Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 62339073127df4579905f551f61c132cf21d2aad / . / docs / FAQ.html
blob: 7294cdd1a4294bbdafc5cdb142202b13d1dbe3e9 [file] [log] [blame]
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]5 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]6 <title>LLVM: Frequently Asked Questions</title>
Misha Brukman7ce62cc2004-06-01 18:51:03 +0000[diff] [blame]7 <style type="text/css">
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]8 @import url("llvm.css");
9 .question { font-weight: bold }
10 .answer { margin-left: 2em }
11 </style>
12</head>
13<body>
John Criswellc310f622003-10-13 16:13:06 +0000[diff] [blame]14
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]15<h1>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]16 LLVM: Frequently Asked Questions
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]17</h1>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]18
19<ol>
20 <li><a href="#license">License</a>
21 <ol>
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]22 <li>Why are the LLVM source code and the front-end distributed under
23 different licenses?</li>
24
25 <li>Does the University of Illinois Open Source License really qualify as an
26 "open source" license?</li>
27
28 <li>Can I modify LLVM source code and redistribute the modified source?</li>
29
30 <li>Can I modify LLVM source code and redistribute binaries or other tools
31 based on it, without redistributing the source?</li>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]32 </ol></li>
33
34 <li><a href="#source">Source code</a>
35 <ol>
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]36 <li>In what language is LLVM written?</li>
37
38 <li>How portable is the LLVM source code?</li>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]39 </ol></li>
40
41 <li><a href="#build">Build Problems</a>
42 <ol>
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]43 <li>When I run configure, it finds the wrong C compiler.</li>
44
45 <li>The <tt>configure</tt> script finds the right C compiler, but it uses
46 the LLVM linker from a previous build. What do I do?</li>
47
48 <li>When creating a dynamic library, I get a strange GLIBC error.</li>
49
50 <li>I've updated my source tree from Subversion, and now my build is trying
51 to use a file/directory that doesn't exist.</li>
52
53 <li>I've modified a Makefile in my source tree, but my build tree keeps
54 using the old version. What do I do?</li>
55
56 <li>I've upgraded to a new version of LLVM, and I get strange build
57 errors.</li>
58
59 <li>I've built LLVM and am testing it, but the tests freeze.</li>
60
61 <li>Why do test results differ when I perform different types of
62 builds?</li>
63
64 <li>Compiling LLVM with GCC 3.3.2 fails, what should I do?</li>
65
66 <li>Compiling LLVM with GCC succeeds, but the resulting tools do not work,
67 what can be wrong?</li>
68
69 <li>When I use the test suite, all of the C Backend tests fail. What is
70 wrong?</li>
71
72 <li>After Subversion update, rebuilding gives the error "No rule to make
73 target".</li>
74
Bill Wendlingf1594022009-04-07 18:51:13 +0000[diff] [blame]75 <li><a href="#srcdir-objdir">When I compile LLVM-GCC with srcdir == objdir,
76 it fails. Why?</a></li>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]77 </ol></li>
John Criswell76c1e382003-11-18 16:08:49 +0000[diff] [blame]78
Reid Spencer501bfee2006-04-26 14:52:19 +0000[diff] [blame]79 <li><a href="#felangs">Source Languages</a>
80 <ol>
81 <li><a href="#langs">What source languages are supported?</a></li>
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]82
Gordon Henriksene5079052008-02-22 21:55:51 +0000[diff] [blame]83 <li><a href="#langirgen">I'd like to write a self-hosting LLVM compiler. How
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]84 should I interface with the LLVM middle-end optimizers and back-end code
85 generators?</a></li>
86
Reid Spencer501bfee2006-04-26 14:52:19 +0000[diff] [blame]87 <li><a href="#langhlsupp">What support is there for higher level source
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]88 language constructs for building a compiler?</a></li>
89
Reid Spencere00906f2006-08-10 20:15:58 +0000[diff] [blame]90 <li><a href="GetElementPtr.html">I don't understand the GetElementPtr
91 instruction. Help!</a></li>
Reid Spencer501bfee2006-04-26 14:52:19 +0000[diff] [blame]92 </ol>
93
Chris Lattnercc33d702003-11-19 05:53:12 +0000[diff] [blame]94 <li><a href="#cfe">Using the GCC Front End</a>
John Criswell76c1e382003-11-18 16:08:49 +0000[diff] [blame]95 <ol>
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]96 <li>When I compile software that uses a configure script, the configure
97 script thinks my system has all of the header files and libraries it is
98 testing for. How do I get configure to work correctly?</li>
John Criswell76c1e382003-11-18 16:08:49 +0000[diff] [blame]99
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]100 <li>When I compile code using the LLVM GCC front end, it complains that it
101 cannot find libcrtend.a?</li>
Tanya Lattner14fc5c12005-04-25 20:36:56 +0000[diff] [blame]102
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]103 <li>How can I disable all optimizations when compiling code using the LLVM
104 GCC front end?</li>
Tanya Lattner14fc5c12005-04-25 20:36:56 +0000[diff] [blame]105
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]106 <li><a href="#translatecxx">Can I use LLVM to convert C++ code to C
107 code?</a></li>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]108
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]109 <li><a href="#platformindependent">Can I compile C or C++ code to
110 platform-independent LLVM bitcode?</a></li>
John Criswell76c1e382003-11-18 16:08:49 +0000[diff] [blame]111 </ol>
112 </li>
Chris Lattnercc33d702003-11-19 05:53:12 +0000[diff] [blame]113
114 <li><a href="#cfe_code">Questions about code generated by the GCC front-end</a>
115 <ol>
Chris Lattner5a53c5d2005-02-25 20:30:21 +0000[diff] [blame]116 <li><a href="#iosinit">What is this <tt>llvm.global_ctors</tt> and
Chris Lattnerc50bbc92004-03-29 19:14:35 +0000[diff] [blame]117 <tt>_GLOBAL__I__tmp_webcompile...</tt> stuff that happens when I
Chris Lattner5a53c5d2005-02-25 20:30:21 +0000[diff] [blame]118 #include &lt;iostream&gt;?</a></li>
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]119
Chris Lattner5a53c5d2005-02-25 20:30:21 +0000[diff] [blame]120 <li><a href="#codedce">Where did all of my code go??</a></li>
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]121
122 <li><a href="#undef">What is this "<tt>undef</tt>" thing that shows up in
123 my code?</a></li>
Chris Lattner2c6f9f72009-06-30 17:10:19 +0000[diff] [blame]124
125 <li><a href="#callconvwrong">Why does instcombine + simplifycfg turn
126 a call to a function with a mismatched calling convention into "unreachable"?
