the supported features, command line options, etc. If
92you are interested in using Clang to build a tool that processes code, please
93see <a href="InternalsManual.html">the Clang Internals Manual</a>. If you are
94interested in the <a href="http://clang.llvm.org/StaticAnalysis.html">Clang
95Static Analyzer</a>, please see its web page.</p>
96
97<p>Clang is designed to support the C family of programming languages, which
98includes <a href="#c">C</a>, <a href="#objc">Objective-C</a>, <a
99href="#cxx">C++</a>, and <a href="#objcxx">Objective-C++</a> as well as many
100dialects of those. For language-specific information, please see the
101corresponding language specific section:</p>
102
103<ul>
104<li><a href="#c">C Language</a>: K&amp;R C, ANSI C89, ISO C90, ISO C94
105 (C89+AMD1), ISO C99 (+TC1, TC2, TC3). </li>
106<li><a href="#objc">Objective-C Language</a>: ObjC 1, ObjC 2, ObjC 2.1, plus
107 variants depending on base language.</li>
108<li><a href="#cxx">C++ Language Features</a></li>
109<li><a href="#objcxx">Objective C++ Language</a></li>
110</ul>
111
112<p>In addition to these base languages and their dialects, Clang supports a
113broad variety of language extensions, which are documented in the corresponding
114language section. These extensions are provided to be compatible with the GCC,
115Microsoft, and other popular compilers as well as to improve functionality
116through Clang-specific features. The Clang driver and language features are
117intentionally designed to be as compatible with the GNU GCC compiler as
118reasonably possible, easing migration from GCC to Clang. In most cases, code
119"just works".</p>
120
121<p>In addition to language specific features, Clang has a variety of features
122that depend on what CPU architecture or operating system is being compiled for.
Douglas Gregorcd5a5052009-11-09 15:15:41 +0000[diff] [blame]123Please see the <a href="#target_features">Target-Specific Features and
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]124Limitations</a> section for more details.</p>
125
126<p>The rest of the introduction introduces some basic <a
127href="#terminology">compiler terminology</a> that is used throughout this manual
128and contains a basic <a href="#basicusage">introduction to using Clang</a>
129as a command line compiler.</p>
130
131<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
132<h3 id="terminology">Terminology</h3>
133<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
134
135<p>Front end, parser, backend, preprocessor, undefined behavior, diagnostic,
136 optimizer</p>
137
138<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
139<h3 id="basicusage">Basic Usage</h3>
140<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
141
142<p>Intro to how to use a C compiler for newbies.</p>
143<p>
144compile + link
145
146compile then link
147
148debug info
149
150enabling optimizations
151
152picking a language to use, defaults to C99 by default. Autosenses based on
153extension.
154
155using a makefile
156</p>
157
158
159<!-- ======================================================================= -->
160<h2 id="commandline">Command Line Options</h2>
161<!-- ======================================================================= -->
162
163<p>
164This section is generally an index into other sections. It does not go into
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]165depth on the ones that are covered by other sections. However, the first part
166introduces the language selection and other high level options like -c, -g, etc.
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]167</p>
168
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]169
170<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
171<h3 id="cl_diagnostics">Options to Control Error and Warning Messages</h3>
172<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
173
174<p><b>-Werror</b>: Turn warnings into errors.</p>
175<p><b>-Werror=foo</b>: Turn warning "foo" into an error.</p>
176<p><b>-Wno-error=foo</b>: Turn warning "foo" into an warning even if -Werror is
177 specified.</p>
178<p><b>-Wfoo</b>: Enable warning foo</p>
179<p><b>-Wno-foo</b>: Disable warning foo</p>
180<p><b>-w</b>: Disable all warnings.</p>
181<p><b>-pedantic</b>: Warn on language extensions.</p>
182<p><b>-pedantic-errors</b>: Error on language extensions.</p>
183<p><b>-Wsystem-headers</b>: Enable warnings from system headers.</p>
184
Chris Lattner0f0c9632010-04-07 20:49:23 +0000[diff] [blame]185<p><b>-ferror-limit=123</b>: Stop emitting diagnostics after 123 errors have
186 been produced. The default is 20, and the error limit can be disabled with
187 -ferror-limit=0.</p>
188
Douglas Gregor575cf372010-04-20 07:18:24 +0000[diff] [blame]189<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>
190
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]191<!-- ================================================= -->
192<h4 id="cl_diag_formatting">Formatting of Diagnostics</h4>
193<!-- ================================================= -->
194
195<p>Clang aims to produce beautiful diagnostics by default, particularly for new
196users that first come to Clang. However, different people have different
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]197preferences, and sometimes Clang is driven by another program that wants to
198parse simple and consistent output, not a person. For these cases, Clang
199provides a wide range of options to control the exact output format of the
200diagnostics that it generates.</p>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]201
202<dl>
203
204<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
205<dt id="opt_fshow-column"><b>-f[no-]show-column</b>: Print column number in
206diagnostic.</dt>
207<dd>This option, which defaults to on, controls whether or not Clang prints the
208column number of a diagnostic. For example, when this is enabled, Clang will
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]209print something like:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]210
211<pre>
212 test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
213 #endif bad
214 ^
215 //
216</pre>
217
218<p>When this is disabled, Clang will print "test.c:28: warning..." with no
219column number.</p>
220</dd>
221
222<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
223<dt id="opt_fshow-source-location"><b>-f[no-]show-source-location</b>: Print
224source file/line/column information in diagnostic.</dt>
225<dd>This option, which defaults to on, controls whether or not Clang prints the
226filename, line number and column number of a diagnostic. For example,
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]227when this is enabled, Clang will print something like:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]228
229<pre>
230 test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
231 #endif bad
232 ^
233 //
234</pre>
235
236<p>When this is disabled, Clang will not print the "test.c:28:8: " part.</p>
237</dd>
238
239<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
240<dt id="opt_fcaret-diagnostics"><b>-f[no-]caret-diagnostics</b>: Print source
241line and ranges from source code in diagnostic.</dt>
242<dd>This option, which defaults to on, controls whether or not Clang prints the
243source line, source ranges, and caret when emitting a diagnostic. For example,
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]244when this is enabled, Clang will print something like:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]245
246<pre>
247 test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
248 #endif bad
249 ^
250 //
251</pre>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]252</dd>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]253<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
Douglas Gregore8d44dd2010-07-09 16:31:58 +0000[diff] [blame]254<dt id="opt_fcolor_diagnostics"><b>-f[no-]color-diagnostics</b>: </dt>
255<dd>This option, which defaults to on when a color-capable terminal is
256 detected, controls whether or not Clang prints diagnostics in color.
