Gitiles
Code Review Sign In
gerrit-public.fairphone.software / toolchain / llvm-project / 80f843d31d6bf058c69f2f369b3cbf1b38fa1ca4 / . / llvm / docs / HowToSubmitABug.html
blob: f01c5c72a10064052668051dbb11722a764f5589 [file] [log] [blame]
Misha Brukmanff919402003-10-31 18:37:20 +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>
5 <title>How to submit an LLVM bug report</title>
6 <link rel="stylesheet" href="llvm.css" type="text/css">
7</head>
8<body>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]9
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]10<div class="doc_title">
11 How to submit an LLVM bug report
12</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]13
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]14<table class="layout" style="width: 90%" >
15<tr class="layout">
16 <td class="left">
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]17<ol>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]18 <li><a href="#introduction">Introduction - Got bugs?</a></li>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]19 <li><a href="#crashers">Crashing Bugs</a>
20 <ul>
21 <li><a href="#front-end">Front-end bugs</a>
22 <li><a href="#gccas">GCCAS bugs</a>
23 <li><a href="#gccld">GCCLD bugs</a>
24 <li><a href="#passes">Bugs in LLVM passes</a>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]25 </ul></li>
26 <li><a href="#miscompilations">Miscompilations</a></li>
27 <li><a href="#codegen">Incorrect code generation (JIT and LLC)</a></li>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]28</ol>
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]29<div class="doc_author">
30 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a> and
31 <a href="http://misha.brukman.net">Misha Brukman</a></p>
Reid Spencer4d212422004-11-01 08:24:13 +0000[diff] [blame]32</div>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]33</td>
Reid Spencerc3c4c4f2004-11-01 08:19:36 +0000[diff] [blame]34<td class="right">
Misha Brukman3c1f8e72004-05-12 21:26:16 +0000[diff] [blame]35 <img src="img/Debugging.gif" alt="Debugging" width="444" height="314">
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]36</td>
37</tr>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]38</table>
39
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]40<!-- *********************************************************************** -->
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]41<div class="doc_section">
42 <a name="introduction">Introduction - Got bugs?</a>
43</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]44<!-- *********************************************************************** -->
45
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]46<div class="doc_text">
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]47
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]48<p>If you're working with LLVM and run into a bug, we definitely want to know
49about it. This document describes what you can do to increase the odds of
50getting it fixed quickly.</p>
51
52<p>Basically you have to do two things at a minimum. First, decide whether the
53bug <a href="#crashers">crashes the compiler</a> (or an LLVM pass), or if the
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]54compiler is <a href="#miscompilations">miscompiling</a> the program. Based on
55what type of bug it is, follow the instructions in the linked section to narrow
56down the bug so that the person who fixes it will be able to find the problem
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]57more easily.</p>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]58
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]59<p>Once you have a reduced test-case, go to <a
Reid Spencerca058542006-03-14 05:39:39 +0000[diff] [blame]60href="http://llvm.org/bugs/enter_bug.cgi">the LLVM Bug Tracking
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]61System</a>, select the category in which the bug falls, and fill out the form
62with the necessary details. The bug description should contain the following
63information:</p>
64
John Criswell56a850a2003-10-22 15:06:11 +0000[diff] [blame]65<ul>
Misha Brukmanb00e1062003-10-22 17:01:44 +0000[diff] [blame]66 <li>All information necessary to reproduce the problem.</li>
67 <li>The reduced test-case that triggers the bug.</li>
68 <li>The location where you obtained LLVM (if not from our CVS
69 repository).</li>
John Criswell56a850a2003-10-22 15:06:11 +0000[diff] [blame]70</ul>
71
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]72<p>Thanks for helping us make LLVM better!</p>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]73
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]74</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]75
76<!-- *********************************************************************** -->
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]77<div class="doc_section">
78 <a name="crashers">Crashing Bugs</a>
79</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]80<!-- *********************************************************************** -->
81
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]82<div class="doc_text">
83
84<p>More often than not, bugs in the compiler cause it to crash - often due to an
Reid Spencer80f843d2006-08-28 00:34:19 +0000[diff] [blame^]85assertion failure of some sort. If you are running <tt><b>opt</b></tt>
86directly, and something crashes, jump to the section on
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]87<a href="#passes">bugs in LLVM passes</a>. Otherwise, the most important
88piece of the puzzle is to figure out if it is the GCC-based front-end that is
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]89buggy or if it's one of the LLVM tools that has problems.</p>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]90
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]91<p>To figure out which program is crashing (the front-end,
92<tt><b>gccas</b></tt>, or <tt><b>gccld</b></tt>), run the
93<tt><b>llvm-gcc</b></tt> command line as you were when the crash occurred, but
94add a <tt>-v</tt> option to the command line. The compiler will print out a
John Criswellfaba4622004-03-12 20:42:16 +0000[diff] [blame]95bunch of stuff, and should end with telling you that one of
96<tt><b>cc1</b>/<b>cc1plus</b></tt>, <tt><b>gccas</b></tt>, or
97<tt><b>gccld</b></tt> crashed.</p>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]98
99<ul>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]100
John Criswellfaba4622004-03-12 20:42:16 +0000[diff] [blame]101 <li>If <tt><b>cc1</b></tt> or <tt><b>cc1plus</b></tt> crashed, you found a
102 problem with the front-end.