127 Why not make the verifier reject it?</a></li>
Chris Lattnercc33d702003-11-19 05:53:12 +0000[diff] [blame]128 </ol>
129 </li>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]130</ol>
131
Chris Lattner7911ce22004-05-23 21:07:27 +0000[diff] [blame]132<div class="doc_author">
NAKAMURA Takumib9a33632011-04-09 02:13:37 +0000[diff] [blame]133 <p>Written by <a href="http://llvm.org/">The LLVM Team</a></p>
Chris Lattner7911ce22004-05-23 21:07:27 +0000[diff] [blame]134</div>
135
136
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]137<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]138<h2>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]139 <a name="license">License</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]140</h2>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]141<!-- *********************************************************************** -->
142
143<div class="question">
144<p>Why are the LLVM source code and the front-end distributed under different
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]145 licenses?</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]146</div>
147
148<div class="answer">
149<p>The C/C++ front-ends are based on GCC and must be distributed under the GPL.
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]150 Our aim is to distribute LLVM source code under a <em>much less
151 restrictive</em> license, in particular one that does not compel users who
152 distribute tools based on modifying the source to redistribute the modified
153 source code as well.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]154</div>
155
156<div class="question">
157<p>Does the University of Illinois Open Source License really qualify as an
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]158 "open source" license?</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]159</div>
160
161<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]162<p>Yes, the license
163 is <a href="http://www.opensource.org/licenses/UoI-NCSA.php">certified</a> by
164 the Open Source Initiative (OSI).</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]165</div>
166
167<div class="question">
168<p>Can I modify LLVM source code and redistribute the modified source?</p>
169</div>
170
171<div class="answer">
172<p>Yes. The modified source distribution must retain the copyright notice and
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]173 follow the three bulletted conditions listed in
174 the <a href="http://llvm.org/svn/llvm-project/llvm/trunk/LICENSE.TXT">LLVM
175 license</a>.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]176</div>
177
178<div class="question">
179<p>Can I modify LLVM source code and redistribute binaries or other tools based
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]180 on it, without redistributing the source?</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]181</div>
182
183<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]184<p>Yes. This is why we distribute LLVM under a less restrictive license than
185 GPL, as explained in the first question above.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]186</div>
187
188<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]189<h2>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]190 <a name="source">Source Code</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]191</h2>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]192<!-- *********************************************************************** -->
193
194<div class="question">
195<p>In what language is LLVM written?</p>
196</div>
197
198<div class="answer">
199<p>All of the LLVM tools and libraries are written in C++ with extensive use of
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]200 the STL.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]201</div>
202
203<div class="question">
204<p>How portable is the LLVM source code?</p>
205</div>
206
207<div class="answer">
208<p>The LLVM source code should be portable to most modern UNIX-like operating
209systems. Most of the code is written in standard C++ with operating system
210services abstracted to a support library. The tools required to build and test
211LLVM have been ported to a plethora of platforms.</p>
212
213<p>Some porting problems may exist in the following areas:</p>
214
215<ul>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]216 <li>The GCC front end code is not as portable as the LLVM suite, so it may not
Bill Wendlinge9a6c352007-09-22 09:54:47 +0000[diff] [blame]217 compile as well on unsupported platforms.</li>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]218
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]219 <li>The LLVM build system relies heavily on UNIX shell tools, like the Bourne
Bill Wendlinge9a6c352007-09-22 09:54:47 +0000[diff] [blame]220 Shell and sed. Porting to systems without these tools (MacOS 9, Plan 9)
221 will require more effort.</li>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]222</ul>
223
224</div>
225
226<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]227<h2>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]228 <a name="build">Build Problems</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]229</h2>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]230<!-- *********************************************************************** -->
231
232<div class="question">
233<p>When I run configure, it finds the wrong C compiler.</p>
234</div>
235
236<div class="answer">
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]237<p>The <tt>configure</tt> script attempts to locate first <tt>gcc</tt> and then
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]238 <tt>cc</tt>, unless it finds compiler paths set in <tt>CC</tt>
239 and <tt>CXX</tt> for the C and C++ compiler, respectively.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]240
241<p>If <tt>configure</tt> finds the wrong compiler, either adjust your
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]242 <tt>PATH</tt> environment variable or set <tt>CC</tt> and <tt>CXX</tt>
243 explicitly.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]244
245</div>
246
247<div class="question">
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]248<p>The <tt>configure</tt> script finds the right C compiler, but it uses the
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]249 LLVM linker from a previous build. What do I do?</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]250</div>
251
252<div class="answer">
253<p>The <tt>configure</tt> script uses the <tt>PATH</tt> to find executables, so
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]254 if it's grabbing the wrong linker/assembler/etc, there are two ways to fix
255 it:</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]256
257<ol>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]258 <li><p>Adjust your <tt>PATH</tt> environment variable so that the correct
Bill Wendlinge9a6c352007-09-22 09:54:47 +0000[diff] [blame]259 program appears first in the <tt>PATH</tt>. This may work, but may not be
260 convenient when you want them <i>first</i> in your path for other
261 work.</p></li>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]262
263 <li><p>Run <tt>configure</tt> with an alternative <tt>PATH</tt> that is
Bill Wendlinge9a6c352007-09-22 09:54:47 +0000[diff] [blame]264 correct. In a Borne compatible shell, the syntax would be:</p>
265
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]266<pre class="doc_code">
Bill Wendlinge9a6c352007-09-22 09:54:47 +0000[diff] [blame]267% PATH=[the path without the bad program] ./configure ...
