Gitiles
Code Review Sign In
gerrit-public.fairphone.software / platform / external / llvm / 86f42bdad93677fa0ca33b27afb0f493028376cb / . / docs / SourceLevelDebugging.html
blob: 70677a71e71fe65a20b33ef36ce4c2f3a3199007 [file] [log] [blame]
Chris Lattnerbdfb3392004-01-05 05:06:33 +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>Source Level Debugging with LLVM</title>
6 <link rel="stylesheet" href="llvm.css" type="text/css">
7</head>
8<body>
9
10<div class="doc_title">Source Level Debugging with LLVM</div>
11
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]12<table border="0" width="100%">
13<tr>
14<td valign="top">
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]15<ul>
16
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]17 <li><a href="#introduction">Introduction</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]18 <ol>
19 <li><a href="#phil">Philosophy behind LLVM debugging information</a></li>
20 <li><a href="#debugopt">Debugging optimized code</a></li>
21 <li><a href="#future">Future work</a></li>
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]22 </ol></li>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]23 <li><a href="#llvm-db">Using the <tt>llvm-db</tt> tool</a>
24 <ol>
25 <li><a href="#limitations">Limitations of <tt>llvm-db</tt></a></li>
26 <li><a href="#sample">A sample <tt>llvm-db</tt> session</a></li>
27 <li><a href="#startup">Starting the debugger</a></li>
28 <li><a href="#commands">Commands recognized by the debugger</a></li>
29 </ol></li>
30
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]31 <li><a href="#architecture">Architecture of the LLVM debugger</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]32 <ol>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]33 <li><a href="#arch_debugger">The Debugger and InferiorProcess classes</a></li>
34 <li><a href="#arch_info">The RuntimeInfo, ProgramInfo, and SourceLanguage classes</a></li>
35 <li><a href="#arch_llvm-db">The <tt>llvm-db</tt> tool</a></li>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]36 <li><a href="#arch_todo">Short-term TODO list</a></li>
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]37 </ol></li>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]38
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]39 <li><a href="#format">Debugging information format</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]40 <ol>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]41 <li><a href="#format_common_anchors">Anchors for global objects</a></li>
42 <li><a href="#format_common_stoppoint">Representing stopping points in the source program</a></li>
43 <li><a href="#format_common_lifetime">Object lifetimes and scoping</a></li>
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]44 <li><a href="#format_common_descriptors">Object descriptor formats</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]45 <ul>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]46 <li><a href="#format_common_source_files">Representation of source files</a></li>
47 <li><a href="#format_common_program_objects">Representation of program objects</a></li>
48 <li><a href="#format_common_object_contexts">Program object contexts</a></li>
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]49 </ul></li>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]50 <li><a href="#format_common_intrinsics">Debugger intrinsic functions</a></li>
51 <li><a href="#format_common_tags">Values for debugger tags</a></li>
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]52 </ol></li>
53 <li><a href="#ccxx_frontend">C/C++ front-end specific debug information</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]54 <ol>
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]55 <li><a href="#ccxx_pse">Program Scope Entries</a>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]56 <ul>
57 <li><a href="#ccxx_compilation_units">Compilation unit entries</a></li>
58 <li><a href="#ccxx_modules">Module, namespace, and importing entries</a></li>
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]59 </ul></li>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]60 <li><a href="#ccxx_dataobjects">Data objects (program variables)</a></li>
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]61 </ol></li>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]62</ul>
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]63</td>
64<td align="right" valign="top">
Misha Brukmane849a1a2004-05-12 21:26:16 +0000[diff] [blame]65<img src="img/venusflytrap.jpg" alt="A leafy and green bug eater" width="247"
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]66height="369">
67</td>
68</tr>
69</table>
70
Chris Lattner7911ce22004-05-23 21:07:27 +0000[diff] [blame]71<div class="doc_author">
72 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a></p>
73</div>
74
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]75
76<!-- *********************************************************************** -->
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]77<div class="doc_section"><a name="introduction">Introduction</a></div> <!--
78*********************************************************************** -->
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]79
80<div class="doc_text">
81
82<p>This document is the central repository for all information pertaining to
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]83debug information in LLVM. It describes the <a href="#llvm-db">user
84interface</a> for the <a href="CommandGuide/llvm-db.html"><tt>llvm-db</tt>
85tool</a>, which provides a powerful <a href="#llvm-db">source-level debugger</a>
86to users of LLVM-based compilers. It then describes the <a
87href="#architecture">various components</a> that make up the debugger and the
88libraries which future clients may use. Finally, it describes the <a
89href="#format">actual format that the LLVM debug information</a> takes,
90which is useful for those interested in creating front-ends or dealing directly
91with the information.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]92
93</div>
94
95<!-- ======================================================================= -->
96<div class="doc_subsection">
97 <a name="phil">Philosophy behind LLVM debugging information</a>
98</div>
99
100<div class="doc_text">
101
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]102<p>The idea of the LLVM debugging information is to capture how the important
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]103pieces of the source-language's Abstract Syntax Tree map onto LLVM code.
104Several design aspects have shaped the solution that appears here. The
105important ones are:</p>
106
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]107<ul>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]108<li>Debugging information should have very little impact on the rest of the
109compiler. No transformations, analyses, or code generators should need to be
110modified because of debugging information.</li>
111
112<li>LLVM optimizations should interact in <a href="#debugopt">well-defined and
113easily described ways</a> with the debugging information.</li>
114
115<li>Because LLVM is designed to support arbitrary programming languages,
116LLVM-to-LLVM tools should not need to know anything about the semantics of the
117source-level-language.</li>
118
119<li>Source-level languages are often <b>widely</b> different from one another.
120LLVM should not put any restrictions of the flavor of the source-language, and
121the debugging information should work with any language.</li>
122
123<li>With code generator support, it should be possible to use an LLVM compiler
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]124to compile a program to native machine code and standard debugging formats.