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]103 Jump ahead to the section on <a href="#front-end">front-end bugs</a>.</li>
104
105 <li>If <tt><b>gccas</b></tt> crashed, you found a bug in <a href="#gccas">one
106 of the passes in <tt><b>gccas</b></tt></a>.</li>
107
108 <li>If <tt><b>gccld</b></tt> crashed, you found a bug in <a href="#gccld">one
109 of the passes in <tt><b>gccld</b></tt></a>.</li>
110
111 <li>Otherwise, something really weird happened. Email the list with what you
112 have at this point.</li>
113
114</ul>
115
116</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]117
118<!-- ======================================================================= -->
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]119<div class="doc_subsection">
120 <a name="front-end">Front-end bugs</a>
121</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]122
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]123<div class="doc_text">
124
125<p>If the problem is in the front-end, you should re-run the same
Brian Gaeke627b9f42003-05-23 21:03:50 +0000[diff] [blame]126<tt>llvm-gcc</tt> command that resulted in the crash, but add the
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]127<tt>-save-temps</tt> option. The compiler will crash again, but it will leave
128behind a <tt><i>foo</i>.i</tt> file (containing preprocessed C source code) and
129possibly <tt><i>foo</i>.s</tt> (containing LLVM assembly code), for each
130compiled <tt><i>foo</i>.c</tt> file. Send us the <tt><i>foo</i>.i</tt> file,
Chris Lattner6f5a00f2006-02-08 17:01:37 +0000[diff] [blame]131along with a brief description of the error it caused. A tool that might help
132you reduce a front-end testcase to a more manageable size is
133<a href="http://delta.tigris.org/">delta</a>.
134</p>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]135
136</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]137
138<!-- ======================================================================= -->
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]139<div class="doc_subsection">
140 <a name="gccas">GCCAS bugs</a>
141</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]142
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]143<div class="doc_text">
144
145<p>If you find that a bug crashes in the <tt><b>gccas</b></tt> stage of
Brian Gaeke627b9f42003-05-23 21:03:50 +0000[diff] [blame]146compilation, compile your test-case to a <tt>.s</tt> file with the
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]147<tt>-save-temps</tt> option to <tt><b>llvm-gcc</b></tt>. Then run:</p>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]148
Misha Brukmanc21921c2004-04-15 20:49:32 +0000[diff] [blame]149<div class="doc_code">
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]150<p><tt><b>gccas</b> -debug-pass=Arguments &lt; /dev/null -o - &gt; /dev/null</tt></p>
Misha Brukmanc21921c2004-04-15 20:49:32 +0000[diff] [blame]151</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]152
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]153<p>... which will print a list of arguments, indicating the list of passes that
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]154<tt><b>gccas</b></tt> runs. Once you have the input file and the list of
155passes, go to the section on <a href="#passes">debugging bugs in LLVM
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]156passes</a>.</p>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]157
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]158</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]159
160<!-- ======================================================================= -->
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]161<div class="doc_subsection">
162 <a name="gccld">GCCLD bugs</a>
163</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]164
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]165<div class="doc_text">
166
167<p>If you find that a bug crashes in the <tt><b>gccld</b></tt> stage of
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]168compilation, gather all of the <tt>.o</tt> bytecode files and libraries that are
169being linked together (the "<tt><b>llvm-gcc</b> -v</tt>" output should include
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]170the full list of objects linked). Then run:</p>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]171
Misha Brukmanc21921c2004-04-15 20:49:32 +0000[diff] [blame]172<div class="doc_code">
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]173<p><tt><b>llvm-as</b> &lt; /dev/null &gt; null.bc<br>
174<b>gccld</b> -debug-pass=Arguments null.bc</tt>
175</p>
Misha Brukmanc21921c2004-04-15 20:49:32 +0000[diff] [blame]176</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]177
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]178<p>... which will print a list of arguments, indicating the list of passes that
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]179<tt><b>gccld</b></tt> runs. Once you have the input files and the list of
180passes, go to the section on <a href="#passes">debugging bugs in LLVM
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]181passes</a>.</p>
182
183</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]184
185<!-- ======================================================================= -->
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]186<div class="doc_subsection">
187 <a name="passes">Bugs in LLVM passes</a>
188</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]189
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]190<div class="doc_text">
191