268</pre>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]269
270 <p>This is still somewhat inconvenient, but it allows <tt>configure</tt>
Bill Wendlinge9a6c352007-09-22 09:54:47 +0000[diff] [blame]271 to do its work without having to adjust your <tt>PATH</tt>
272 permanently.</p></li>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]273</ol>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]274</div>
275
276<div class="question">
277<p>When creating a dynamic library, I get a strange GLIBC error.</p>
278</div>
279
280<div class="answer">
281<p>Under some operating systems (i.e. Linux), libtool does not work correctly if
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]282 GCC was compiled with the --disable-shared option. To work around this,
283 install your own version of GCC that has shared libraries enabled by
284 default.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]285</div>
286
287<div class="question">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]288<p>I've updated my source tree from Subversion, and now my build is trying to
289 use a file/directory that doesn't exist.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]290</div>
291
292<div class="answer">
293<p>You need to re-run configure in your object directory. When new Makefiles
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]294 are added to the source tree, they have to be copied over to the object tree
295 in order to be used by the build.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]296</div>
297
298<div class="question">
299<p>I've modified a Makefile in my source tree, but my build tree keeps using the
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]300 old version. What do I do?</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]301</div>
302
303<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]304<p>If the Makefile already exists in your object tree, you can just run the
305 following command in the top level directory of your object tree:</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]306
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]307<pre class="doc_code">
308% ./config.status &lt;relative path to Makefile&gt;
309</pre>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]310
311<p>If the Makefile is new, you will have to modify the configure script to copy
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]312 it over.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]313</div>
314
315<div class="question">
316<p>I've upgraded to a new version of LLVM, and I get strange build errors.</p>
317</div>
318
319<div class="answer">
320
321<p>Sometimes, changes to the LLVM source code alters how the build system works.
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]322 Changes in libtool, autoconf, or header file dependencies are especially
323 prone to this sort of problem.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]324
325<p>The best thing to try is to remove the old files and re-build. In most
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]326 cases, this takes care of the problem. To do this, just type <tt>make
327 clean</tt> and then <tt>make</tt> in the directory that fails to build.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]328</div>
329
330<div class="question">
331<p>I've built LLVM and am testing it, but the tests freeze.</p>
332</div>
333
334<div class="answer">
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]335<p>This is most likely occurring because you built a profile or release
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]336 (optimized) build of LLVM and have not specified the same information on the
337 <tt>gmake</tt> command line.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]338
339<p>For example, if you built LLVM with the command:</p>
340
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]341<pre class="doc_code">
342% gmake ENABLE_PROFILING=1
343</pre>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]344
345<p>...then you must run the tests with the following commands:</p>
346
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]347<pre class="doc_code">
Bill Wendlingd6a68eb2007-05-29 09:24:33 +0000[diff] [blame]348% cd llvm/test
349% gmake ENABLE_PROFILING=1
350</pre>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]351</div>
352
353<div class="question">
354<p>Why do test results differ when I perform different types of builds?</p>
355</div>
356
357<div class="answer">
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]358<p>The LLVM test suite is dependent upon several features of the LLVM tools and
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]359 libraries.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]360
361<p>First, the debugging assertions in code are not enabled in optimized or
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]362 profiling builds. Hence, tests that used to fail may pass.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]363
364<p>Second, some tests may rely upon debugging options or behavior that is only
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]365 available in the debug build. These tests will fail in an optimized or
366 profile build.</p>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]367</div>
368
Chris Lattner8a0b9242003-12-08 05:43:19 +0000[diff] [blame]369<div class="question">
Chris Lattner306acee2003-12-22 04:06:12 +0000[diff] [blame]370<p>Compiling LLVM with GCC 3.3.2 fails, what should I do?</p>
Chris Lattner8a0b9242003-12-08 05:43:19 +0000[diff] [blame]371</div>
372
373<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]374<p>This is <a href="http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13392">a bug in
375 GCC</a>, and affects projects other than LLVM. Try upgrading or downgrading
376 your GCC.</p>
Chris Lattner8a0b9242003-12-08 05:43:19 +0000[diff] [blame]377</div>
378
John Criswelld1799612004-03-29 20:23:11 +0000[diff] [blame]379<div class="question">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]380<p>Compiling LLVM with GCC succeeds, but the resulting tools do not work, what
381 can be wrong?</p>
Gabor Greif54820ce2009-03-02 19:08:05 +0000[diff] [blame]382</div>
383
384<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]385<p>Several versions of GCC have shown a weakness in miscompiling the LLVM
386 codebase. Please consult your compiler version (<tt>gcc --version</tt>) to
387 find out whether it is <a href="GettingStarted.html#brokengcc">broken</a>.