257 When this option is enabled, Clang will use colors to highlight
258 specific parts of the diagnostic, e.g.,
Chris Lattner4e1c53d2011-01-24 03:47:34 +0000[diff] [blame]259 <pre>
260 <b><font color="black">test.c:28:8: <font color="magenta">warning</font>: extra tokens at end of #endif directive [-Wextra-tokens]</font></b>
261 #endif bad
262 <font color="green">^</font>
263 <font color="green">//</font>
Douglas Gregore8d44dd2010-07-09 16:31:58 +0000[diff] [blame]264</pre>
Chris Lattner4e1c53d2011-01-24 03:47:34 +0000[diff] [blame]265
266<p>When this is disabled, Clang will just print:</p>
267
268<pre>
269 test.c:2:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
270 #endif bad
271 ^
272 //
273</pre>
274</dd>
Douglas Gregore8d44dd2010-07-09 16:31:58 +0000[diff] [blame]275<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
Douglas Gregorc9471b02011-05-21 17:07:29 +0000[diff] [blame^]276<dt id="opt_fdiagnostics-format"><b>-fdiagnostics-format=clang/msvc/vi</b>:
277Changes diagnostic output format to better match IDEs and command line tools.</dt>
278<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:
279
280 <dl>
281 <dt><b>clang</b> (default)</dt>
282 <dd>
283 <pre>t.c:3:11: warning: conversion specifies type 'char *' but the argument has type 'int'</pre>
284 </dd>
285
286 <dt><b>msvc</b></dt>
287 <dd>
288 <pre>t.c(3,11) : warning: conversion specifies type 'char *' but the argument has type 'int'</pre>
289 </dd>
290
291 <dt><b>vi</b></dt>
292 <dd>
293 <pre>t.c +3:11: warning: conversion specifies type 'char *' but the argument has type 'int'</pre>
294 </dd>
295 </dl>
296</dd>
297
298<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]299<dt id="opt_fdiagnostics-show-option"><b>-f[no-]diagnostics-show-option</b>:
300Enable <tt>[-Woption]</tt> information in diagnostic line.</dt>
301<dd>This option, which defaults to on,
302controls whether or not Clang prints the associated <A
303href="#cl_diag_warning_groups">warning group</a> option name when outputting
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]304a warning diagnostic. For example, in this output:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]305
306<pre>
307 test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
308 #endif bad
309 ^
310 //
311</pre>
312
313<p>Passing <b>-fno-diagnostics-show-option</b> will prevent Clang from printing
314the [<a href="#opt_Wextra-tokens">-Wextra-tokens</a>] information in the
315diagnostic. This information tells you the flag needed to enable or disable the
316diagnostic, either from the command line or through <a
317href="#pragma_GCC_diagnostic">#pragma GCC diagnostic</a>.</dd>
318
Chris Lattner28a43a42010-05-05 01:35:28 +0000[diff] [blame]319<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
320<dt id="opt_fdiagnostics-show-category"><b>-fdiagnostics-show-category=none/id/name</b>:
321Enable printing category information in diagnostic line.</dt>
322<dd>This option, which defaults to "none",
323controls whether or not Clang prints the category associated with a diagnostic
324when emitting it. Each diagnostic may or many not have an associated category,
325if it has one, it is listed in the diagnostic categorization field of the
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]326diagnostic line (in the []'s).
Chris Lattner28a43a42010-05-05 01:35:28 +0000[diff] [blame]327
328<p>For example, a format string warning will produce these three renditions
329based on the setting of this option:</p>
330
331<pre>
332 t.c:3:11: warning: conversion specifies type 'char *' but the argument has type 'int' [-Wformat]
333 t.c:3:11: warning: conversion specifies type 'char *' but the argument has type 'int' [-Wformat<b>,1</b>]
334 t.c:3:11: warning: conversion specifies type 'char *' but the argument has type 'int' [-Wformat<b>,Format String</b>]
335</pre>
336
337<p>This category can be used by clients that want to group diagnostics by
338category, so it should be a high level category. We want dozens of these, not
339hundreds or thousands of them.</p>
340</dd>
341
342
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]343
344<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
345<dt id="opt_fdiagnostics-fixit-info"><b>-f[no-]diagnostics-fixit-info</b>:
346Enable "FixIt" information in the diagnostics output.</dt>
347<dd>This option, which defaults to on, controls whether or not Clang prints the
348information on how to fix a specific diagnostic underneath it when it knows.
