blob: 62417f10b92a960c0f086454c2c4549629cff296 [file] [log] [blame]
Chris Lattnerbdfb3392004-01-05 05:06:33 +00001<!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
Reid Spencerd3f876c2004-11-01 08:19:36 +000012<table class="layout" style="width:100%">
13 <tr class="layout">
14 <td class="left">
Chris Lattnerbdfb3392004-01-05 05:06:33 +000015<ul>
Misha Brukman82873732004-05-12 19:21:57 +000016 <li><a href="#introduction">Introduction</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +000017 <ol>
18 <li><a href="#phil">Philosophy behind LLVM debugging information</a></li>
Jim Laskey383e0092006-03-23 17:54:33 +000019 <li><a href="#consumers">Debug information consumers</a></li>
Chris Lattnerbdfb3392004-01-05 05:06:33 +000020 <li><a href="#debugopt">Debugging optimized code</a></li>
Misha Brukman82873732004-05-12 19:21:57 +000021 </ol></li>
Misha Brukman82873732004-05-12 19:21:57 +000022 <li><a href="#format">Debugging information format</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +000023 <ol>
Jim Laskeycec12a52006-03-14 18:08:46 +000024 <li><a href="#debug_info_descriptors">Debug information descriptors</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +000025 <ul>
Jim Laskeycec12a52006-03-14 18:08:46 +000026 <li><a href="#format_anchors">Anchor descriptors</a></li>
27 <li><a href="#format_compile_units">Compile unit descriptors</a></li>
28 <li><a href="#format_global_variables">Global variable descriptors</a></li>
29 <li><a href="#format_subprograms">Subprogram descriptors</a></li>
Jim Laskey3d11bee2006-03-15 19:10:52 +000030 <li><a href="#format_blocks">Block descriptors</a></li>
Jim Laskeycec12a52006-03-14 18:08:46 +000031 <li><a href="#format_basic_type">Basic type descriptors</a></li>
32 <li><a href="#format_derived_type">Derived type descriptors</a></li>
33 <li><a href="#format_composite_type">Composite type descriptors</a></li>
34 <li><a href="#format_subrange">Subrange descriptors</a></li>
35 <li><a href="#format_enumeration">Enumerator descriptors</a></li>
Jim Laskey383e0092006-03-23 17:54:33 +000036 <li><a href="#format_variables">Local variables</a></li>
Misha Brukman82873732004-05-12 19:21:57 +000037 </ul></li>
Jim Laskeycec12a52006-03-14 18:08:46 +000038 <li><a href="#format_common_intrinsics">Debugger intrinsic functions</a>
39 <ul>
40 <li><a href="#format_common_stoppoint">llvm.dbg.stoppoint</a></li>
41 <li><a href="#format_common_func_start">llvm.dbg.func.start</a></li>
42 <li><a href="#format_common_region_start">llvm.dbg.region.start</a></li>
43 <li><a href="#format_common_region_end">llvm.dbg.region.end</a></li>
44 <li><a href="#format_common_declare">llvm.dbg.declare</a></li>
45 </ul></li>
46 <li><a href="#format_common_stoppoints">Representing stopping points in the
47 source program</a></li>
Misha Brukman82873732004-05-12 19:21:57 +000048 </ol></li>
49 <li><a href="#ccxx_frontend">C/C++ front-end specific debug information</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +000050 <ol>
Jim Laskeycec12a52006-03-14 18:08:46 +000051 <li><a href="#ccxx_compile_units">C/C++ source file information</a></li>
52 <li><a href="#ccxx_global_variable">C/C++ global variable information</a></li>
53 <li><a href="#ccxx_subprogram">C/C++ function information</a></li>
54 <li><a href="#ccxx_basic_types">C/C++ basic types</a></li>
55 <li><a href="#ccxx_derived_types">C/C++ derived types</a></li>
56 <li><a href="#ccxx_composite_types">C/C++ struct/union types</a></li>
57 <li><a href="#ccxx_enumeration_types">C/C++ enumeration types</a></li>
Misha Brukman82873732004-05-12 19:21:57 +000058 </ol></li>
Chris Lattnerbdfb3392004-01-05 05:06:33 +000059</ul>
Misha Brukman82873732004-05-12 19:21:57 +000060</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +000061<td class="right">
Misha Brukmane849a1a2004-05-12 21:26:16 +000062<img src="img/venusflytrap.jpg" alt="A leafy and green bug eater" width="247"
Misha Brukman82873732004-05-12 19:21:57 +000063height="369">
64</td>
Reid Spencerd3f876c2004-11-01 08:19:36 +000065</tr></table>
Misha Brukman82873732004-05-12 19:21:57 +000066