388 If so, your only option is to upgrade GCC to a known good version.</p>
Gabor Greif54820ce2009-03-02 19:08:05 +0000[diff] [blame]389</div>
390
391<div class="question">
Bill Wendlinge9a6c352007-09-22 09:54:47 +0000[diff] [blame]392<p>After Subversion update, rebuilding gives the error "No rule to make
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]393 target".</p>
Misha Brukman1739aec2004-09-09 16:36:47 +0000[diff] [blame]394</div>
395
396<div class="answer">
397<p>If the error is of the form:</p>
398
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]399<pre class="doc_code">
Misha Brukman1739aec2004-09-09 16:36:47 +0000[diff] [blame]400gmake[2]: *** No rule to make target `/path/to/somefile', needed by
401`/path/to/another/file.d'.<br>
402Stop.
Bill Wendlingd6a68eb2007-05-29 09:24:33 +0000[diff] [blame]403</pre>
Misha Brukman1739aec2004-09-09 16:36:47 +0000[diff] [blame]404
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]405<p>This may occur anytime files are moved within the Subversion repository or
406 removed entirely. In this case, the best solution is to erase all
407 <tt>.d</tt> files, which list dependencies for source files, and rebuild:</p>
Misha Brukman1739aec2004-09-09 16:36:47 +0000[diff] [blame]408
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]409<pre class="doc_code">
Misha Brukman1739aec2004-09-09 16:36:47 +0000[diff] [blame]410% cd $LLVM_OBJ_DIR
411% rm -f `find . -name \*\.d`
412% gmake
413</pre>
Misha Brukman1739aec2004-09-09 16:36:47 +0000[diff] [blame]414
415<p>In other cases, it may be necessary to run <tt>make clean</tt> before
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]416 rebuilding.</p>
Misha Brukman1739aec2004-09-09 16:36:47 +0000[diff] [blame]417</div>
418
Bill Wendlingf1594022009-04-07 18:51:13 +0000[diff] [blame]419<div class="question">
Bill Wendlinga0bd8102009-04-07 18:54:06 +0000[diff] [blame]420<p><a name="srcdir-objdir">When I compile LLVM-GCC with srcdir == objdir, it
Bill Wendlingf1594022009-04-07 18:51:13 +0000[diff] [blame]421 fails. Why?</a></p>
422</div>
423
424<div class="answer">
425<p>The <tt>GNUmakefile</tt> in the top-level directory of LLVM-GCC is a special
426 <tt>Makefile</tt> used by Apple to invoke the <tt>build_gcc</tt> script after
Dan Gohmanabbf7cb2010-02-25 23:41:41 +0000[diff] [blame]427 setting up a special environment. This has the unfortunate side-effect that
Bill Wendlingf1594022009-04-07 18:51:13 +0000[diff] [blame]428 trying to build LLVM-GCC with srcdir == objdir in a "non-Apple way" invokes
429 the <tt>GNUmakefile</tt> instead of <tt>Makefile</tt>. Because the
430 environment isn't set up correctly to do this, the build fails.</p>
431
432<p>People not building LLVM-GCC the "Apple way" need to build LLVM-GCC with
433 srcdir != objdir, or simply remove the GNUmakefile entirely.</p>
434
435<p>We regret the inconvenience.</p>
436</div>
437
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]438<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]439<h2>
440 <a name="felangs">Source Languages</a>
441</h2>
Reid Spencer501bfee2006-04-26 14:52:19 +0000[diff] [blame]442
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]443<div class="question">
444<p><a name="langs">What source languages are supported?</a></p>
Reid Spencer501bfee2006-04-26 14:52:19 +0000[diff] [blame]445</div>
Gordon Henriksen58366822008-02-22 20:58:29 +0000[diff] [blame]446
Gordon Henriksen58366822008-02-22 20:58:29 +0000[diff] [blame]447<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]448<p>LLVM currently has full support for C and C++ source languages. These are
449 available through a special version of GCC that LLVM calls the
450 <a href="#cfe">C Front End</a></p>
451
452<p>There is an incomplete version of a Java front end available in the
453 <tt>java</tt> module. There is no documentation on this yet so you'll need to
454 download the code, compile it, and try it.</p>
455
456<p>The PyPy developers are working on integrating LLVM into the PyPy backend so
457 that PyPy language can translate to LLVM.</p>
Gordon Henriksen58366822008-02-22 20:58:29 +0000[diff] [blame]458</div>
459
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]460<div class="question">
461<p><a name="langirgen">I'd like to write a self-hosting LLVM compiler. How
462 should I interface with the LLVM middle-end optimizers and back-end code
463 generators?</a></p>
Reid Spencer501bfee2006-04-26 14:52:19 +0000[diff] [blame]464</div>
Chris Lattner33bef482006-08-15 00:43:35 +0000[diff] [blame]465
Chris Lattner33bef482006-08-15 00:43:35 +0000[diff] [blame]466<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]467<p>Your compiler front-end will communicate with LLVM by creating a module in
468 the LLVM intermediate representation (IR) format. Assuming you want to write
469 your language's compiler in the language itself (rather than C++), there are
470 3 major ways to tackle generating LLVM IR from a front-end:</p>
471
472<ul>
473 <li><strong>Call into the LLVM libraries code using your language's FFI
474 (foreign function interface).</strong>
475
476 <ul>
477 <li><em>for:</em> best tracks changes to the LLVM IR, .ll syntax, and .bc
478 format</li>
479
480 <li><em>for:</em> enables running LLVM optimization passes without a
481 emit/parse overhead</li>
482
483 <li><em>for:</em> adapts well to a JIT context</li>
484
485 <li><em>against:</em> lots of ugly glue code to write</li>
486 </ul></li>
487
488 <li> <strong>Emit LLVM assembly from your compiler's native language.</strong>
489 <ul>
490 <li><em>for:</em> very straightforward to get started</li>
491
492 <li><em>against:</em> the .ll parser is slower than the bitcode reader
493 when interfacing to the middle end</li>
494
495 <li><em>against:</em> you'll have to re-engineer the LLVM IR object model
496 and asm writer in your language</li>
497
498 <li><em>against:</em> it may be harder to track changes to the IR</li>
499 </ul></li>
500
501 <li><strong>Emit LLVM bitcode from your compiler's native language.</strong>
502
503 <ul>
504 <li><em>for:</em> can use the more-efficient bitcode reader when