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]349For example, in this output:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]350
351<pre>
352 test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
353 #endif bad
354 ^
355 //
356</pre>
357
358<p>Passing <b>-fno-diagnostics-fixit-info</b> will prevent Clang from printing
359the "//" line at the end of the message. This information is useful for users
360who may not understand what is wrong, but can be confusing for machine
361parsing.</p>
362</dd>
363
364<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
Chris Lattner2a9cc232009-04-21 05:35:32 +0000[diff] [blame]365<dt id="opt_fdiagnostics-print-source-range-info">
366<b>-f[no-]diagnostics-print-source-range-info</b>:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]367Print machine parsable information about source ranges.</dt>
368<dd>This option, which defaults to off, controls whether or not Clang prints
369information about source ranges in a machine parsable format after the
370file/line/column number information. The information is a simple sequence of
371brace enclosed ranges, where each range lists the start and end line/column
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]372locations. For example, in this output:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]373
374<pre>
375exprs.c:47:15:{47:8-47:14}{47:17-47:24}: error: invalid operands to binary expression ('int *' and '_Complex float')
376 P = (P-42) + Gamma*4;
377 ~~~~~~ ^ ~~~~~~~
378</pre>
379
Chris Lattner2a9cc232009-04-21 05:35:32 +0000[diff] [blame]380<p>The {}'s are generated by -fdiagnostics-print-source-range-info.</p>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]381</dd>
382
Douglas Gregor4786c152010-08-19 20:24:43 +0000[diff] [blame]383<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
384<dt id="opt_fdiagnostics-parseable-fixits">
385<b>-fdiagnostics-parseable-fixits</b>:
386Print Fix-Its in a machine parseable form.</dt>
387<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>
388
389<pre>
Douglas Gregorbf5e09d2010-08-20 03:17:33 +0000[diff] [blame]390 fix-it:"t.cpp":{7:25-7:29}:"Gamma"
Douglas Gregor4786c152010-08-19 20:24:43 +0000[diff] [blame]391</pre>
392
NAKAMURA Takumi44626362011-04-05 00:57:02 +0000[diff] [blame]393<p>The range printed is a half-open range, so in this example the characters at
394column 25 up to but not including column 29 on line 7 in t.cpp should be
395replaced with the string &quot;Gamma&quot;. Either the range or the replacement
396string may be empty (representing strict insertions and strict erasures,
397respectively). Both the file name and the insertion string escape backslash (as
398&quot;\\&quot;), tabs (as &quot;\t&quot;), newlines (as &quot;\n&quot;), double
399quotes(as &quot;\&quot;&quot;) and non-printable characters (as octal
400&quot;\xxx&quot;).</p>
Douglas Gregor4786c152010-08-19 20:24:43 +0000[diff] [blame]401</dd>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]402
403</dl>
404
405
406
407
408<!-- ===================================================== -->
409<h4 id="cl_diag_warning_groups">Individual Warning Groups</h4>
410<!-- ===================================================== -->
411
412<p>TODO: Generate this from tblgen. Define one anchor per warning group.</p>
413
414
415<dl>
416
417
418<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
419<dt id="opt_Wextra-tokens"><b>-Wextra-tokens</b>: Warn about excess tokens at
420 the end of a preprocessor directive.</dt>
421<dd>This option, which defaults to on, enables warnings about extra tokens at
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]422the end of preprocessor directives. For example:
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]423
424<pre>
425 test.c:28:8: warning: extra tokens at end of #endif directive [-Wextra-tokens]
426 #endif bad
427 ^
428</pre>
429
430<p>These extra tokens are not strictly conforming, and are usually best handled
431by commenting them out.</p>
432
433<p>This option is also enabled by <a href="">-Wfoo</a>, <a href="">-Wbar</a>,
434 and <a href="">-Wbaz</a>.</p>
435</dd>
436
Jeffrey Yasskin21d07e42010-06-05 01:39:57 +0000[diff] [blame]437<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
438<dt id="opt_Wambiguous-member-template"><b>-Wambiguous-member-template</b>:
439Warn about unqualified uses of a member template whose name resolves
440to another template at the location of the use.</dt>
441<dd>This option, which defaults to on, enables a warning in the
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]442following code:
Jeffrey Yasskin21d07e42010-06-05 01:39:57 +0000[diff] [blame]443
444<pre>
445template&lt;typename T> struct set{};
446template&lt;typename T> struct trait { typedef const T& type; };
447struct Value {
448 template&lt;typename T> void set(typename trait&lt;T>::type value) {}
449};
450void foo() {
451 Value v;
452 v.set&lt;double>(3.2);
453}
454</pre>
455
456<p>C++ [basic.lookup.classref] requires this to be an error, but,
457because it's hard to work around, Clang downgrades it to a warning as
458an extension.</p>
459</dd>
460
Jeffrey Yasskin57d12fd2010-06-07 15:58:05 +0000[diff] [blame]461<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
462<dt id="opt_Wbind-to-temporary-copy"><b>-Wbind-to-temporary-copy</b>: Warn about
463an unusable copy constructor when binding a reference to a temporary.</dt>
464<dd>This option, which defaults to on, enables warnings about binding a
465reference to a temporary when the temporary doesn't have a usable copy
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]466constructor. For example:
Jeffrey Yasskin57d12fd2010-06-07 15:58:05 +0000[diff] [blame]467
468<pre>
469 struct NonCopyable {
470 NonCopyable();
471 private:
472 NonCopyable(const NonCopyable&);
473 };
474 void foo(const NonCopyable&);
475 void bar() {
476 foo(NonCopyable()); // Disallowed in C++98; allowed in C++0x.
477 }
478</pre>
479<pre>
480 struct NonCopyable2 {
481 NonCopyable2();
482 NonCopyable2(NonCopyable2&);
483 };
484 void foo(const NonCopyable2&);
485 void bar() {
486 foo(NonCopyable2()); // Disallowed in C++98; allowed in C++0x.