192<p>At this point, you should have some number of LLVM assembly files or bytecode
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]193files and a list of passes which crash when run on the specified input. In
194order to reduce the list of passes (which is probably large) and the input to
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]195something tractable, use the <tt><b>bugpoint</b></tt> tool as follows:</p>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]196
Misha Brukmanc21921c2004-04-15 20:49:32 +0000[diff] [blame]197<div class="doc_code">
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]198<p><tt><b>bugpoint</b> &lt;input files&gt; &lt;list of passes&gt;</tt></p>
Misha Brukmanc21921c2004-04-15 20:49:32 +0000[diff] [blame]199</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]200
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]201<p><tt><b>bugpoint</b></tt> will print a bunch of output as it reduces the
202test-case, but it should eventually print something like this:</p>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]203
Misha Brukman4825e212004-04-19 03:07:29 +0000[diff] [blame]204<div class="doc_code">
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]205<p><tt>
206...<br>
207Emitted bytecode to 'bugpoint-reduced-simplified.bc'<br>
208<br>
209*** You can reproduce the problem with: opt bugpoint-reduced-simplified.bc -licm<br>
210</tt></p>
Misha Brukman4825e212004-04-19 03:07:29 +0000[diff] [blame]211</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]212
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]213<p>Once you complete this, please send the LLVM bytecode file and the command
214line to reproduce the problem to the llvmbugs mailing list.</p>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]215
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]216</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]217
218<!-- *********************************************************************** -->
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]219<div class="doc_section">
220 <a name="miscompilations">Miscompilations</a>
221</div>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]222<!-- *********************************************************************** -->
223
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]224<div class="doc_text">
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]225
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]226<p>A miscompilation occurs when a pass does not correctly transform a program,
227thus producing errors that are only noticed during execution. This is different
228from producing invalid LLVM code (i.e., code not in SSA form, using values
229before defining them, etc.) which the verifier will check for after a pass
230finishes its run.</p>
231
Chris Lattner41553242004-05-11 19:34:27 +0000[diff] [blame]232<p>If it looks like the LLVM compiler is miscompiling a program, the very first
Misha Brukmanb797a4f2004-05-12 00:57:24 +0000[diff] [blame]233thing to check is to make sure it is not using undefined behavior. In
234particular, check to see if the program <a
235href="http://valgrind.kde.org/">valgrind</a>s clean, passes purify, or some
Chris Lattner41553242004-05-11 19:34:27 +0000[diff] [blame]236other memory checker tool. Many of the "LLVM bugs" that we have chased down
237ended up being bugs in the program being compiled, not LLVM.</p>
238
239<p>Once you determine that the program itself is not buggy, you should choose
240which code generator you wish to compile the program with (e.g. C backend, the
241JIT, or LLC) and optionally a series of LLVM passes to run. For example:</p>
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]242
Misha Brukmanc21921c2004-04-15 20:49:32 +0000[diff] [blame]243<div class="doc_code">
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]244<p><tt>
245<b>bugpoint</b> -run-cbe [... optzn passes ...] file-to-test.bc --args -- [program arguments]</tt></p>
Misha Brukmanc21921c2004-04-15 20:49:32 +0000[diff] [blame]246</div>
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]247
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]248<p><tt>bugpoint</tt> will try to narrow down your list of passes to the one pass
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]249that causes an error, and simplify the bytecode file as much as it can to assist
250you. It will print a message letting you know how to reproduce the resulting
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]251error.</p>
252
253</div>
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]254
255<!-- *********************************************************************** -->
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]256<div class="doc_section">
257 <a name="codegen">Incorrect code generation</a>
258</div>
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]259<!-- *********************************************************************** -->
260
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]261<div class="doc_text">
262
263<p>Similarly to debugging incorrect compilation by mis-behaving passes, you can
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]264debug incorrect code generation by either LLC or the JIT, using
265<tt>bugpoint</tt>. The process <tt>bugpoint</tt> follows in this case is to try
266to narrow the code down to a function that is miscompiled by one or the other
267method, but since for correctness, the entire program must be run,
268<tt>bugpoint</tt> will compile the code it deems to not be affected with the C