505 interfacing to the middle end</li>
506
507 <li><em>against:</em> you'll have to re-engineer the LLVM IR object
508 model and bitcode writer in your language</li>
509
510 <li><em>against:</em> it may be harder to track changes to the IR</li>
511 </ul></li>
512</ul>
513
514<p>If you go with the first option, the C bindings in include/llvm-c should help
515 a lot, since most languages have strong support for interfacing with C. The
516 most common hurdle with calling C from managed code is interfacing with the
517 garbage collector. The C interface was designed to require very little memory
518 management, and so is straightforward in this regard.</p>
519</div>
520
521<div class="question">
522<p><a name="langhlsupp">What support is there for a higher level source language
523 constructs for building a compiler?</a></p>
524</div>
525
526<div class="answer">
527<p>Currently, there isn't much. LLVM supports an intermediate representation
528 which is useful for code representation but will not support the high level
529 (abstract syntax tree) representation needed by most compilers. There are no
Eric Christophereae5a132011-09-20 00:34:27 +0000[diff] [blame]530 facilities for lexical nor semantic analysis.</p>
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]531</div>
532
533<div class="question">
534<p><a name="getelementptr">I don't understand the GetElementPtr
535 instruction. Help!</a></p>
536</div>
537
538<div class="answer">
539<p>See <a href="GetElementPtr.html">The Often Misunderstood GEP
Bill Wendlinge9a6c352007-09-22 09:54:47 +0000[diff] [blame]540 Instruction</a>.</p>
Chris Lattner33bef482006-08-15 00:43:35 +0000[diff] [blame]541</div>
542
Reid Spencer501bfee2006-04-26 14:52:19 +0000[diff] [blame]543<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]544<h2>
Chris Lattnercc33d702003-11-19 05:53:12 +0000[diff] [blame]545 <a name="cfe">Using the GCC Front End</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]546</h2>
John Criswell6ea30b02003-11-18 16:05:23 +0000[diff] [blame]547
548<div class="question">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]549<p>When I compile software that uses a configure script, the configure script
550 thinks my system has all of the header files and libraries it is testing for.
551 How do I get configure to work correctly?</p>
John Criswell6ea30b02003-11-18 16:05:23 +0000[diff] [blame]552</div>
553
554<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]555<p>The configure script is getting things wrong because the LLVM linker allows
556 symbols to be undefined at link time (so that they can be resolved during JIT
557 or translation to the C back end). That is why configure thinks your system
558 "has everything."</p>
559
560<p>To work around this, perform the following steps:</p>
561
John Criswell6ea30b02003-11-18 16:05:23 +0000[diff] [blame]562<ol>
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]563 <li>Make sure the CC and CXX environment variables contains the full path to
564 the LLVM GCC front end.</li>
John Criswell6ea30b02003-11-18 16:05:23 +0000[diff] [blame]565
Reid Spencer434262a2007-02-09 15:59:08 +0000[diff] [blame]566 <li>Make sure that the regular C compiler is first in your PATH. </li>
John Criswell6ea30b02003-11-18 16:05:23 +0000[diff] [blame]567
Reid Spencer434262a2007-02-09 15:59:08 +0000[diff] [blame]568 <li>Add the string "-Wl,-native" to your CFLAGS environment variable.</li>
John Criswell6ea30b02003-11-18 16:05:23 +0000[diff] [blame]569</ol>
570
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]571<p>This will allow the <tt>llvm-ld</tt> linker to create a native code
572 executable instead of shell script that runs the JIT. Creating native code
573 requires standard linkage, which in turn will allow the configure script to
574 find out if code is not linking on your system because the feature isn't
575 available on your system.</p>
John Criswell6ea30b02003-11-18 16:05:23 +0000[diff] [blame]576</div>
577
Chris Lattnercc33d702003-11-19 05:53:12 +0000[diff] [blame]578<div class="question">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]579<p>When I compile code using the LLVM GCC front end, it complains that it cannot
580 find libcrtend.a.
John Criswell6ea30b02003-11-18 16:05:23 +0000[diff] [blame]581</p>
582</div>
583
Chris Lattnercc33d702003-11-19 05:53:12 +0000[diff] [blame]584<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]585<p>The only way this can happen is if you haven't installed the runtime
586 library. To correct this, do:</p>
Bill Wendlingd6a68eb2007-05-29 09:24:33 +0000[diff] [blame]587
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]588<pre class="doc_code">
Bill Wendlingd6a68eb2007-05-29 09:24:33 +0000[diff] [blame]589% cd llvm/runtime
590% make clean ; make install-bytecode
Reid Spencerf96eb572004-12-15 00:14:01 +0000[diff] [blame]591</pre>
John Criswell6ea30b02003-11-18 16:05:23 +0000[diff] [blame]592</div>
Chris Lattnercc33d702003-11-19 05:53:12 +0000[diff] [blame]593
Tanya Lattner14fc5c12005-04-25 20:36:56 +0000[diff] [blame]594<div class="question">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]595<p>How can I disable all optimizations when compiling code using the LLVM GCC
596 front end?</p>
Tanya Lattner14fc5c12005-04-25 20:36:56 +0000[diff] [blame]597</div>
598
599<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]600<p>Passing "-Wa,-disable-opt -Wl,-disable-opt" will disable *all* cleanup and
601 optimizations done at the llvm level, leaving you with the truly horrible
602 code that you desire.</p>
Tanya Lattner14fc5c12005-04-25 20:36:56 +0000[diff] [blame]603</div>
Chris Lattnercc33d702003-11-19 05:53:12 +0000[diff] [blame]604
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]605
606<div class="question">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]607<p><a name="translatecxx">Can I use LLVM to convert C++ code to C code?</a></p>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]608</div>
609
610<div class="answer">
611<p>Yes, you can use LLVM to convert code from any language LLVM supports to C.