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]125This allows compatibility with traditional machine-code level debuggers, like
126GDB or DBX.</li>
127
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]128</ul>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]129
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]130<p>The approach used by the LLVM implementation is to use a small set of <a
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]131href="#format_common_intrinsics">intrinsic functions</a> to define a mapping
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]132between LLVM program objects and the source-level objects. The description of
133the source-level program is maintained in LLVM global variables in an <a
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]134href="#ccxx_frontend">implementation-defined format</a> (the C/C++ front-end
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]135currently uses working draft 7 of the <a
136href="http://www.eagercon.com/dwarf/dwarf3std.htm">Dwarf 3 standard</a>).</p>
137
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]138<p>When a program is debugged, the debugger interacts with the user and turns
139the stored debug information into source-language specific information. As
140such, the debugger must be aware of the source-language, and is thus tied to a
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]141specific language of family of languages. The <a href="#llvm-db">LLVM
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]142debugger</a> is designed to be modular in its support for source-languages.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]143
144</div>
145
146
147<!-- ======================================================================= -->
148<div class="doc_subsection">
149 <a name="debugopt">Debugging optimized code</a>
150</div>
151
152<div class="doc_text">
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]153
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]154<p>An extremely high priority of LLVM debugging information is to make it
155interact well with optimizations and analysis. In particular, the LLVM debug
156information provides the following guarantees:</p>
157
158<ul>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]159
160<li>LLVM debug information <b>always provides information to accurately read the
161source-level state of the program</b>, regardless of which LLVM optimizations
162have been run, and without any modification to the optimizations themselves.
163However, some optimizations may impact the ability to modify the current state
164of the program with a debugger, such as setting program variables, or calling
165function that have been deleted.</li>
166
167<li>LLVM optimizations gracefully interact with debugging information. If they
168are not aware of debug information, they are automatically disabled as necessary
169in the cases that would invalidate the debug info. This retains the LLVM
170features making it easy to write new transformations.</li>
171
172<li>As desired, LLVM optimizations can be upgraded to be aware of the LLVM
173debugging information, allowing them to update the debugging information as they
174perform aggressive optimizations. This means that, with effort, the LLVM
175optimizers could optimize debug code just as well as non-debug code.</li>
176
177<li>LLVM debug information does not prevent many important optimizations from
178happening (for example inlining, basic block reordering/merging/cleanup, tail
179duplication, etc), further reducing the amount of the compiler that eventually
180is "aware" of debugging information.</li>
181
182<li>LLVM debug information is automatically optimized along with the rest of the
183program, using existing facilities. For example, duplicate information is
184automatically merged by the linker, and unused information is automatically
185removed.</li>
186
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]187</ul>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]188
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]189<p>Basically, the debug information allows you to compile a program with
190"<tt>-O0 -g</tt>" and get full debug information, allowing you to arbitrarily
191modify the program as it executes from the debugger. Compiling a program with
192"<tt>-O3 -g</tt>" gives you full debug information that is always available and
193accurate for reading (e.g., you get accurate stack traces despite tail call
194elimination and inlining), but you might lose the ability to modify the program
195and call functions where were optimized out of the program, or inlined away
196completely.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]197
198</div>
199
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]200<!-- ======================================================================= -->
201<div class="doc_subsection">
202 <a name="future">Future work</a>
203</div>
204
205<div class="doc_text">
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]206<p>There are several important extensions that could be eventually added to the
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]207LLVM debugger. The most important extension would be to upgrade the LLVM code
208generators to support debugging information. This would also allow, for
209example, the X86 code generator to emit native objects that contain debugging
210information consumable by traditional source-level debuggers like GDB or
211DBX.</p>
212
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]213<p>Additionally, LLVM optimizations can be upgraded to incrementally update the
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]214debugging information, <a href="#commands">new commands</a> can be added to the
215debugger, and thread support could be added to the debugger.</p>
216
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]217<p>The "SourceLanguage" modules provided by <tt>llvm-db</tt> could be
218substantially improved to provide good support for C++ language features like
219namespaces and scoping rules.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]220
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]221<p>After working with the debugger for a while, perhaps the nicest improvement
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]222would be to add some sort of line editor, such as GNU readline (but one that is
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]223compatible with the LLVM license).</p>
224
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]225<p>For someone so inclined, it should be straight-forward to write different
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]226front-ends for the LLVM debugger, as the LLVM debugging engine is cleanly
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]227separated from the <tt>llvm-db</tt> front-end. A new LLVM GUI debugger or IDE
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]228would be nice. :)</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]229
230</div>
231
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]232<!-- *********************************************************************** -->
233<div class="doc_section">
234 <a name="llvm-db">Using the <tt>llvm-db</tt> tool</a>
235</div>
236<!-- *********************************************************************** -->
237
238<div class="doc_text">
239
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]240<p>The <tt>llvm-db</tt> tool provides a GDB-like interface for source-level
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]241debugging of programs. This tool provides many standard commands for inspecting
242and modifying the program as it executes, loading new programs, single stepping,
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]243placing breakpoints, etc. This section describes how to use the debugger.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]244
245<p><tt>llvm-db</tt> has been designed to be as similar to GDB in its user
246interface as possible. This should make it extremely easy to learn
247<tt>llvm-db</tt> if you already know <tt>GDB</tt>. In general, <tt>llvm-db</tt>
248provides the subset of GDB commands that are applicable to LLVM debugging users.
249If there is a command missing that make a reasonable amount of sense within the
250<a href="#limitations">limitations of <tt>llvm-db</tt></a>, please report it as
251a bug or, better yet, submit a patch to add it. :)</p>
252
253</div>
254
255<!-- ======================================================================= -->
256<div class="doc_subsection">
257 <a name="limitations">Limitations of <tt>llvm-db</tt></a>
258</div>
259
260<div class="doc_text">
261
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]262<p><tt>llvm-db</tt> is designed to be modular and easy to extend. This
263extensibility was key to getting the debugger up-and-running quickly, because we
264can start with simple-but-unsophisicated implementations of various components.