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]269Backend, and then link in the shared object it generates.</p>
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]270
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]271<p>To debug the JIT:</p>
272
Misha Brukmanc21921c2004-04-15 20:49:32 +0000[diff] [blame]273<div class="doc_code">
Misha Brukmanc050f9f2004-05-14 18:57:24 +0000[diff] [blame]274<pre>
275bugpoint -run-jit -output=[correct output file] [bytecode file] \
276 --tool-args -- [arguments to pass to lli] \
277 --args -- [program arguments]
278</pre>
Misha Brukmanc21921c2004-04-15 20:49:32 +0000[diff] [blame]279</div>
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]280
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]281<p>Similarly, to debug the LLC, one would run:</p>
282
Misha Brukmanc21921c2004-04-15 20:49:32 +0000[diff] [blame]283<div class="doc_code">
Misha Brukmanc050f9f2004-05-14 18:57:24 +0000[diff] [blame]284<pre>
285bugpoint -run-llc -output=[correct output file] [bytecode file] \
286 --tool-args -- [arguments to pass to llc] \
287 --args -- [program arguments]
288</pre>
Misha Brukmanc21921c2004-04-15 20:49:32 +0000[diff] [blame]289</div>
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]290
Misha Brukman272322e2004-04-15 21:01:21 +0000[diff] [blame]291<p><b>Special note:</b> if you are debugging MultiSource or SPEC tests that
292already exist in the <tt>llvm/test</tt> hierarchy, there is an easier way to
293debug the JIT, LLC, and CBE, using the pre-written Makefile targets, which
294will pass the program options specified in the Makefiles:</p>
295
296<div class="doc_code">
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]297<p><tt>
298cd llvm/test/../../program<br>
Misha Brukman67abadb2004-05-14 19:04:07 +0000[diff] [blame]299make bugpoint-jit
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]300</tt></p>
Misha Brukman272322e2004-04-15 21:01:21 +0000[diff] [blame]301</div>
302
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]303<p>At the end of a successful <tt>bugpoint</tt> run, you will be presented
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]304with two bytecode files: a <em>safe</em> file which can be compiled with the C
305backend and the <em>test</em> file which either LLC or the JIT
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]306mis-codegenerates, and thus causes the error.</p>
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]307
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]308<p>To reproduce the error that <tt>bugpoint</tt> found, it is sufficient to do
309the following:</p>
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]310
311<ol>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]312
Misha Brukman67abadb2004-05-14 19:04:07 +0000[diff] [blame]313<li><p>Regenerate the shared object from the safe bytecode file:</p>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]314
Misha Brukman67abadb2004-05-14 19:04:07 +0000[diff] [blame]315<div class="doc_code">
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]316<p><tt>
317<b>llc</b> -march=c safe.bc -o safe.c<br>
Misha Brukman67abadb2004-05-14 19:04:07 +0000[diff] [blame]318<b>gcc</b> -shared safe.c -o safe.so
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]319</tt></p>
Misha Brukman67abadb2004-05-14 19:04:07 +0000[diff] [blame]320</div></li>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]321
Misha Brukman67abadb2004-05-14 19:04:07 +0000[diff] [blame]322<li><p>If debugging LLC, compile test bytecode native and link with the shared
323 object:</p>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]324
Misha Brukman67abadb2004-05-14 19:04:07 +0000[diff] [blame]325<div class="doc_code">
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]326<p><tt>
327<b>llc</b> test.bc -o test.s -f<br>
328<b>gcc</b> test.s safe.so -o test.llc<br>
Misha Brukman67abadb2004-05-14 19:04:07 +0000[diff] [blame]329./test.llc [program options]
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]330</tt></p>
Misha Brukman67abadb2004-05-14 19:04:07 +0000[diff] [blame]331</div></li>
332
333<li><p>If debugging the JIT, load the shared object and supply the test
334 bytecode:</p>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]335
Misha Brukman67abadb2004-05-14 19:04:07 +0000[diff] [blame]336<div class="doc_code">
Chris Lattner411eba12004-05-23 21:05:39 +0000[diff] [blame]337<p><tt><b>lli</b> -load=safe.so test.bc [program options]</tt></p>
Misha Brukman67abadb2004-05-14 19:04:07 +0000[diff] [blame]338</div></li>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]339
Misha Brukmana3fc5d92003-09-17 18:51:47 +0000[diff] [blame]340</ol>
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]341
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]342</div>
343
Chris Lattner9525a332003-05-21 22:21:07 +0000[diff] [blame]344<!-- *********************************************************************** -->
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]345<hr>
Misha Brukman67db86a2004-01-15 19:03:47 +0000[diff] [blame]346<address>
347 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
348 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
349 <a href="http://validator.w3.org/check/referer"><img
350 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
351
352 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencerca058542006-03-14 05:39:39 +0000[diff] [blame]353 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]354 <br>
355 Last modified: $Date$
Misha Brukman67db86a2004-01-15 19:03:47 +0000[diff] [blame]356</address>
Misha Brukmanff919402003-10-31 18:37:20 +0000[diff] [blame]357
358</body>
359</html>