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]612 Note that the generated C code will be very low level (all loops are lowered
613 to gotos, etc) and not very pretty (comments are stripped, original source
614 formatting is totally lost, variables are renamed, expressions are
615 regrouped), so this may not be what you're looking for. Also, there are
616 several limitations noted below.<p>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]617
618<p>Use commands like this:</p>
619
620<ol>
Chris Lattner860e0b42010-05-04 18:15:33 +0000[diff] [blame]621 <li><p>Compile your program with llvm-g++:</p>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]622
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]623<pre class="doc_code">
Chris Lattner860e0b42010-05-04 18:15:33 +0000[diff] [blame]624% llvm-g++ -emit-llvm x.cpp -o program.bc -c
Bill Wendlingd6a68eb2007-05-29 09:24:33 +0000[diff] [blame]625</pre>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]626
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]627 <p>or:</p>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]628
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]629<pre class="doc_code">
Chris Lattner860e0b42010-05-04 18:15:33 +0000[diff] [blame]630% llvm-g++ a.cpp -c -emit-llvm
631% llvm-g++ b.cpp -c -emit-llvm
632% llvm-ld a.o b.o -o program
Bill Wendlingd6a68eb2007-05-29 09:24:33 +0000[diff] [blame]633</pre>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]634
Chris Lattner860e0b42010-05-04 18:15:33 +0000[diff] [blame]635 <p>This will generate program and program.bc. The .bc
636 file is the LLVM version of the program all linked together.</p></li>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]637
Bill Wendlinge9a6c352007-09-22 09:54:47 +0000[diff] [blame]638 <li><p>Convert the LLVM code to C code, using the LLC tool with the C
639 backend:</p>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]640
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]641<pre class="doc_code">
Bill Wendlingd6a68eb2007-05-29 09:24:33 +0000[diff] [blame]642% llc -march=c program.bc -o program.c
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]643</pre></li>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]644
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]645 <li><p>Finally, compile the C file:</p>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]646
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]647<pre class="doc_code">
Chris Lattnercd465012010-05-04 18:16:00 +0000[diff] [blame]648% cc x.c -lstdc++
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]649</pre></li>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]650
651</ol>
652
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]653<p>Using LLVM does not eliminate the need for C++ library support. If you use
654 the llvm-g++ front-end, the generated code will depend on g++'s C++ support
655 libraries in the same way that code generated from g++ would. If you use
656 another C++ front-end, the generated code will depend on whatever library
657 that front-end would normally require.</p>
Chris Lattnerb495fb02006-08-31 04:26:31 +0000[diff] [blame]658
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]659<p>If you are working on a platform that does not provide any C++ libraries, you
660 may be able to manually compile libstdc++ to LLVM bitcode, statically link it
661 into your program, then use the commands above to convert the whole result
662 into C code. Alternatively, you might compile the libraries and your
663 application into two different chunks of C code and link them.</p>
Chris Lattnerb495fb02006-08-31 04:26:31 +0000[diff] [blame]664
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]665<p>Note that, by default, the C back end does not support exception handling.
666 If you want/need it for a certain program, you can enable it by passing
667 "-enable-correct-eh-support" to the llc program. The resultant code will use
668 setjmp/longjmp to implement exception support that is relatively slow, and
669 not C++-ABI-conforming on most platforms, but otherwise correct.</p>
Dan Gohmand5b455f2009-01-25 16:04:50 +0000[diff] [blame]670
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]671<p>Also, there are a number of other limitations of the C backend that cause it
672 to produce code that does not fully conform to the C++ ABI on most
673 platforms. Some of the C++ programs in LLVM's test suite are known to fail
Benjamin Kramer8040cd32009-10-12 14:46:08 +0000[diff] [blame]674 when compiled with the C back end because of ABI incompatibilities with
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]675 standard C++ libraries.</p>
Chris Lattneraf7fd202006-07-19 18:19:59 +0000[diff] [blame]676</div>
677
Dan Gohmancfbcd592009-02-10 17:26:53 +0000[diff] [blame]678<div class="question">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]679<p><a name="platformindependent">Can I compile C or C++ code to
680 platform-independent LLVM bitcode?</a></p>
Dan Gohmancfbcd592009-02-10 17:26:53 +0000[diff] [blame]681</div>
682
683<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]684<p>No. C and C++ are inherently platform-dependent languages. The most obvious
685 example of this is the preprocessor. A very common way that C code is made
686 portable is by using the preprocessor to include platform-specific code. In
687 practice, information about other platforms is lost after preprocessing, so
688 the result is inherently dependent on the platform that the preprocessing was
Benjamin Kramer8040cd32009-10-12 14:46:08 +0000[diff] [blame]689 targeting.</p>
Dan Gohmancfbcd592009-02-10 17:26:53 +0000[diff] [blame]690
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]691<p>Another example is <tt>sizeof</tt>. It's common for <tt>sizeof(long)</tt> to
692 vary between platforms. In most C front-ends, <tt>sizeof</tt> is expanded to