265Because of this, it is currently missing many features, though they should be
266easy to add over time (patches welcomed!). The biggest inherent limitations of
267<tt>llvm-db</tt> are currently due to extremely simple <a
268href="#arch_debugger">debugger backend</a> (implemented in
269"lib/Debugger/UnixLocalInferiorProcess.cpp") which is designed to work without
270any cooperation from the code generators. Because it is so simple, it suffers
271from the following inherent limitations:</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]272
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]273<ul>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]274
275<li>Running a program in <tt>llvm-db</tt> is a bit slower than running it with
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]276<tt>lli</tt> (i.e., in the JIT).</li>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]277
278<li>Inspection of the target hardware is not supported. This means that you
279cannot, for example, print the contents of X86 registers.</li>
280
281<li>Inspection of LLVM code is not supported. This means that you cannot print
282the contents of arbitrary LLVM values, or use commands such as <tt>stepi</tt>.
283This also means that you cannot debug code without debug information.</li>
284
285<li>Portions of the debugger run in the same address space as the program being
286debugged. This means that memory corruption by the program could trample on
287portions of the debugger.</li>
288
289<li>Attaching to existing processes and core files is not currently
290supported.</li>
291
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]292</ul>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]293
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]294<p>That said, the debugger is still quite useful, and all of these limitations
295can be eliminated by integrating support for the debugger into the code
296generators, and writing a new <a href="#arch_debugger">InferiorProcess</a>
297subclass to use it. See the <a href="#future">future work</a> section for ideas
298of how to extend the LLVM debugger despite these limitations.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]299
300</div>
301
302
303<!-- ======================================================================= -->
304<div class="doc_subsection">
305 <a name="sample">A sample <tt>llvm-db</tt> session</a>
306</div>
307
308<div class="doc_text">
309
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]310<p>TODO: this is obviously lame, when more is implemented, this can be much
311better.</p>
312
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]313<pre>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]314$ <b>llvm-db funccall</b>
315llvm-db: The LLVM source-level debugger
316Loading program... successfully loaded 'funccall.bc'!
317(llvm-db) <b>create</b>
318Starting program: funccall.bc
319main at funccall.c:9:2
3209 -> q = 0;
321(llvm-db) <b>list main</b>
3224 void foo() {
3235 int t = q;
3246 q = t + 1;
3257 }
3268 int main() {
3279 -> q = 0;
32810 foo();
32911 q = q - 1;
33012
33113 return q;
332(llvm-db) <b>list</b>
33314 }
334(llvm-db) <b>step</b>
33510 -> foo();
336(llvm-db) <b>s</b>
337foo at funccall.c:5:2
3385 -> int t = q;
339(llvm-db) <b>bt</b>
340#0 -> 0x85ffba0 in foo at funccall.c:5:2
341#1 0x85ffd98 in main at funccall.c:10:2
342(llvm-db) <b>finish</b>
343main at funccall.c:11:2
34411 -> q = q - 1;
345(llvm-db) <b>s</b>
34613 -> return q;
347(llvm-db) <b>s</b>
348The program stopped with exit code 0
349(llvm-db) <b>quit</b>
350$
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]351</pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]352
353</div>
354
355
356
357<!-- ======================================================================= -->
358<div class="doc_subsection">
359 <a name="startup">Starting the debugger</a>
360</div>
361
362<div class="doc_text">
363
364<p>There are three ways to start up the <tt>llvm-db</tt> debugger:</p>
365
366<p>When run with no options, just <tt>llvm-db</tt>, the debugger starts up
367without a program loaded at all. You must use the <a
368href="#c_file"><tt>file</tt> command</a> to load a program, and the <a
Misha Brukman1ed83482004-06-03 23:33:10 +0000[diff] [blame]369href="#c_set_args"><tt>set args</tt></a> or <a href="#c_run"><tt>run</tt></a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]370commands to specify the arguments for the program.</p>
371
372<p>If you start the debugger with one argument, as <tt>llvm-db
373&lt;program&gt;</tt>, the debugger will start up and load in the specified
374program. You can then optionally specify arguments to the program with the <a
Misha Brukman1ed83482004-06-03 23:33:10 +0000[diff] [blame]375href="#c_set_args"><tt>set args</tt></a> or <a href="#c_run"><tt>run</tt></a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]376commands.</p>
377
378<p>The third way to start the program is with the <tt>--args</tt> option. This
379option allows you to specify the program to load and the arguments to start out
380with. <!-- No options to <tt>llvm-db</tt> may be specified after the
381<tt>-args</tt> option. --> Example use: <tt>llvm-db --args ls /home</tt></p>
382
383</div>
384
385<!-- ======================================================================= -->
386<div class="doc_subsection">
387 <a name="commands">Commands recognized by the debugger</a>
388</div>
389
390<div class="doc_text">
391
392<p>FIXME: this needs work obviously. See the <a
393href="http://sources.redhat.com/gdb/documentation/">GDB documentation</a> for
394information about what these do, or try '<tt>help [command]</tt>' within
395<tt>llvm-db</tt> to get information.</p>
396
397<p>
398<h2>General usage:</h2>
399<ul>
400<li>help [command]</li>
401<li>quit</li>
402<li><a name="c_file">file</a> [program]</li>
403</ul>
404
405<h2>Program inspection and interaction:</h2>
406<ul>
407<li>create (start the program, stopping it ASAP in <tt>main</tt>)</li>
408<li>kill</li>
409<li>run [args]</li>
410<li>step [num]</li>
411<li>next [num]</li>
412<li>cont</li>
413<li>finish</li>
414
415<li>list [start[, end]]</li>
416<li>info source</li>
417<li>info sources</li>
418<li>info functions</li>
419</ul>
420
421<h2>Call stack inspection:</h2>
422<ul>
423<li>backtrace</li>
424<li>up [n]</li>
425<li>down [n]</li>
426<li>frame [n]</li>
427</ul>
428
429
430<h2>Debugger inspection and interaction:</h2>
431<ul>
432<li>info target</li>
433<li>show prompt</li>
434<li>set prompt</li>
435<li>show listsize</li>
436<li>set listsize</li>
437<li>show language</li>
438<li>set language</li>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]439<li>show args</li>
440<li>set args [args]</li>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]441</ul>
442
443<h2>TODO:</h2>
444<ul>
445<li>info frame</li>
446<li>break</li>
447<li>print</li>
448<li>ptype</li>
449
450<li>info types</li>
451<li>info variables</li>
452<li>info program</li>
453
454<li>info args</li>
455<li>info locals</li>
456<li>info catch</li>
457<li>... many others</li>
458</ul>