Chris Lattner7911ce22004-05-23 21:07:27 +000067<div class="doc_author">
Jim Laskeycec12a52006-03-14 18:08:46 +000068 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
69 and <a href="mailto:jlaskey@apple.com">Jim Laskey</a></p>
Chris Lattner7911ce22004-05-23 21:07:27 +000070</div>
71
Chris Lattnerbdfb3392004-01-05 05:06:33 +000072
73<!-- *********************************************************************** -->
Misha Brukman94218a72004-12-09 20:27:37 +000074<div class="doc_section"><a name="introduction">Introduction</a></div>
75<!-- *********************************************************************** -->
Chris Lattnerbdfb3392004-01-05 05:06:33 +000076
77<div class="doc_text">
78
79<p>This document is the central repository for all information pertaining to
Jim Laskeycec12a52006-03-14 18:08:46 +000080debug information in LLVM. It describes the <a href="#format">actual format
81that the LLVM debug information</a> takes, which is useful for those interested
82in creating front-ends or dealing directly with the information. Further, this
83document provides specifc examples of what debug information for C/C++.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +000084
85</div>
86
87<!-- ======================================================================= -->
88<div class="doc_subsection">
89 <a name="phil">Philosophy behind LLVM debugging information</a>
90</div>
91
92<div class="doc_text">
93
Misha Brukman82873732004-05-12 19:21:57 +000094<p>The idea of the LLVM debugging information is to capture how the important
Chris Lattnerbdfb3392004-01-05 05:06:33 +000095pieces of the source-language's Abstract Syntax Tree map onto LLVM code.
96Several design aspects have shaped the solution that appears here. The
97important ones are:</p>
98
Misha Brukman82873732004-05-12 19:21:57 +000099<ul>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000100<li>Debugging information should have very little impact on the rest of the
101compiler. No transformations, analyses, or code generators should need to be
102modified because of debugging information.</li>
103
104<li>LLVM optimizations should interact in <a href="#debugopt">well-defined and
105easily described ways</a> with the debugging information.</li>
106
107<li>Because LLVM is designed to support arbitrary programming languages,
108LLVM-to-LLVM tools should not need to know anything about the semantics of the
109source-level-language.</li>
110
111<li>Source-level languages are often <b>widely</b> different from one another.
112LLVM should not put any restrictions of the flavor of the source-language, and
113the debugging information should work with any language.</li>
114
115<li>With code generator support, it should be possible to use an LLVM compiler
Chris Lattner8ff75902004-01-06 05:31:32 +0000116to compile a program to native machine code and standard debugging formats.
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000117This allows compatibility with traditional machine-code level debuggers, like
118GDB or DBX.</li>
119
Misha Brukman82873732004-05-12 19:21:57 +0000120</ul>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000121
Misha Brukman82873732004-05-12 19:21:57 +0000122<p>The approach used by the LLVM implementation is to use a small set of <a
Chris Lattner8ff75902004-01-06 05:31:32 +0000123href="#format_common_intrinsics">intrinsic functions</a> to define a mapping
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000124between LLVM program objects and the source-level objects. The description of
125the source-level program is maintained in LLVM global variables in an <a
Chris Lattner8ff75902004-01-06 05:31:32 +0000126href="#ccxx_frontend">implementation-defined format</a> (the C/C++ front-end
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000127currently uses working draft 7 of the <a
128href="http://www.eagercon.com/dwarf/dwarf3std.htm">Dwarf 3 standard</a>).</p>
129
Jim Laskeycec12a52006-03-14 18:08:46 +0000130<p>When a program is being debugged, a debugger interacts with the user and
131turns the stored debug information into source-language specific information.