487 }
488</pre>
489
490<p>Note that if <tt>NonCopyable2::NonCopyable2()</tt> has a default
491argument whose instantiation produces a compile error, that error will
492still be a hard error in C++98 mode even if this warning is turned
493off.</p>
494
495</dd>
496
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]497</dl>
498
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]499<!-- ======================================================================= -->
500<h2 id="general_features">Language and Target-Independent Features</h2>
501<!-- ======================================================================= -->
502
503
504<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
505<h3 id="diagnostics">Controlling Errors and Warnings</h3>
506<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
507
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]508<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]509emit errors and warning messages, and how they are displayed to the console.</p>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]510
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]511<h4 id="diagnostics_display">Controlling How Clang Displays Diagnostics</h4>
Chris Lattner65a795b2009-04-20 06:00:23 +0000[diff] [blame]512
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]513<p>When Clang emits a diagnostic, it includes rich information in the output,
514and gives you fine-grain control over which information is printed. Clang has
515the ability to print this information, and these are the options that control
516it:</p>
517
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]518<ol>
519<li>A file/line/column indicator that shows exactly where the diagnostic occurs
520 in your code [<a href="#opt_fshow-column">-fshow-column</a>, <a
521 href="#opt_fshow-source-location">-fshow-source-location</a>].</li>
522<li>A categorization of the diagnostic as a note, warning, error, or fatal
523 error.</li>
524<li>A text string that describes what the problem is.</li>
525<li>An option that indicates how to control the diagnostic (for diagnostics that
526 support it) [<a
527 href="#opt_fdiagnostics-show-option">-fdiagnostics-show-option</a>].</li>
Chris Lattner3f145382010-05-24 21:35:18 +0000[diff] [blame]528<li>A <a href="#diagnostics_categories">high-level category</a> for the
529 diagnostic for clients that want to group diagnostics by class (for
530 diagnostics that support it) [<a
Chris Lattner28a43a42010-05-05 01:35:28 +0000[diff] [blame]531 href="#opt_fdiagnostics-show-category">-fdiagnostics-show-category</a>].</li>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]532<li>The line of source code that the issue occurs on, along with a caret and
533 ranges that indicate the important locations [<a
534 href="opt_fcaret-diagnostics">-fcaret-diagnostics</a>].</li>
535<li>"FixIt" information, which is a concise explanation of how to fix the
536 problem (when Clang is certain it knows) [<a
537 href="opt_fdiagnostics-fixit-info">-fdiagnostics-fixit-info</a>].</li>
538<li>A machine-parsable representation of the ranges involved (off by
539 default) [<a
Chris Lattner2a9cc232009-04-21 05:35:32 +0000[diff] [blame]540 href="opt_fdiagnostics-print-source-range-info">-fdiagnostics-print-source-range-info</a>].</li>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]541</ol>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]542
543<p>For more information please see <a href="#cl_diag_formatting">Formatting of
544Diagnostics</a>.</p>
545
Chris Lattner3f145382010-05-24 21:35:18 +0000[diff] [blame]546
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]547<h4 id="diagnostics_mappings">Diagnostic Mappings</h4>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]548
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]549<p>All diagnostics are mapped into one of these 5 classes:</p>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]550
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]551<ul>
552<li>Ignored</li>
553<li>Note</li>
554<li>Warning</li>
555<li>Error</li>
556<li>Fatal</li>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]557</ul>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]558
Chris Lattner3f145382010-05-24 21:35:18 +0000[diff] [blame]559<h4 id="diagnostics_categories">Diagnostic Categories</h4>
560
561<p>Though not shown by default, diagnostics may each be associated with a
562 high-level category. This category is intended to make it possible to triage
563 builds that produce a large number of errors or warnings in a grouped way.
564</p>
565
566<p>Categories are not shown by default, but they can be turned on with the
567<a href="#opt_fdiagnostics-show-category">-fdiagnostics-show-category</a> option.
568When set to "<tt>name</tt>", the category is printed textually in the diagnostic
569output. When it is set to "<tt>id</tt>", a category number is printed. The
570mapping of category names to category id's can be obtained by running '<tt>clang
571 --print-diagnostic-categories</tt>'.
572</p>
573
574<h4 id="diagnostics_commandline">Controlling Diagnostics via Command Line
575 Flags</h4>
Chris Lattner8217f4e2009-04-20 06:26:18 +0000[diff] [blame]576
577<p>-W flags, -pedantic, etc</p>
578
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]579<h4 id="diagnostics_pragmas">Controlling Diagnostics via Pragmas</h4>
580
581<p>Clang can also control what diagnostics are enabled through the use of
582pragmas in the source code. This is useful for turning off specific warnings
583in a section of source code. Clang supports GCC's pragma for compatibility
584with existing source code, as well as several extensions. </p>
585
586<p>The pragma may control any warning that can be used from the command line.
587Warnings may be set to ignored, warning, error, or fatal. The following
588example code will tell Clang or GCC to ignore the -Wall warnings:</p>
589
590<pre>
591#pragma GCC diagnostic ignored "-Wall"
592</pre>
593
Chris Lattnerd462b6a2011-01-24 03:47:59 +0000[diff] [blame]594<p>In addition to all of the functionality provided by GCC's pragma, Clang
Chris Lattner3401cf82009-07-12 21:22:10 +0000[diff] [blame]595also allows you to push and pop the current warning state. This is particularly
596useful when writing a header file that will be compiled by other people, because
597you don't know what warning flags they build with.</p>
598
599<p>In the below example
600-Wmultichar is ignored for only a single line of code, after which the
601diagnostics return to whatever state had previously existed.</p>
602
603<pre>
604#pragma clang diagnostic push
605#pragma clang diagnostic ignored "-Wmultichar"
606
607char b = 'df'; // no warning.
608
609#pragma clang diagnostic pop
610</pre>
611
612<p>The push and pop pragmas will save and restore the full diagnostic state of
613the compiler, regardless of how it was set. That means that it is possible to
614use push and pop around GCC compatible diagnostics and Clang will push and pop
615them appropriately, while GCC will ignore the pushes and pops as unknown
616pragmas. It should be noted that while Clang supports the GCC pragma, Clang and
617GCC do not support the exact same set of warnings, so even when using GCC
618compatible #pragmas there is no guarantee that they will have identical behaviour
619on both compilers. </p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]620
Ted Kremenek2fb11eb2010-08-24 18:12:35 +0000[diff] [blame]621<h4 id="analyzer_diagnositics">Controlling Static Analyzer Diagnostics</h4>
622
623<p>While not strictly part of the compiler, the diagnostics from Clang's <a
624href="http://clang-analyzer.llvm.org">static analyzer</a> can also be influenced
625by the user via changes to the source code. This can be done in two ways:
626
627<ul>
628
629<li id="analyzer_annotations"><b>Annotations</b>: The static analyzer recognizes various GCC-style
630attributes (e.g., <tt>__attribute__((nonnull)))</tt>) that can either suppress
631static analyzer warnings or teach the analyzer about code invariants which
632enable it to find more bugs. While many of these attributes are standard GCC
Chris Lattnerd462b6a2011-01-24 03:47:59 +0000[diff] [blame]633attributes, additional ones have been added to Clang to specifically support the
Ted Kremenek2fb11eb2010-08-24 18:12:35 +0000[diff] [blame]634static analyzer. Detailed information on these annotations can be found in the
635<a href="http://clang-analyzer.llvm.org/annotations.html">analyzer's
636documentation</a>.</li>
637
638<li><b><tt>__clang_analyzer__</tt></b>: When the static analyzer is using Clang
639to parse source files, it implicitly defines the preprocessor macro
640<tt>__clang_analyzer__</tt>. While discouraged, code can use this macro to
641selectively exclude code the analyzer examines. Here is an example:
642
643<pre>
644#ifndef __clang_analyzer__
645// Code not to be analyzed
646#endif
647</pre>
648
649In general, this usage is discouraged. Instead, we prefer that users file bugs
650against the analyzer when it flags false positives. There is also active
651discussion of allowing users in the future to selectively silence specific
Chris Lattnerd462b6a2011-01-24 03:47:59 +0000[diff] [blame]652analyzer warnings (some of which can already be done using <a
Ted Kremenek2fb11eb2010-08-24 18:12:35 +0000[diff] [blame]653href="analyzer_annotations">annotations</a>).</li>
654
655</ul>
656
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]657<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
658<h3 id="precompiledheaders">Precompiled Headers</h3>
659<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
660
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]661<p><a href="http://en.wikipedia.org/wiki/Precompiled_header">Precompiled
662headers</a> are a general approach employed by many compilers to reduce
663compilation time. The underlying motivation of the approach is that it is
664common for the same (and often large) header files to be included by
665multiple source files. Consequently, compile times can often be greatly improved
666by caching some of the (redundant) work done by a compiler to process headers.