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]459
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]460</div>
461
462<!-- *********************************************************************** -->
463<div class="doc_section">
464 <a name="architecture">Architecture of the LLVM debugger</a>
465</div>
466<!-- *********************************************************************** -->
467
468<div class="doc_text">
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]469<p>The LLVM debugger is built out of three distinct layers of software. These
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]470layers provide clients with different interface options depending on what pieces
471of they want to implement themselves, and it also promotes code modularity and
472good design. The three layers are the <a href="#arch_debugger">Debugger
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]473interface</a>, the <a href="#arch_info">"info" interfaces</a>, and the <a
474href="#arch_llvm-db"><tt>llvm-db</tt> tool</a> itself.</p>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]475</div>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]476
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]477<!-- ======================================================================= -->
478<div class="doc_subsection">
479 <a name="arch_debugger">The Debugger and InferiorProcess classes</a>
480</div>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]481
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]482<div class="doc_text">
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]483<p>The Debugger class (defined in the <tt>include/llvm/Debugger/</tt> directory)
484is a low-level class which is used to maintain information about the loaded
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]485program, as well as start and stop the program running as necessary. This class
486does not provide any high-level analysis or control over the program, only
487exposing simple interfaces like <tt>load/unloadProgram</tt>,
488<tt>create/killProgram</tt>, <tt>step/next/finish/contProgram</tt>, and
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]489low-level methods for installing breakpoints.</p>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]490
491<p>
492The Debugger class is itself a wrapper around the lowest-level InferiorProcess
493class. This class is used to represent an instance of the program running under
494debugger control. The InferiorProcess class can be implemented in different
495ways for different targets and execution scenarios (e.g., remote debugging).
496The InferiorProcess class exposes a small and simple collection of interfaces
497which are useful for inspecting the current state of the program (such as
498collecting stack trace information, reading the memory image of the process,
499etc). The interfaces in this class are designed to be as low-level and simple
500as possible, to make it easy to create new instances of the class.
501</p>
502
503<p>
504The Debugger class exposes the currently active instance of InferiorProcess
505through the <tt>Debugger::getRunningProcess</tt> method, which returns a
506<tt>const</tt> reference to the class. This means that clients of the Debugger
507class can only <b>inspect</b> the running instance of the program directly. To
508change the executing process in some way, they must use the interces exposed by
509the Debugger class.
510</p>
511</div>
512
513<!-- ======================================================================= -->
514<div class="doc_subsection">
515 <a name="arch_info">The RuntimeInfo, ProgramInfo, and SourceLanguage classes</a>
516</div>
517
518<div class="doc_text">
519<p>
520The next-highest level of debugger abstraction is provided through the
521ProgramInfo, RuntimeInfo, SourceLanguage and related classes (also defined in
522the <tt>include/llvm/Debugger/</tt> directory). These classes efficiently
523decode the debugging information and low-level interfaces exposed by
524InferiorProcess into a higher-level representation, suitable for analysis by the
525debugger.
526</p>
527
528<p>
529The ProgramInfo class exposes a variety of different kinds of information about
530the program objects in the source-level-language. The SourceFileInfo class
531represents a source-file in the program (e.g. a .cpp or .h file). The
532SourceFileInfo class captures information such as which SourceLanguage was used
533to compile the file, where the debugger can get access to the actual file text
534(which is lazily loaded on demand), etc. The SourceFunctionInfo class
535represents a... <b>FIXME: finish</b>. The ProgramInfo class provides interfaces
536to lazily find and decode the information needed to create the Source*Info
537classes requested by the debugger.
538</p>
539
540<p>
541The RuntimeInfo class exposes information about the currently executed program,
542by decoding information from the InferiorProcess and ProgramInfo classes. It
543provides a StackFrame class which provides an easy-to-use interface for
544inspecting the current and suspended stack frames in the program.
545</p>
546
547<p>
548The SourceLanguage class is an abstract interface used by the debugger to
549perform all source-language-specific tasks. For example, this interface is used
550by the ProgramInfo class to decode language-specific types and functions and by
551the debugger front-end (such as <a href="#arch_llvm-db"><tt>llvm-db</tt></a> to
552evaluate source-langauge expressions typed into the debugger. This class uses
553the RuntimeInfo &amp; ProgramInfo classes to get information about the current
554execution context and the loaded program, respectively.
555</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]556
557</div>
558
559<!-- ======================================================================= -->
560<div class="doc_subsection">
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]561 <a name="arch_llvm-db">The <tt>llvm-db</tt> tool</a>
562</div>
563
564<div class="doc_text">
565<p>
566The <tt>llvm-db</tt> is designed to be a debugger providing an interface as <a
567href="#llvm-db">similar to GDB</a> as reasonable, but no more so than that.
568Because the <a href="#arch_debugger">Debugger</a> and <a
569href="#arch_info">info</a> classes implement all of the heavy lifting and
570analysis, <tt>llvm-db</tt> (which lives in <tt>llvm/tools/llvm-db</tt>) consists
571mainly of of code to interact with the user and parse commands. The CLIDebugger
572constructor registers all of the builtin commands for the debugger, and each
573command is implemented as a CLIDebugger::[name]Command method.
574</p>
575</div>
576
577
578<!-- ======================================================================= -->
579<div class="doc_subsection">
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]580 <a name="arch_todo">Short-term TODO list</a>
581</div>
582
583<div class="doc_text">
584
585<p>
586FIXME: this section will eventually go away. These are notes to myself of
587things that should be implemented, but haven't yet.