Jim Laskey3d11bee2006-03-15 19:10:52 +0000132As such, a debugger must be aware of the source-language, and is thus tied to
Jim Laskeycec12a52006-03-14 18:08:46 +0000133a specific language of family of languages.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000134
135</div>
136
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000137<!-- ======================================================================= -->
138<div class="doc_subsection">
Jim Laskey383e0092006-03-23 17:54:33 +0000139 <a name="consumers">Debug information consumers</a>
140</div>
141
142<div class="doc_text">
143<p>The role of debug information is to provide meta information normally
144stripped away during the compilation process. This meta information provides an
145llvm user a relationship between generated code and the original program source
146code.</p>
147
148<p>Currently, debug information is consumed by the DwarfWriter to produce dwarf
149information used by the gdb debugger. Other targets could use the same
150information to produce stabs or other debug forms.</p>
151
152<p>It would also be reasonable to use debug information to feed profiling tools
153for analysis of generated code, or, tools for reconstructing the original source
154from generated code.</p>
155
156<p>TODO - expound a bit more.</p>
157
158</div>
159
160<!-- ======================================================================= -->
161<div class="doc_subsection">
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000162 <a name="debugopt">Debugging optimized code</a>
163</div>
164
165<div class="doc_text">
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000166
Misha Brukman82873732004-05-12 19:21:57 +0000167<p>An extremely high priority of LLVM debugging information is to make it
168interact well with optimizations and analysis. In particular, the LLVM debug
169information provides the following guarantees:</p>
170
171<ul>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000172
173<li>LLVM debug information <b>always provides information to accurately read the
174source-level state of the program</b>, regardless of which LLVM optimizations
175have been run, and without any modification to the optimizations themselves.
176However, some optimizations may impact the ability to modify the current state
177of the program with a debugger, such as setting program variables, or calling
178function that have been deleted.</li>
179
180<li>LLVM optimizations gracefully interact with debugging information. If they
181are not aware of debug information, they are automatically disabled as necessary
182in the cases that would invalidate the debug info. This retains the LLVM
183features making it easy to write new transformations.</li>
184
185<li>As desired, LLVM optimizations can be upgraded to be aware of the LLVM
186debugging information, allowing them to update the debugging information as they
187perform aggressive optimizations. This means that, with effort, the LLVM
188optimizers could optimize debug code just as well as non-debug code.</li>
189
190<li>LLVM debug information does not prevent many important optimizations from
191happening (for example inlining, basic block reordering/merging/cleanup, tail
192duplication, etc), further reducing the amount of the compiler that eventually
193is "aware" of debugging information.</li>
194
195<li>LLVM debug information is automatically optimized along with the rest of the
196program, using existing facilities. For example, duplicate information is
197automatically merged by the linker, and unused information is automatically
198removed.</li>
199
Misha Brukman82873732004-05-12 19:21:57 +0000200</ul>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000201
Misha Brukman82873732004-05-12 19:21:57 +0000202<p>Basically, the debug information allows you to compile a program with
203"<tt>-O0 -g</tt>" and get full debug information, allowing you to arbitrarily
Jim Laskey3d11bee2006-03-15 19:10:52 +0000204modify the program as it executes from a debugger. Compiling a program with
Misha Brukman82873732004-05-12 19:21:57 +0000205"<tt>-O3 -g</tt>" gives you full debug information that is always available and
206accurate for reading (e.g., you get accurate stack traces despite tail call
207elimination and inlining), but you might lose the ability to modify the program
208and call functions where were optimized out of the program, or inlined away
209completely.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000210
211</div>
212
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000213<!-- *********************************************************************** -->
214<div class="doc_section">
Chris Lattner8ff75902004-01-06 05:31:32 +0000215 <a name="format">Debugging information format</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000216</div>
217<!-- *********************************************************************** -->
218
219<div class="doc_text">
220
221<p>LLVM debugging information has been carefully designed to make it possible
222for the optimizer to optimize the program and debugging information without