667Precompiled header files, which represent one of many ways to implement
668this optimization, are literally files that represent an on-disk cache that
669contains the vital information necessary to reduce some of the work
670needed to process a corresponding header file. While details of precompiled
Chris Lattnerd462b6a2011-01-24 03:47:59 +0000[diff] [blame]671headers vary between compilers, precompiled headers have been shown to be
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]672highly effective at speeding up program compilation on systems with very large
673system headers (e.g., Mac OS/X).</p>
674
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]675<h4>Generating a PCH File</h4>
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]676
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]677<p>To generate a PCH file using Clang, one invokes Clang with
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]678the <b><tt>-x <i>&lt;language&gt;</i>-header</tt></b> option. This mirrors the
679interface in GCC for generating PCH files:</p>
680
681<pre>
682 $ gcc -x c-header test.h -o test.h.gch
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]683 $ clang -x c-header test.h -o test.h.pch
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]684</pre>
685
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]686<h4>Using a PCH File</h4>
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]687
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]688<p>A PCH file can then be used as a prefix header when a
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]689<b><tt>-include</tt></b> option is passed to <tt>clang</tt>:</p>
690
691<pre>
692 $ clang -include test.h test.c -o test
693</pre>
694
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]695<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]696is 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]697will be processed from the PCH file. Otherwise, Clang falls back to
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]698directly processing the content of <tt>test.h</tt>. This mirrors the behavior of
699GCC.</p>
700
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]701<p><b>NOTE:</b> Clang does <em>not</em> automatically use PCH files
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]702for headers that are directly included within a source file. For example:</p>
703
704<pre>
Chris Lattnere42ec542009-06-13 20:35:58 +0000[diff] [blame]705 $ clang -x c-header test.h -o test.h.pch
Chris Lattner5c3074f2009-04-20 04:37:38 +0000[diff] [blame]706 $ cat test.c
707 #include "test.h"
708 $ clang test.c -o test
709</pre>
710
Douglas Gregorf4d59532009-06-03 22:37:00 +0000[diff] [blame]711<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]712<tt>test.h</tt> since <tt>test.h</tt> was included directly in the source file
713and not specified on the command line using <tt>-include</tt>.</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]714
Douglas Gregore650c8c2009-07-07 00:12:59 +0000[diff] [blame]715<h4>Relocatable PCH Files</h4>
716<p>It is sometimes necessary to build a precompiled header from headers that
717are not yet in their final, installed locations. For example, one might build a
718precompiled header within the build tree that is then meant to be installed
719alongside the headers. Clang permits the creation of "relocatable" precompiled
720headers, which are built with a given path (into the build directory) and can
721later be used from an installed location.</p>
722
723<p>To build a relocatable precompiled header, place your headers into a
724subdirectory whose structure mimics the installed location. For example, if you
725want to build a precompiled header for the header <code>mylib.h</code> that
726will be installed into <code>/usr/include</code>, create a subdirectory
727<code>build/usr/include</code> and place the header <code>mylib.h</code> into
728that subdirectory. If <code>mylib.h</code> depends on other headers, then
729they can be stored within <code>build/usr/include</code> in a way that mimics
730the installed location.</p>
731
732<p>Building a relocatable precompiled header requires two additional arguments.
733First, pass the <code>--relocatable-pch</code> flag to indicate that the
734resulting PCH file should be relocatable. Second, pass
735<code>-isysroot /path/to/build</code>, which makes all includes for your
736library relative to the build directory. For example:</p>
737
738<pre>
739 # clang -x c-header --relocatable-pch -isysroot /path/to/build /path/to/build/mylib.h mylib.h.pch
740</pre>
741
742<p>When loading the relocatable PCH file, the various headers used in the PCH
743file are found from the system header root. For example, <code>mylib.h</code>
744can be found in <code>/usr/include/mylib.h</code>. If the headers are installed
745in some other system root, the <code>-isysroot</code> option can be used provide
746a different system root from which the headers will be based. For example,
747<code>-isysroot /Developer/SDKs/MacOSX10.4u.sdk</code> will look for
748<code>mylib.h</code> in
749<code>/Developer/SDKs/MacOSX10.4u.sdk/usr/include/mylib.h</code>.</p>
750
751<p>Relocatable precompiled headers are intended to be used in a limited number
752of cases where the compilation environment is tightly controlled and the
753precompiled header cannot be generated after headers have been installed.
754Relocatable precompiled headers also have some performance impact, because
755the difference in location between the header locations at PCH build time vs.