Misha Brukman7e0fc8a2009-04-10 20:48:27 +0000[diff] [blame]693 a constant immediately, thus hard-wiring a platform-specific detail.</p>
Dan Gohmancfbcd592009-02-10 17:26:53 +0000[diff] [blame]694
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]695<p>Also, since many platforms define their ABIs in terms of C, and since LLVM is
696 lower-level than C, front-ends currently must emit platform-specific IR in
697 order to have the result conform to the platform ABI.</p>
Dan Gohmancfbcd592009-02-10 17:26:53 +0000[diff] [blame]698</div>
699
Chris Lattnercc33d702003-11-19 05:53:12 +0000[diff] [blame]700<!-- *********************************************************************** -->
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]701<h2>
Chris Lattnercc33d702003-11-19 05:53:12 +0000[diff] [blame]702 <a name="cfe_code">Questions about code generated by the GCC front-end</a>
NAKAMURA Takumi05d02652011-04-18 23:59:50 +0000[diff] [blame]703</h2>
Chris Lattnercc33d702003-11-19 05:53:12 +0000[diff] [blame]704
Misha Brukman237dc2a2004-12-03 23:58:18 +0000[diff] [blame]705<div class="question">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]706<p><a name="iosinit">What is this <tt>llvm.global_ctors</tt> and
707 <tt>_GLOBAL__I__tmp_webcompile...</tt> stuff that happens when I <tt>#include
708 &lt;iostream&gt;</tt>?</a></p>
Misha Brukman237dc2a2004-12-03 23:58:18 +0000[diff] [blame]709</div>
Chris Lattnerc50bbc92004-03-29 19:14:35 +0000[diff] [blame]710
711<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]712<p>If you <tt>#include</tt> the <tt>&lt;iostream&gt;</tt> header into a C++
713 translation unit, the file will probably use
714 the <tt>std::cin</tt>/<tt>std::cout</tt>/... global objects. However, C++
715 does not guarantee an order of initialization between static objects in
716 different translation units, so if a static ctor/dtor in your .cpp file
717 used <tt>std::cout</tt>, for example, the object would not necessarily be
718 automatically initialized before your use.</p>
Chris Lattnerc50bbc92004-03-29 19:14:35 +0000[diff] [blame]719
Misha Brukman237dc2a2004-12-03 23:58:18 +0000[diff] [blame]720<p>To make <tt>std::cout</tt> and friends work correctly in these scenarios, the
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]721 STL that we use declares a static object that gets created in every
722 translation unit that includes <tt>&lt;iostream&gt;</tt>. This object has a
723 static constructor and destructor that initializes and destroys the global
724 iostream objects before they could possibly be used in the file. The code
725 that you see in the .ll file corresponds to the constructor and destructor
726 registration code.
Chris Lattnerc50bbc92004-03-29 19:14:35 +0000[diff] [blame]727</p>
728
Misha Brukman237dc2a2004-12-03 23:58:18 +0000[diff] [blame]729<p>If you would like to make it easier to <b>understand</b> the LLVM code
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]730 generated by the compiler in the demo page, consider using <tt>printf()</tt>
731 instead of <tt>iostream</tt>s to print values.</p>
Chris Lattnerc50bbc92004-03-29 19:14:35 +0000[diff] [blame]732</div>
733
Chris Lattner5a53c5d2005-02-25 20:30:21 +0000[diff] [blame]734<!--=========================================================================-->
735
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]736<div class="question">
737<p><a name="codedce">Where did all of my code go??</a></p>
738</div>
Chris Lattner5a53c5d2005-02-25 20:30:21 +0000[diff] [blame]739
740<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]741<p>If you are using the LLVM demo page, you may often wonder what happened to
742 all of the code that you typed in. Remember that the demo script is running
743 the code through the LLVM optimizers, so if your code doesn't actually do
744 anything useful, it might all be deleted.</p>
Chris Lattner5a53c5d2005-02-25 20:30:21 +0000[diff] [blame]745
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]746<p>To prevent this, make sure that the code is actually needed. For example, if
747 you are computing some expression, return the value from the function instead
748 of leaving it in a local variable. If you really want to constrain the
749 optimizer, you can read from and assign to <tt>volatile</tt> global
750 variables.</p>
Chris Lattner5a53c5d2005-02-25 20:30:21 +0000[diff] [blame]751</div>
752
753<!--=========================================================================-->
754
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]755<div class="question">
756<p><a name="undef">What is this "<tt>undef</tt>" thing that shows up in my
Bill Wendlingb1a61bd2009-04-07 18:52:30 +0000[diff] [blame]757 code?</a></p>
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]758</div>
Chris Lattner5a53c5d2005-02-25 20:30:21 +0000[diff] [blame]759
760<div class="answer">
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]761<p><a href="LangRef.html#undef"><tt>undef</tt></a> is the LLVM way of
762 representing a value that is not defined. You can get these if you do not
763 initialize a variable before you use it. For example, the C function:</p>
Chris Lattner5a53c5d2005-02-25 20:30:21 +0000[diff] [blame]764
Misha Brukmana54d4b22008-12-17 18:11:40 +0000[diff] [blame]765<pre class="doc_code">
Bill Wendlingd6a68eb2007-05-29 09:24:33 +0000[diff] [blame]766int X() { int i; return i; }
767</pre>
Chris Lattner5a53c5d2005-02-25 20:30:21 +0000[diff] [blame]768
Bill Wendling290235f2009-04-07 18:40:56 +0000[diff] [blame]769<p>Is compiled to "<tt>ret i32 undef</tt>" because "<tt>i</tt>" never has a
770 value specified for it.</p>
Chris Lattner5a53c5d2005-02-25 20:30:21 +0000[diff] [blame]771</div>
772
Chris Lattner2c6f9f72009-06-30 17:10:19 +0000[diff] [blame]773<!--=========================================================================-->
774
775<div class="question">
776<p><a name="callconvwrong">Why does instcombine + simplifycfg turn
777 a call to a function with a mismatched calling convention into "unreachable"?