223necessarily having to know anything about debugging information. In particular,
224the global constant merging pass automatically eliminates duplicated debugging
225information (often caused by header files), the global dead code elimination
226pass automatically deletes debugging information for a function if it decides to
227delete the function, and the linker eliminates debug information when it merges
228<tt>linkonce</tt> functions.</p>
229
230<p>To do this, most of the debugging information (descriptors for types,
231variables, functions, source files, etc) is inserted by the language front-end
232in the form of LLVM global variables. These LLVM global variables are no
233different from any other global variables, except that they have a web of LLVM
234intrinsic functions that point to them. If the last references to a particular
235piece of debugging information are deleted (for example, by the
236<tt>-globaldce</tt> pass), the extraneous debug information will automatically
237become dead and be removed by the optimizer.</p>
238
Jim Laskeycec12a52006-03-14 18:08:46 +0000239<p>Debug information is designed to be agnostic about the target debugger and
240debugging information representation (e.g. DWARF/Stabs/etc). It uses a generic
241machine debug information pass to decode the information that represents
242variables, types, functions, namespaces, etc: this allows for arbitrary
243source-language semantics and type-systems to be used, as long as there is a
244module written for the target debugger to interpret the information. In
245addition, debug global variables are declared in the <tt>"llvm.metadata"</tt>
246section. All values declared in this section are stripped away after target
247debug information is constructed and before the program object is emitted.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000248
Misha Brukman82873732004-05-12 19:21:57 +0000249<p>To provide basic functionality, the LLVM debugger does have to make some
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000250assumptions about the source-level language being debugged, though it keeps
251these to a minimum. The only common features that the LLVM debugger assumes
Jim Laskeycec12a52006-03-14 18:08:46 +0000252exist are <a href="#format_compile_units">source files</a>, and <a
253href="#format_global_variables">program objects</a>. These abstract objects are
Jim Laskey3d11bee2006-03-15 19:10:52 +0000254used by a debugger to form stack traces, show information about local
Misha Brukman82873732004-05-12 19:21:57 +0000255variables, etc.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000256
257<p>This section of the documentation first describes the representation aspects
Chris Lattner8ff75902004-01-06 05:31:32 +0000258common to any source-language. The <a href="#ccxx_frontend">next section</a>
259describes the data layout conventions used by the C and C++ front-ends.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000260
261</div>
262
263<!-- ======================================================================= -->
264<div class="doc_subsection">
Jim Laskeycec12a52006-03-14 18:08:46 +0000265 <a name="debug_info_descriptors">Debug information descriptors</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000266</div>
267
268<div class="doc_text">
Jim Laskeycec12a52006-03-14 18:08:46 +0000269<p>In consideration of the complexity and volume of debug information, LLVM
270provides a specification for well formed debug global variables. The constant
271value of each of these globals is one of a limited set of structures, known as
272debug descriptors.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000273
Jim Laskeycec12a52006-03-14 18:08:46 +0000274<p>Consumers of LLVM debug information expect the descriptors for program
275objects to start in a canonical format, but the descriptors can include
Jim Laskey383e0092006-03-23 17:54:33 +0000276additional information appended at the end that is source-language specific. All
277LLVM debugging information is versioned, allowing backwards compatibility in the
278case that the core structures need to change in some way. Also, all debugging
279information objects start with a tag to indicate what type of object it is. The
280source-language is allowed to define its own objects, by using unreserved tag
281numbers. We recommend using with tags in the range 0x1000 thru 0x2000 (there is
282a defined enum DW_TAG_user_base = 0x1000.)</p>
Jim Laskeycec12a52006-03-14 18:08:46 +0000283
284<p>The fields of debug descriptors used internally by LLVM (MachineDebugInfo)
285are restricted to only the simple data types <tt>int</tt>, <tt>uint</tt>,
286<tt>bool</tt>, <tt>float</tt>, <tt>double</tt>, <tt>sbyte*</tt> and <tt> { }*
287</tt>. References to arbitrary values are handled using a <tt> { }* </tt> and a
288cast to <tt> { }* </tt> expression; typically references to other field
289descriptors, arrays of descriptors or global variables.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000290
Misha Brukman82873732004-05-12 19:21:57 +0000291<pre>
Jim Laskeycec12a52006-03-14 18:08:46 +0000292 %llvm.dbg.object.type = type {
293 uint, ;; A tag
294 ...