756at the time of PCH use requires one of the PCH optimizations,
757<code>stat()</code> caching, to be disabled. However, this change is only
758likely to affect PCH files that reference a large number of headers.</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]759
Mike Stump53664ca2009-12-14 23:53:10 +0000[diff] [blame]760<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
761<h3 id="codegen">Controlling Code Generation</h3>
762<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
763
764<p>Clang provides a number of ways to control code generation. The options are listed below.</p>
765
766<!-- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -->
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]767<dl>
Mike Stump53664ca2009-12-14 23:53:10 +0000[diff] [blame]768<dt id="opt_fcatch-undefined-behavior"><b>-fcatch-undefined-behavior</b>: Turn
769on runtime code generation to check for undefined behavior.</dt>
770
771<dd>This option, which defaults to off, controls whether or not Clang
Chris Lattnerd462b6a2011-01-24 03:47:59 +0000[diff] [blame]772adds runtime checks for undefined runtime behavior. If a check fails,
Mike Stump53664ca2009-12-14 23:53:10 +0000[diff] [blame]773<tt>__builtin_trap()</tt> is used to indicate failure.
774The checks are:
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]775<ul>
Chris Lattnerd462b6a2011-01-24 03:47:59 +0000[diff] [blame]776<li>Subscripting where the static type of one operand is a variable
Mike Stump88b2a172009-12-16 03:25:12 +0000[diff] [blame]777 which is decayed from an array type and the other operand is
778 greater than the size of the array or less than zero.</li>
Mike Stump53664ca2009-12-14 23:53:10 +0000[diff] [blame]779<li>Shift operators where the amount shifted is greater or equal to the
780 promoted bit-width of the left-hand-side or less than zero.</li>
Mike Stump8f6a3ed2009-12-16 03:18:14 +0000[diff] [blame]781<li>If control flow reaches __builtin_unreachable.
782<li>When llvm implements more __builtin_object_size support, reads and
783 writes for objects that __builtin_object_size indicates we aren't
784 accessing valid memory. Bit-fields and vectors are not yet checked.
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]785</ul>
Mike Stump53664ca2009-12-14 23:53:10 +0000[diff] [blame]786</dd>
787
Nuno Lopesaa526242009-12-17 10:00:52 +0000[diff] [blame]788<dt id="opt_fno-assume-sane-operator-new"><b>-fno-assume-sane-operator-new</b>:
789Don't assume that the C++'s new operator is sane.</dt>
Nuno Lopesb23f20d2009-12-17 10:15:49 +0000[diff] [blame]790<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]791operator will always return a pointer that does not
Nuno Lopesaa526242009-12-17 10:00:52 +0000[diff] [blame]792alias any other pointer when the function returns.</dd>
Evan Chengfda026b2011-04-08 22:18:01 +0000[diff] [blame]793
Evan Cheng4cf4b592011-04-08 22:34:21 +0000[diff] [blame]794<dt id="opt_ftrap-function"><b>-ftrap-function=[name]</b>: Instruct code
Evan Chengfda026b2011-04-08 22:18:01 +0000[diff] [blame]795generator to emit a function call to the specified function name for
796<tt>__builtin_trap()</tt>.</dt>
797
798<dd>LLVM code generator translates <tt>__builtin_trap()</tt> to a trap
799instruction if it is supported by the target ISA. Otherwise, the builtin is
800translated into a call to <tt>abort</tt>. If this option is set, then the code
801generator will always lower the builtin to a call to the specified function
802regardless of whether the target ISA has a trap instruction. This option is
Chris Lattner02e266b2011-04-08 22:58:43 +0000[diff] [blame]803useful for environments (e.g. deeply embedded) where a trap cannot be properly
804handled, or when some custom behavior is desired.</dd>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]805</dl>
Nuno Lopesaa526242009-12-17 10:00:52 +0000[diff] [blame]806
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]807<!-- ======================================================================= -->
808<h2 id="c">C Language Features</h2>
809<!-- ======================================================================= -->
810
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]811<p>The support for standard C in clang is feature-complete except for the C99
812floating-point pragmas.</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]813
814<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]815<h3 id="c_ext">Extensions supported by clang</h3>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]816<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
817
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]818<p>See <a href="LanguageExtensions.html">clang language extensions</a>.</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]819
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]820<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
821<h3 id="c_modes">Differences between various standard modes</h3>
822<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]823
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]824<p>clang supports the -std option, which changes what language mode clang uses.
825The supported modes for C are c89, gnu89, c94, c99, gnu99 and various aliases
826for those modes. If no -std option is specified, clang defaults to gnu99 mode.
827</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]828
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]829<p>Differences between all c* and gnu* modes:</p>
830<ul>
831<li>c* modes define "__STRICT_ANSI__".</li>
Eli Friedman26fa0ed2009-05-27 23:02:57 +0000[diff] [blame]832<li>Target-specific defines not prefixed by underscores, like "linux", are
833defined in gnu* modes.</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]834<li>Trigraphs default to being off in gnu* modes; they can be enabled by the
835-trigraphs option.</li>
836<li>The parser recognizes "asm" and "typeof" as keywords in gnu* modes; the
837variants "__asm__" and "__typeof__" are recognized in all modes.</li>
Eli Friedmanb0ac5452009-05-16 23:17:30 +0000[diff] [blame]838<li>The Apple "blocks" extension is recognized by default in gnu* modes
839on some platforms; it can be enabled in any mode with the "-fblocks"
840option.</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]841</ul>
842
843<p>Differences between *89 and *99 modes:</p>
844<ul>
845<li>The *99 modes default to implementing "inline" as specified in C99, while
846the *89 modes implement the GNU version. This can be overridden for individual
847functions with the __gnu_inline__ attribute.</li>
Eli Friedman26fa0ed2009-05-27 23:02:57 +0000[diff] [blame]848<li>Digraphs are not recognized in c89 mode.</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]849<li>The scope of names defined inside a "for", "if", "switch", "while", or "do"
850statement is different. (example: "if ((struct x {int x;}*)0) {}".)</li>
851<li>__STDC_VERSION__ is not defined in *89 modes.</li>
Eli Friedman26fa0ed2009-05-27 23:02:57 +0000[diff] [blame]852<li>"inline" is not recognized as a keyword in c89 mode.</li>
853<li>"restrict" is not recognized as a keyword in *89 modes.</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]854<li>Commas are allowed in integer constant expressions in *99 modes.</li>
855<li>Arrays which are not lvalues are not implicitly promoted to pointers in
856*89 modes.</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]857<li>Some warnings are different.</li>
858</ul>
859
860<p>c94 mode is identical to c89 mode except that digraphs are enabled in
861c94 mode (FIXME: And __STDC_VERSION__ should be defined!).</p>
862
863<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
864<h3 id="c_unimpl_gcc">GCC extensions not implemented yet</h3>
865<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
866
867<p>clang tries to be compatible with gcc as much as possible, but some gcc
868extensions are not implemented yet:</p>
869
870<ul>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]871
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]872<li>clang does not support #pragma weak
Eli Friedman4da92552009-06-02 08:21:31 +0000[diff] [blame]873(<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]874the uses described in the bug, this is likely to be implemented at some
875point, at least partially.</li>
876
Eli Friedman4da92552009-06-02 08:21:31 +0000[diff] [blame]877<li>clang does not support code generation for local variables pinned to
878registers (<a href="http://llvm.org/bugs/show_bug.cgi?id=3933">bug 3933</a>).