588</p>
589
590<p>
591<b>Breakpoints:</b> Support is already implemented in the 'InferiorProcess'
592class, though it hasn't been tested yet. To finish breakpoint support, we need
593to implement breakCommand (which should reuse the linespec parser from the list
594command), and handle the fact that 'break foo' or 'break file.c:53' may insert
595multiple breakpoints. Also, if you say 'break file.c:53' and there is no
596stoppoint on line 53, the breakpoint should go on the next available line. My
597idea was to have the Debugger class provide a "Breakpoint" class which
598encapsulated this messiness, giving the debugger front-end a simple interface.
599The debugger front-end would have to map the really complex semantics of
600temporary breakpoints and 'conditional' breakpoints onto this intermediate
601level. Also, breakpoints should survive as much as possible across program
602reloads.
603</p>
604
605<p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]606<b>UnixLocalInferiorProcess.cpp speedup</b>: There is no reason for the debugged
607process to code gen the globals corresponding to debug information. The
608IntrinsicLowering object could instead change descriptors into constant expr
609casts of the constant address of the LLVM objects for the descriptors. This
610would also allow us to eliminate the mapping back and forth between physical
611addresses that must be done.</p>
612
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]613<p>
614<b>Process deaths</b>: The InferiorProcessDead exception should be extended to
615know "how" a process died, i.e., it was killed by a signal. This is easy to
616collect in the UnixLocalInferiorProcess, we just need to represent it.</p>
617
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]618</div>
619
620<!-- *********************************************************************** -->
621<div class="doc_section">
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]622 <a name="format">Debugging information format</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]623</div>
624<!-- *********************************************************************** -->
625
626<div class="doc_text">
627
628<p>LLVM debugging information has been carefully designed to make it possible
629for the optimizer to optimize the program and debugging information without
630necessarily having to know anything about debugging information. In particular,
631the global constant merging pass automatically eliminates duplicated debugging
632information (often caused by header files), the global dead code elimination
633pass automatically deletes debugging information for a function if it decides to
634delete the function, and the linker eliminates debug information when it merges
635<tt>linkonce</tt> functions.</p>
636
637<p>To do this, most of the debugging information (descriptors for types,
638variables, functions, source files, etc) is inserted by the language front-end
639in the form of LLVM global variables. These LLVM global variables are no
640different from any other global variables, except that they have a web of LLVM
641intrinsic functions that point to them. If the last references to a particular
642piece of debugging information are deleted (for example, by the
643<tt>-globaldce</tt> pass), the extraneous debug information will automatically
644become dead and be removed by the optimizer.</p>
645
646<p>The debugger is designed to be agnostic about the contents of most of the
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]647debugging information. It uses a <a href="#arch_info">source-language-specific
648module</a> to decode the information that represents variables, types,
649functions, namespaces, etc: this allows for arbitrary source-language semantics
650and type-systems to be used, as long as there is a module written for the
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]651debugger to interpret the information.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]652
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]653<p>To provide basic functionality, the LLVM debugger does have to make some
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]654assumptions about the source-level language being debugged, though it keeps
655these to a minimum. The only common features that the LLVM debugger assumes
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]656exist are <a href="#format_common_source_files">source files</a>, and <a
657href="#format_program_objects">program objects</a>. These abstract objects are
658used by the debugger to form stack traces, show information about local
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]659variables, etc.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]660
661<p>This section of the documentation first describes the representation aspects
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]662common to any source-language. The <a href="#ccxx_frontend">next section</a>
663describes the data layout conventions used by the C and C++ front-ends.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]664
665</div>
666
667<!-- ======================================================================= -->
668<div class="doc_subsection">
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]669 <a name="format_common_anchors">Anchors for global objects</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]670</div>
671
672<div class="doc_text">
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]673<p>One important aspect of the LLVM debug representation is that it allows the
674LLVM debugger to efficiently index all of the global objects without having the
675scan the program. To do this, all of the global objects use "anchor" globals of
676type "<tt>{}</tt>", with designated names. These anchor objects obviously do
677not contain any content or meaning by themselves, but all of the global objects
678of a particular type (e.g., source file descriptors) contain a pointer to the
679anchor. This pointer allows the debugger to use def-use chains to find all
680global objects of that type.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]681
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]682<p>So far, the following names are recognized as anchors by the LLVM
683debugger:</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]684
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]685<pre>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]686 %<a href="#format_common_source_files">llvm.dbg.translation_units</a> = linkonce global {} {}
687 %<a href="#format_program_objects">llvm.dbg.globals</a> = linkonce global {} {}
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]688</pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]689
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]690<p>Using anchors in this way (where the source file descriptor points to the
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]691anchors, as opposed to having a list of source file descriptors) allows for the
692standard dead global elimination and merging passes to automatically remove
693unused debugging information. If the globals were kept track of through lists,
694there would always be an object pointing to the descriptors, thus would never be
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]695deleted.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]696
697</div>
698
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]699<!-- ======================================================================= -->
700<div class="doc_subsection">
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]701 <a name="format_common_stoppoint">
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]702 Representing stopping points in the source program
703 </a>
704</div>
705
706<div class="doc_text">
707
708<p>LLVM debugger "stop points" are a key part of the debugging representation
709that allows the LLVM to maintain simple semantics for <a
710href="#debugopt">debugging optimized code</a>. The basic idea is that the
711front-end inserts calls to the <tt>%llvm.dbg.stoppoint</tt> intrinsic function
712at every point in the program where the debugger should be able to inspect the
713program (these correspond to places the debugger stops when you "<tt>step</tt>"
714through it). The front-end can choose to place these as fine-grained as it
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]715would like (for example, before every subexpression evaluated), but it is
716recommended to only put them after every source statement that includes