295 }
Misha Brukman82873732004-05-12 19:21:57 +0000296</pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000297
Jim Laskeycec12a52006-03-14 18:08:46 +0000298<p>The first field of a descriptor is always an <tt>uint</tt> containing a tag
299value identifying the content of the descriptor. The remaining fields are
300specific to the descriptor. The values of tags are loosely bound to the tag
301values of Dwarf information entries. However, that does not restrict the use of
302the information supplied to Dwarf targets.</p>
303
304<p>The details of the various descriptors follow.</p>
305
306</div>
307
308<!-- ======================================================================= -->
309<div class="doc_subsubsection">
310 <a name="format_anchors">Anchor descriptors</a>
311</div>
312
313<div class="doc_text">
314
315<pre>
316 %<a href="#format_anchors">llvm.dbg.anchor.type</a> = type {
317 uint, ;; Tag = 0
318 uint ;; Tag of descriptors grouped by the anchor
319 }
320</pre>
321
322<p>One important aspect of the LLVM debug representation is that it allows the
323LLVM debugger to efficiently index all of the global objects without having the
324scan the program. To do this, all of the global objects use "anchor"
325descriptors with designated names. All of the global objects of a particular
326type (e.g., compile units) contain a pointer to the anchor. This pointer allows
Jim Laskey3d11bee2006-03-15 19:10:52 +0000327a debugger to use def-use chains to find all global objects of that type.</p>
Jim Laskeycec12a52006-03-14 18:08:46 +0000328
329<p>The following names are recognized as anchors by LLVM:</p>
330
331<pre>
332 %<a href="#format_compile_units">llvm.dbg.compile_units</a> = linkonce constant %<a href="#format_anchors">llvm.dbg.anchor.type</a> { uint 0, uint 17 } ;; DW_TAG_compile_unit
333 %<a href="#format_global_variables">llvm.dbg.global_variables</a> = linkonce constant %<a href="#format_anchors">llvm.dbg.anchor.type</a> { uint 0, uint 52 } ;; DW_TAG_variable
334 %<a href="#format_subprograms">llvm.dbg.subprograms</a> = linkonce constant %<a href="#format_anchors">llvm.dbg.anchor.type</a> { uint 0, uint 46 } ;; DW_TAG_subprogram
335</pre>
336
337<p>Using anchors in this way (where the compile unit descriptor points to the
338anchors, as opposed to having a list of compile unit descriptors) allows for the
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000339standard dead global elimination and merging passes to automatically remove
340unused debugging information. If the globals were kept track of through lists,
341there would always be an object pointing to the descriptors, thus would never be
Misha Brukman82873732004-05-12 19:21:57 +0000342deleted.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000343
344</div>
345
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000346<!-- ======================================================================= -->
Jim Laskeycec12a52006-03-14 18:08:46 +0000347<div class="doc_subsubsection">
348 <a name="format_compile_units">Compile unit descriptors</a>
349</div>
350
351<div class="doc_text">
352
353<pre>
354 %<a href="#format_compile_units">llvm.dbg.compile_unit.type</a> = type {
355 uint, ;; Tag = 17 (DW_TAG_compile_unit)
356 { }*, ;; Compile unit anchor = cast = (%<a href="#format_anchors">llvm.dbg.anchor.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_units</a> to { }*)
Jim Laskey383e0092006-03-23 17:54:33 +0000357 uint, ;; LLVM debug version number = 3
Jim Laskeycec12a52006-03-14 18:08:46 +0000358 uint, ;; Dwarf language identifier (ex. DW_LANG_C89)
359 sbyte*, ;; Source file name
360 sbyte*, ;; Source file directory (includes trailing slash)
361 sbyte* ;; Producer (ex. "4.0.1 LLVM (LLVM research group)")
362 }
363</pre>
364
365<p>These descriptors contain the version number for the debug info (currently
Jim Laskey383e0092006-03-23 17:54:33 +00003663), a source language ID for the file (we use the Dwarf 3.0 ID numbers, such as
Jim Laskeycec12a52006-03-14 18:08:46 +0000367<tt>DW_LANG_C89</tt>, <tt>DW_LANG_C_plus_plus</tt>, <tt>DW_LANG_Cobol74</tt>,
368etc), three strings describing the filename, working directory of the compiler,
369and an identifier string for the compiler that produced it.</p>
370
371<p> Compile unit descriptors provide the root context for objects declared in a
372specific source file. Global variables and top level functions would be defined