879This is a relatively small feature, so it is likely to be implemented
880relatively soon.</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]881
882<li>clang does not support decimal floating point types (_Decimal32 and
883friends) or fixed-point types (_Fract and friends); nobody has expressed
884interest in these features yet, so it's hard to say when they will be
885implemented.</li>
886
887<li>clang does not support nested functions; this is a complex feature which
888is infrequently used, so it is unlikely to be implemented anytime soon.</li>
889
Daniel Dunbar05fa6292009-05-12 23:12:07 +0000[diff] [blame]890<li>clang does not support global register variables, this is unlikely
Eli Friedmanca130bd2009-06-12 20:04:25 +0000[diff] [blame]891to be implemented soon because it requires additional LLVM backend support.
892</li>
Daniel Dunbar05fa6292009-05-12 23:12:07 +0000[diff] [blame]893
894<li>clang does not support static initialization of flexible array
895members. This appears to be a rarely used extension, but could be
896implemented pending user demand.</li>
897
Eli Friedmanca130bd2009-06-12 20:04:25 +0000[diff] [blame]898<li>clang does not support __builtin_va_arg_pack/__builtin_va_arg_pack_len.
899This is used rarely, but in some potentially interesting places, like the
900glibc headers, so it may be implemented pending user demand. Note that
901because clang pretends to be like GCC 4.2, and this extension was introduced
902in 4.3, the glibc headers will not try to use this extension with clang at
903the moment.</li>
904
905<li>clang does not support the gcc extension for forward-declaring function
906parameters; this has not showed up in any real-world code yet, though, so it
907might never be implemented.</li>
908
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]909</ul>
910
911<p>This is not a complete list; if you find an unsupported extension
912missing from this list, please send an e-mail to cfe-dev. This list
913currently excludes C++; see <a href="#cxx">C++ Language Features</a>.
914Also, this list does not include bugs in mostly-implemented features; please
915see 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">
916bug tracker</a> for known existing bugs (FIXME: Is there a section for
917bug-reporting guidelines somewhere?).</p>
918
919<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
920<h3 id="c_unsupp_gcc">Intentionally unsupported GCC extensions</h3>
921<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
922
Eli Friedman0b326002009-06-12 20:11:05 +0000[diff] [blame]923<ul>
924
Eli Friedman2f3fca02009-06-12 20:08:48 +0000[diff] [blame]925<li>clang does not support the gcc extension that allows variable-length arrays
Chris Lattnerd462b6a2011-01-24 03:47:59 +0000[diff] [blame]926in structures. This is for a few reasons: one, it is tricky
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]927to implement, two, the extension is completely undocumented, and three, the
Chris Lattnerb9d511c2010-10-28 02:20:32 +0000[diff] [blame]928extension appears to be rarely used. Note that clang <em>does</em> support
929flexible array members (arrays with a zero or unspecified size at the end of
930a structure).</li>
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]931
Eli Friedman2f3fca02009-06-12 20:08:48 +0000[diff] [blame]932<li>clang does not support duplicate definitions of a function where one is
Daniel Dunbarc5f928c2009-06-04 18:37:52 +0000[diff] [blame]933inline. This complicates clients of the AST which normally can expect there is
934at most one definition for each function. Source code using this feature should
935be changed to define the inline and out-of-line definitions in separate
Eli Friedman2f3fca02009-06-12 20:08:48 +0000[diff] [blame]936translation units.</li>
Daniel Dunbarc5f928c2009-06-04 18:37:52 +0000[diff] [blame]937
Eli Friedmanca130bd2009-06-12 20:04:25 +0000[diff] [blame]938<li>clang does not have an equivalent to gcc's "fold"; this means that
939clang doesn't accept some constructs gcc might accept in contexts where a
940constant expression is required, like "x-x" where x is a variable, or calls
941to C library functions like strlen.</li>
942
Eli Friedman2f3fca02009-06-12 20:08:48 +0000[diff] [blame]943<li>clang does not support multiple alternative constraints in inline asm; this
Eli Friedmanca130bd2009-06-12 20:04:25 +0000[diff] [blame]944is an extremely obscure feature which would be complicated to implement
Eli Friedman2f3fca02009-06-12 20:08:48 +0000[diff] [blame]945correctly.</li>
Eli Friedmanca130bd2009-06-12 20:04:25 +0000[diff] [blame]946
947<li>clang does not support __builtin_apply and friends; this extension is
948extremely obscure and difficult to implement reliably.</li>
949
Eli Friedman0b326002009-06-12 20:11:05 +0000[diff] [blame]950</ul>
951
Eli Friedman660a5ac2009-04-28 18:48:34 +0000[diff] [blame]952<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
953<h3 id="c_ms">Microsoft extensions</h3>
954<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
955
956<p>clang has some experimental support for extensions from
957Microsoft Visual C++; to enable it, use the -fms-extensions command-line
Eli Friedmana1821b52009-06-08 06:21:03 +0000[diff] [blame]958option. This is the default for Windows targets. Note that the
959support is incomplete; enabling Microsoft extensions will silently drop
960certain constructs (including __declspec and Microsoft-style asm statements).