717executable code.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]718
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]719<p>Using calls to this intrinsic function to demark legal points for the
720debugger to inspect the program automatically disables any optimizations that
721could potentially confuse debugging information. To non-debug-information-aware
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]722transformations, these calls simply look like calls to an external function,
723which they must assume to do anything (including reading or writing to any part
724of reachable memory). On the other hand, it does not impact many optimizations,
725such as code motion of non-trapping instructions, nor does it impact
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]726optimization of subexpressions, code duplication transformations, or basic-block
727reordering transformations.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]728
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]729<p>An important aspect of the calls to the <tt>%llvm.dbg.stoppoint</tt>
730intrinsic is that the function-local debugging information is woven together
731with use-def chains. This makes it easy for the debugger to, for example,
732locate the 'next' stop point. For a concrete example of stop points, see the
733example in <a href="#format_common_lifetime">the next section</a>.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]734
735</div>
736
737
738<!-- ======================================================================= -->
739<div class="doc_subsection">
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]740 <a name="format_common_lifetime">Object lifetimes and scoping</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]741</div>
742
743<div class="doc_text">
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]744<p>In many languages, the local variables in functions can have their lifetime
745or scope limited to a subset of a function. In the C family of languages, for
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]746example, variables are only live (readable and writable) within the source block
747that they are defined in. In functional languages, values are only readable
748after they have been defined. Though this is a very obvious concept, it is also
749non-trivial to model in LLVM, because it has no notion of scoping in this sense,
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]750and does not want to be tied to a language's scoping rules.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]751
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]752<p>In order to handle this, the LLVM debug format uses the notion of "regions"
753of a function, delineated by calls to intrinsic functions. These intrinsic
754functions define new regions of the program and indicate when the region
755lifetime expires. Consider the following C fragment, for example:</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]756
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]757<pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]7581. void foo() {
7592. int X = ...;
7603. int Y = ...;
7614. {
7625. int Z = ...;
7636. ...
7647. }
7658. ...
7669. }
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]767</pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]768
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]769<p>Compiled to LLVM, this function would be represented like this (FIXME: CHECK
770AND UPDATE THIS):</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]771
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]772<pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]773void %foo() {
774 %X = alloca int
775 %Y = alloca int
776 %Z = alloca int
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]777 <a name="#icl_ex_D1">%D1</a> = call {}* %llvm.dbg.func.start(<a href="#format_program_objects">%lldb.global</a>* %d.foo)
778 %D2 = call {}* <a href="#format_common_stoppoint">%llvm.dbg.stoppoint</a>({}* %D1, uint 2, uint 2, <a href="#format_common_source_files">%lldb.compile_unit</a>* %file)
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]779
780 %D3 = call {}* %llvm.dbg.DEFINEVARIABLE({}* %D2, ...)
781 <i>;; Evaluate expression on line 2, assigning to X.</i>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]782 %D4 = call {}* <a href="#format_common_stoppoint">%llvm.dbg.stoppoint</a>({}* %D3, uint 3, uint 2, <a href="#format_common_source_files">%lldb.compile_unit</a>* %file)
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]783
784 %D5 = call {}* %llvm.dbg.DEFINEVARIABLE({}* %D4, ...)
785 <i>;; Evaluate expression on line 3, assigning to Y.</i>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]786 %D6 = call {}* <a href="#format_common_stoppoint">%llvm.dbg.stoppoint</a>({}* %D5, uint 5, uint 4, <a href="#format_common_source_files">%lldb.compile_unit</a>* %file)
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]787
788 <a name="#icl_ex_D1">%D7</a> = call {}* %llvm.region.start({}* %D6)
789 %D8 = call {}* %llvm.dbg.DEFINEVARIABLE({}* %D7, ...)
790 <i>;; Evaluate expression on line 5, assigning to Z.</i>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]791 %D9 = call {}* <a href="#format_common_stoppoint">%llvm.dbg.stoppoint</a>({}* %D8, uint 6, uint 4, <a href="#format_common_source_files">%lldb.compile_unit</a>* %file)
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]792
793 <i>;; Code for line 6.</i>
794 %D10 = call {}* %llvm.region.end({}* %D9)
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]795 %D11 = call {}* <a href="#format_common_stoppoint">%llvm.dbg.stoppoint</a>({}* %D10, uint 8, uint 2, <a href="#format_common_source_files">%lldb.compile_unit</a>* %file)
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]796
797 <i>;; Code for line 8.</i>
798 <a name="#icl_ex_D1">%D12</a> = call {}* %llvm.region.end({}* %D11)
799 ret void
800}
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]801</pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]802
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]803<p>This example illustrates a few important details about the LLVM debugging
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]804information. In particular, it shows how the various intrinsics used are woven
805together with def-use and use-def chains, similar to how <a
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]806href="#format_common_anchors">anchors</a> are used with globals. This allows
807the debugger to analyze the relationship between statements, variable
808definitions, and the code used to implement the function.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]809
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]810<p>In this example, two explicit regions are defined, one with the <a
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]811href="#icl_ex_D1">definition of the <tt>%D1</tt> variable</a> and one with the
812<a href="#icl_ex_D7">definition of <tt>%D7</tt></a>. In the case of
813<tt>%D1</tt>, the debug information indicates that the function whose <a
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]814href="#format_program_objects">descriptor</a> is specified as an argument to the
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]815intrinsic. This defines a new stack frame whose lifetime ends when the region
816is ended by <a href="#icl_ex_D12">the <tt>%D12</tt> call</a>.</p>
817
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]818<p>Using regions to represent the boundaries of source-level functions allow
819LLVM interprocedural optimizations to arbitrarily modify LLVM functions without
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]820having to worry about breaking mapping information between the LLVM code and the
821and source-level program. In particular, the inliner requires no modification
822to support inlining with debugging information: there is no explicit correlation
823drawn between LLVM functions and their source-level counterparts (note however,
824that if the inliner inlines all instances of a non-strong-linkage function into
825its caller that it will not be possible for the user to manually invoke the
826inlined function from the debugger).</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]827
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]828<p>Once the function has been defined, the <a
829href="#format_common_stoppoint">stopping point</a> corresponding to line #2 of
830the function is encountered. At this point in the function, <b>no</b> local