778 Why not make the verifier reject it?</a></p>
779</div>
780
781<div class="answer">
782<p>This is a common problem run into by authors of front-ends that are using
783custom calling conventions: you need to make sure to set the right calling
784convention on both the function and on each call to the function. For example,
785this code:</p>
786
787<pre class="doc_code">
788define fastcc void @foo() {
789 ret void
790}
791define void @bar() {
Dan Gohman3dfb3cf2010-05-28 17:07:41 +0000[diff] [blame]792 call void @foo()
Chris Lattner2c6f9f72009-06-30 17:10:19 +0000[diff] [blame]793 ret void
794}
795</pre>
796
797<p>Is optimized to:</p>
798
799<pre class="doc_code">
800define fastcc void @foo() {
801 ret void
802}
803define void @bar() {
804 unreachable
805}
806</pre>
807
808<p>... with "opt -instcombine -simplifycfg". This often bites people because
809"all their code disappears". Setting the calling convention on the caller and
810callee is required for indirect calls to work, so people often ask why not make
811the verifier reject this sort of thing.</p>
812
813<p>The answer is that this code has undefined behavior, but it is not illegal.
814If we made it illegal, then every transformation that could potentially create
815this would have to ensure that it doesn't, and there is valid code that can
816create this sort of construct (in dead code). The sorts of things that can
817cause this to happen are fairly contrived, but we still need to accept them.
818Here's an example:</p>
819
820<pre class="doc_code">
821define fastcc void @foo() {
822 ret void
823}
824define internal void @bar(void()* %FP, i1 %cond) {
825 br i1 %cond, label %T, label %F
826T:
827 call void %FP()
828 ret void
829F:
830 call fastcc void %FP()
831 ret void
832}
833define void @test() {
834 %X = or i1 false, false
835 call void @bar(void()* @foo, i1 %X)
836 ret void
837}
838</pre>
839
840<p>In this example, "test" always passes @foo/false into bar, which ensures that
841 it is dynamically called with the right calling conv (thus, the code is
842 perfectly well defined). If you run this through the inliner, you get this
843 (the explicit "or" is there so that the inliner doesn't dead code eliminate
844 a bunch of stuff):
845</p>
846
847<pre class="doc_code">
848define fastcc void @foo() {
849 ret void
850}
851define void @test() {
852 %X = or i1 false, false
853 br i1 %X, label %T.i, label %F.i
854T.i:
855 call void @foo()
856 br label %bar.exit
857F.i:
858 call fastcc void @foo()
859 br label %bar.exit
860bar.exit:
861 ret void
862}
863</pre>
864
865<p>Here you can see that the inlining pass made an undefined call to @foo with
866 the wrong calling convention. We really don't want to make the inliner have
867 to know about this sort of thing, so it needs to be valid code. In this case,
868 dead code elimination can trivially remove the undefined code. However, if %X
869 was an input argument to @test, the inliner would produce this:
870</p>
871
872<pre class="doc_code">
873define fastcc void @foo() {
874 ret void
875}
876
877define void @test(i1 %X) {
878 br i1 %X, label %T.i, label %F.i
879T.i:
880 call void @foo()
881 br label %bar.exit
882F.i:
883 call fastcc void @foo()
884 br label %bar.exit
885bar.exit:
886 ret void
887}
888</pre>
889
890<p>The interesting thing about this is that %X <em>must</em> be false for the
891code to be well-defined, but no amount of dead code elimination will be able to
892delete the broken call as unreachable. However, since instcombine/simplifycfg
893turns the undefined call into unreachable, we end up with a branch on a
894condition that goes to unreachable: a branch to unreachable can never happen, so
895"-inline -instcombine -simplifycfg" is able to produce:</p>
896
897<pre class="doc_code">
898define fastcc void @foo() {
899 ret void
900}
901define void @test(i1 %X) {
902F.i:
903 call fastcc void @foo()
904 ret void
905}
906</pre>
907
908</div>
909
John Criswell6ea30b02003-11-18 16:05:23 +0000[diff] [blame]910<!-- *********************************************************************** -->
John Criswellc310f622003-10-13 16:13:06 +0000[diff] [blame]911
912<hr>
Misha Brukman7ce62cc2004-06-01 18:51:03 +0000[diff] [blame]913<address>
914 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]915 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukman7ce62cc2004-06-01 18:51:03 +0000[diff] [blame]916 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]917 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Misha Brukman7ce62cc2004-06-01 18:51:03 +0000[diff] [blame]918
NAKAMURA Takumib9a33632011-04-09 02:13:37 +0000[diff] [blame]919 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
Misha Brukmana6538852003-11-06 21:55:44 +0000[diff] [blame]920 Last modified: $Date$
Misha Brukman7ce62cc2004-06-01 18:51:03 +0000[diff] [blame]921</address>
John Criswellf08c5d82003-10-24 22:48:20 +0000[diff] [blame]922
John Criswellc310f622003-10-13 16:13:06 +0000[diff] [blame]923</body>
924</html>