961</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]962
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]963<ul>
Douglas Gregord3b227d2010-12-14 16:21:49 +0000[diff] [blame]964<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]965is the same as Visual C/C++ 2003. Any number is supported and can greatly affect
966what Windows SDK and c++stdlib headers clang can compile. This option will be
967removed when clang supports the full set of MS extensions required for these
968headers.</li>
969
Daniel Dunbar05fa6292009-05-12 23:12:07 +0000[diff] [blame]970<li>clang does not support the Microsoft extension where anonymous
Daniel Dunbar869e0542009-05-13 00:23:53 +0000[diff] [blame]971record members can be declared using user defined typedefs.</li>
Daniel Dunbar05fa6292009-05-12 23:12:07 +0000[diff] [blame]972
Daniel Dunbar9375ed12009-05-13 21:40:49 +0000[diff] [blame]973<li>clang supports the Microsoft "#pragma pack" feature for
974controlling record layout. GCC also contains support for this feature,
975however where MSVC and GCC are incompatible clang follows the MSVC
976definition.</li>
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]977</ul>
Daniel Dunbar9375ed12009-05-13 21:40:49 +0000[diff] [blame]978
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]979<!-- ======================================================================= -->
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]980<h2 id="target_features">Target-Specific Features and Limitations</h2>
981<!-- ======================================================================= -->
982
983
984<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
985<h3 id="target_arch">CPU Architectures Features and Limitations</h3>
986<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
987
988<!-- ======================== -->
989<h4 id="target_arch_x86">X86</h4>
990<!-- ======================== -->
Daniel Dunbarbcaf7aa2010-09-19 19:26:59 +0000[diff] [blame]991
992<p>The support for X86 (both 32-bit and 64-bit) is considered stable on Darwin
993(Mac OS/X), Linux, FreeBSD, and Dragonfly BSD: it has been tested to correctly
994compile many large C, C++, Objective-C, and Objective-C++ codebases.</p>
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]995
NAKAMURA Takumiddf68ff2011-04-05 04:31:31 +0000[diff] [blame]996<p>On x86_64-mingw32, passing i128(by value) is incompatible to Microsoft x64
997calling conversion. You might need to tweak WinX86_64ABIInfo::classify()
998in lib/CodeGen/TargetInfo.cpp.</p>
NAKAMURA Takumia75fdb22011-04-04 15:02:41 +0000[diff] [blame]999
Eli Friedman3b658d32009-06-08 05:12:39 +0000[diff] [blame]1000<!-- ======================== -->
1001<h4 id="target_arch_arm">ARM</h4>
1002<!-- ======================== -->
Daniel Dunbarbcaf7aa2010-09-19 19:26:59 +0000[diff] [blame]1003
1004<p>The support for ARM (specifically ARMv6 and ARMv7) is considered stable on
1005Darwin (iOS): it has been tested to correctly compile many large C, C++,
Bob Wilson4ea8dfa2011-01-10 17:55:14 +0000[diff] [blame]1006Objective-C, and Objective-C++ codebases. Clang only supports a limited number
1007of ARM architectures. It does not yet fully support ARMv5, for example.</p>
Eli Friedman3b658d32009-06-08 05:12:39 +0000[diff] [blame]1008
1009<!-- ======================== -->
1010<h4 id="target_arch_other">Other platforms</h4>
1011<!-- ======================== -->
1012clang currently contains some support for PPC and Sparc; however, significant
1013pieces of code generation are still missing, and they haven't undergone
1014significant testing.
1015
Eli Friedman3b658d32009-06-08 05:12:39 +0000[diff] [blame]1016<p>clang contains limited support for the MSP430 embedded processor, but both
1017the clang support and the LLVM backend support are highly experimental.
1018
1019<p>Other platforms are completely unsupported at the moment. Adding the
1020minimal support needed for parsing and semantic analysis on a new platform
1021is quite easy; see lib/Basic/Targets.cpp in the clang source tree. This level
1022of support is also sufficient for conversion to LLVM IR for simple programs.
1023Proper support for conversion to LLVM IR requires adding code to
1024lib/CodeGen/CGCall.cpp at the moment; this is likely to change soon, though.
1025Generating assembly requires a suitable LLVM backend.
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1026
1027<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1028<h3 id="target_os">Operating System Features and Limitations</h3>
1029<!-- = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = -->
1030
1031<!-- ======================================= -->
1032<h4 id="target_os_darwin">Darwin (Mac OS/X)</h4>
1033<!-- ======================================= -->
1034
1035<p>No __thread support, 64-bit ObjC support requires SL tools.</p>
1036
NAKAMURA Takumia75fdb22011-04-04 15:02:41 +0000[diff] [blame]1037<!-- ======================================= -->
1038<h4 id="target_os_win32">Windows</h4>
1039<!-- ======================================= -->
1040
1041<p>Experimental supports are on Cygming.</p>
1042
1043<h5>Cygwin</h5>
1044
1045<p>Clang works on Cygwin-1.7.</p>
1046
1047<h5>MinGW32</h5>
1048
1049<p>Clang works on some mingw32 distributions.
1050Clang assumes directories as below;</p>
1051
1052<ul>
1053<li><tt>C:/mingw/include</tt></li>
1054<li><tt>C:/mingw/lib</tt></li>
1055<li><tt>C:/mingw/lib/gcc/mingw32/4.[3-5].0/include/c++</tt></li>
1056</ul>
1057
1058<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>
1059
1060<h5>MinGW-w64</h5>
1061
1062<p>For x32(i686-w64-mingw32), it is not supported yet.</p>
1063
Chris Lattnerfc8f0e12011-04-15 05:22:18 +0000[diff] [blame]1064<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>
NAKAMURA Takumia75fdb22011-04-04 15:02:41 +0000[diff] [blame]1065
1066<ul>
1067<li><tt>C:/mingw/x86_64-w64-mingw32/include</tt></li>
1068<li><tt>C:/mingw/x86_64-w64-mingw32/include/c++/4.5.[23]</tt></li>
1069<li>GCC driver &quot;gcc.exe&quot; to build x86_64-w64-mingw32 binary.</li>
1070</ul>
1071
1072<p><a href="http://llvm.org/bugs/show_bug.cgi?id=8833">Some tests might fail</a>
1073on x64.</p>
1074
Chris Lattnercf17d9d2009-04-20 04:23:09 +0000[diff] [blame]1075</div>
1076</body>
1077</html>