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]831variables are live. As lines 2 and 3 of the example are executed, their
832variable definitions are automatically introduced into the program, without the
833need to specify a new region. These variables do not require new regions to be
834introduced because they go out of scope at the same point in the program: line
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]8359.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]836
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]837<p>In contrast, the <tt>Z</tt> variable goes out of scope at a different time,
838on line 7. For this reason, it is defined within <a href="#icl_ex_D7">the
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]839<tt>%D7</tt> region</a>, which kills the availability of <tt>Z</tt> before the
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]840code for line 8 is executed. In this way, regions can support arbitrary
841source-language scoping rules, as long as they can only be nested (ie, one scope
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]842cannot partially overlap with a part of another scope).</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]843
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]844<p>It is worth noting that this scoping mechanism is used to control scoping of
845all declarations, not just variable declarations. For example, the scope of a
846C++ using declaration is controlled with this, and the <tt>llvm-db</tt> C++
847support routines could use this to change how name lookup is performed (though
848this is not implemented yet).</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]849
850</div>
851
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]852<!-- ======================================================================= -->
853<div class="doc_subsection">
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]854 <a name="format_common_descriptors">Object descriptor formats</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]855</div>
856
857<div class="doc_text">
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]858<p>The LLVM debugger expects the descriptors for program objects to start in a
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]859canonical format, but the descriptors can include additional information
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]860appended at the end that is source-language specific. All LLVM debugging
861information is versioned, allowing backwards compatibility in the case that the
862core structures need to change in some way. Also, all debugging information
863objects start with a <a href="#format_common_tags">tag</a> to indicate what type
864of object it is. The source-language is allows to define its own objects, by
865using unreserved tag numbers.</p>
866
867<p>The lowest-level descriptor are those describing <a
868href="#format_common_source_files">the files containing the program source
869code</a>, as most other descriptors (sometimes indirectly) refer to them.
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]870</p>
871</div>
872
873
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]874<!-- ------------------------------------------------------------------------ ->
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]875<div class="doc_subsubsection">
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]876 <a name="format_common_source_files">Representation of source files</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]877</div>
878
879<div class="doc_text">
880<p>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]881Source file descriptors are patterned after the Dwarf "compile_unit" object.
882The descriptor currently is defined to have at least the following LLVM
883type entries:</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]884
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]885<pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]886%lldb.compile_unit = type {
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]887 uint, <i>;; Tag: <a href="#tag_compile_unit">LLVM_COMPILE_UNIT</a></i>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]888 ushort, <i>;; LLVM debug version number</i>
889 ushort, <i>;; Dwarf language identifier</i>
890 sbyte*, <i>;; Filename</i>
891 sbyte*, <i>;; Working directory when compiled</i>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]892 sbyte* <i>;; Producer of the debug information</i>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]893}
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]894</pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]895
896<p>
897These descriptors contain the version number for the debug info, a source
898language ID for the file (we use the Dwarf 3.0 ID numbers, such as
899<tt>DW_LANG_C89</tt>, <tt>DW_LANG_C_plus_plus</tt>, <tt>DW_LANG_Cobol74</tt>,
900etc), three strings describing the filename, working directory of the compiler,
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]901and an identifier string for the compiler that produced it. Note that actual
902compile_unit declarations must also include an <a
903href="#format_common_anchors">anchor</a> to <tt>llvm.dbg.translation_units</tt>,
904but it is not specified where the anchor is to be located. Here is an example
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]905descriptor:
906</p>
907
908<p><pre>
909%arraytest_source_file = internal constant %lldb.compile_unit {
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]910 <a href="#tag_compile_unit">uint 17</a>, ; Tag value
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]911 ushort 0, ; Version #0
912 ushort 1, ; DW_LANG_C89
913 sbyte* getelementptr ([12 x sbyte]* %.str_1, long 0, long 0), ; filename
914 sbyte* getelementptr ([12 x sbyte]* %.str_2, long 0, long 0), ; working dir
915 sbyte* getelementptr ([12 x sbyte]* %.str_3, long 0, long 0), ; producer
916 {}* %llvm.dbg.translation_units ; Anchor
917}
918%.str_1 = internal constant [12 x sbyte] c"arraytest.c\00"
919%.str_2 = internal constant [12 x sbyte] c"/home/sabre\00"
920%.str_3 = internal constant [12 x sbyte] c"llvmgcc 3.4\00"
921</pre></p>
922
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]923<p>
924Note that the LLVM constant merging pass should eliminate duplicate copies of
925the strings that get emitted to each translation unit, such as the producer.
926</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]927
928</div>
929
930
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]931<!-- ----------------------------------------------------------------------- -->
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]932<div class="doc_subsubsection">
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]933 <a name="format_program_objects">Representation of program objects</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]934</div>
935
936<div class="doc_text">
937<p>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]938The LLVM debugger needs to know about some source-language program objects, in
939order to build stack traces, print information about local variables, and other
940related activities. The LLVM debugger differentiates between three different
941types of program objects: subprograms (functions, messages, methods, etc),
942variables (locals and globals), and others. Because source-languages have
943widely varying forms of these objects, the LLVM debugger expects only a few
944fields in the descriptor for each object:
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]945</p>
946
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]947<pre>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]948%lldb.object = type {
949 uint, <i>;; <a href="#format_common_tag">A tag</a></i>
950 <i>any</i>*, <i>;; The <a href="#format_common_object_contexts">context</a> for the object</i>
951 sbyte* <i>;; The object 'name'</i>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]952}
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]953</pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]954
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]955<p>The first field contains a tag for the descriptor. The second field contains
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]956either a pointer to the descriptor for the containing <a
957href="#format_common_source_files">source file</a>, or it contains a pointer to
958another program object whose context pointer eventually reaches a source file.
959Through this <a href="#format_common_object_contexts">context</a> pointer, the
960LLVM debugger can establish the debug version number of the object.</p>
961
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]962<p>The third field contains a string that the debugger can use to identify the
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]963object if it does not contain explicit support for the source-language in use
964(ie, the 'unknown' source language handler uses this string). This should be
965some sort of unmangled string that corresponds to the object, but it is a
966quality of implementation issue what exactly it contains (it is legal, though
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]967not useful, for all of these strings to be null).</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]968
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]969<p>Note again that descriptors can be extended to include
970source-language-specific information in addition to the fields required by the
971LLVM debugger. See the <a href="#ccxx_descriptors">section on the C/C++
972front-end</a> for more information. Also remember that global objects
973(functions, selectors, global variables, etc) must contain an <a
Misha Brukman179bf4b2004-06-03 23:42:24 +0000[diff] [blame]974href="#format_common_anchors">anchor</a> to the <tt>llvm.dbg.globals</tt>
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]975variable.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]976</div>
977
978
979<!-- ======================================================================= -->
980<div class="doc_subsection">
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]981 <a name="format_common_object_contexts">Program object contexts</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]982</div>
983
984<div class="doc_text">
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]985<pre>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]986Allow source-language specific contexts, use to identify namespaces etc
987Must end up in a source file descriptor.
988Debugger core ignores all unknown context objects.
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]989</pre>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]990</div>
991
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]992<!-- ======================================================================= -->
993<div class="doc_subsection">
994 <a name="format_common_intrinsics">Debugger intrinsic functions</a>
995</div>
996
997<div class="doc_text">
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]998<pre>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]999Define each intrinsics, as an extension of the language reference manual.
1000
1001llvm.dbg.stoppoint
1002llvm.dbg.region.start
1003llvm.dbg.region.end
1004llvm.dbg.function.start
1005llvm.dbg.declare
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1006</pre>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]1007</div>
1008
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]1009<!-- ======================================================================= -->
1010<div class="doc_subsection">
1011 <a name="format_common_tags">Values for debugger tags</a>
1012</div>
1013
1014<div class="doc_text">
1015
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1016<p>Happen to be the same value as the similarly named Dwarf-3 tags, this may
1017change in the future.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]1018
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1019<pre>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]1020 <a name="tag_compile_unit">LLVM_COMPILE_UNIT</a> : 17
1021 <a name="tag_subprogram">LLVM_SUBPROGRAM</a> : 46
1022 <a name="tag_variable">LLVM_VARIABLE</a> : 52
1023<!-- <a name="tag_formal_parameter">LLVM_FORMAL_PARAMETER : 5-->
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1024</pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]1025</div>
1026
1027
1028
1029<!-- *********************************************************************** -->
1030<div class="doc_section">
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]1031 <a name="ccxx_frontend">C/C++ front-end specific debug information</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]1032</div>
1033
1034<div class="doc_text">
1035
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1036<p>The C and C++ front-ends represent information about the program in a format
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]1037that is effectively identical to <a
1038href="http://www.eagercon.com/dwarf/dwarf3std.htm">Dwarf 3.0</a> in terms of
1039information content. This allows code generators to trivially support native
1040debuggers by generating standard dwarf information, and contains enough
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]1041information for non-dwarf targets to translate it as needed.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]1042
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1043<p>The basic debug information required by the debugger is (intentionally)
1044designed to be as minimal as possible. This basic information is so minimal
1045that it is unlikely that <b>any</b> source-language could be adequately
1046described by it. Because of this, the debugger format was designed for
1047extension to support source-language-specific information. The extended
1048descriptors are read and interpreted by the <a
1049href="#arch_info">language-specific</a> modules in the debugger if there is
1050support available, otherwise it is ignored.</p>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]1051
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1052<p>This section describes the extensions used to represent C and C++ programs.
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]1053Other languages could pattern themselves after this (which itself is tuned to
1054representing programs in the same way that Dwarf 3 does), or they could choose
1055to provide completely different extensions if they don't fit into the Dwarf
1056model. As support for debugging information gets added to the various LLVM
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1057source-language front-ends, the information used should be documented here.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]1058
1059</div>
1060
1061<!-- ======================================================================= -->
1062<div class="doc_subsection">
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]1063 <a name="ccxx_pse">Program Scope Entries</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]1064</div>
1065
1066<div class="doc_text">
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1067<p>TODO</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]1068</div>
1069
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1070<!-- -------------------------------------------------------------------------->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]1071<div class="doc_subsubsection">
1072 <a name="ccxx_compilation_units">Compilation unit entries</a>
1073</div>
1074
1075<div class="doc_text">
1076<p>
1077Translation units do not add any information over the standard <a
1078href="#format_common_source_files">source file representation</a> already
1079expected by the debugger. As such, it uses descriptors of the type specified,
1080with a trailing <a href="#format_common_anchors">anchor</a>.
1081</p>
1082</div>
1083
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1084<!-- -------------------------------------------------------------------------->
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]1085<div class="doc_subsubsection">
1086 <a name="ccxx_modules">Module, namespace, and importing entries</a>
1087</div>
1088
1089<div class="doc_text">
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1090<p>TODO</p>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]1091</div>
1092
1093<!-- ======================================================================= -->
1094<div class="doc_subsection">
1095 <a name="ccxx_dataobjects">Data objects (program variables)</a>
1096</div>
1097
1098<div class="doc_text">
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1099<p>TODO</p>
Chris Lattner8ff75902004-01-06 05:31:32 +0000[diff] [blame]1100</div>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]1101
1102
1103<!-- *********************************************************************** -->
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1104
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]1105<hr>
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1106<address>
1107 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
1108 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
1109 <a href="http://validator.w3.org/check/referer"><img
1110 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
1111
1112 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
1113 <a href="http://llvm.cs.uiuc.edu">LLVM Compiler Infrastructure</a><br>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]1114 Last modified: $Date$
Misha Brukman82873732004-05-12 19:21:57 +0000[diff] [blame]1115</address>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000[diff] [blame]1116
1117</body>
1118</html>