373using this context. Compile unit descriptors also provide context for source
374line correspondence.</p>
375
376</div>
377
378<!-- ======================================================================= -->
379<div class="doc_subsubsection">
380 <a name="format_global_variables">Global variable descriptors</a>
381</div>
382
383<div class="doc_text">
384
385<pre>
386 %<a href="#format_global_variables">llvm.dbg.global_variable.type</a> = type {
387 uint, ;; Tag = 52 (DW_TAG_variable)
388 { }*, ;; Global variable anchor = cast (%<a href="#format_anchors">llvm.dbg.anchor.type</a>* %<a href="#format_global_variables">llvm.dbg.global_variables</a> to { }*),
Jim Laskey3d11bee2006-03-15 19:10:52 +0000389 { }*, ;; Reference to context descriptor
Jim Laskeycec12a52006-03-14 18:08:46 +0000390 sbyte*, ;; Name
Jim Laskey3d11bee2006-03-15 19:10:52 +0000391 { }*, ;; Reference to compile unit where defined
Jim Laskey383e0092006-03-23 17:54:33 +0000392 uint, ;; Line number where defined
Jim Laskeycec12a52006-03-14 18:08:46 +0000393 { }*, ;; Reference to type descriptor
394 bool, ;; True if the global is local to compile unit (static)
395 bool, ;; True if the global is defined in the compile unit (not extern)
Jim Laskey3d11bee2006-03-15 19:10:52 +0000396 { }* ;; Reference to the global variable
Jim Laskeycec12a52006-03-14 18:08:46 +0000397 }
398</pre>
399
400<p>These descriptors provide debug information about globals variables. The
401provide details such as name, type and where the variable is defined.</p>
402
403</div>
404
405<!-- ======================================================================= -->
406<div class="doc_subsubsection">
407 <a name="format_subprograms">Subprogram descriptors</a>
408</div>
409
410<div class="doc_text">
411
412<pre>
413 %<a href="#format_subprograms">llvm.dbg.subprogram.type</a> = type {
414 uint, ;; Tag = 46 (DW_TAG_subprogram)
415 { }*, ;; Subprogram anchor = cast (%<a href="#format_anchors">llvm.dbg.anchor.type</a>* %<a href="#format_subprograms">llvm.dbg.subprograms</a> to { }*),
Jim Laskey3d11bee2006-03-15 19:10:52 +0000416 { }*, ;; Reference to context descriptor
Jim Laskeycec12a52006-03-14 18:08:46 +0000417 sbyte*, ;; Name
Jim Laskey3d11bee2006-03-15 19:10:52 +0000418 { }*, ;; Reference to compile unit where defined
Jim Laskey383e0092006-03-23 17:54:33 +0000419 uint, ;; Line number where defined
Jim Laskeycec12a52006-03-14 18:08:46 +0000420 { }*, ;; Reference to type descriptor
421 bool, ;; True if the global is local to compile unit (static)
Jim Laskey383e0092006-03-23 17:54:33 +0000422 bool ;; True if the global is defined in the compile unit (not extern)
Jim Laskeycec12a52006-03-14 18:08:46 +0000423 }
Jim Laskeycec12a52006-03-14 18:08:46 +0000424</pre>
425
426<p>These descriptors provide debug information about functions, methods and
Jim Laskey383e0092006-03-23 17:54:33 +0000427subprograms. They provide details such as name, return types and the source
428location where the subprogram is defined.</p>
Jim Laskey3d11bee2006-03-15 19:10:52 +0000429
430</div>
431<!-- ======================================================================= -->
432<div class="doc_subsubsection">
433 <a name="format_blocks">Block descriptors</a>
434</div>
435
436<div class="doc_text">
437
438<pre>
439 %<a href="#format_blocks">llvm.dbg.block</a> = type {
440 uint, ;; Tag = 13 (DW_TAG_lexical_block)
Jim Laskey383e0092006-03-23 17:54:33 +0000441 { }* ;; Reference to context descriptor
Jim Laskey3d11bee2006-03-15 19:10:52 +0000442 }
443</pre>
444
445<p>These descriptors provide debug information about nested blocks within a
446subprogram. The array of member descriptors is used to define local variables
447and deeper nested blocks.</p>
448
Jim Laskeycec12a52006-03-14 18:08:46 +0000449</div>
450
451<!-- ======================================================================= -->
452<div class="doc_subsubsection">
453 <a name="format_basic_type">Basic type descriptors</a>
454</div>
455
456<div class="doc_text">
457
458<pre>
459 %<a href="#format_basic_type">llvm.dbg.basictype.type</a> = type {
460 uint, ;; Tag = 36 (DW_TAG_base_type)
461 { }*, ;; Reference to context (typically a compile unit)
462 sbyte*, ;; Name (may be "" for anonymous types)
463 { }*, ;; Reference to compile unit where defined (may be NULL)
Jim Laskey383e0092006-03-23 17:54:33 +0000464 uint, ;; Line number where defined (may be 0)
Jim Laskeycec12a52006-03-14 18:08:46 +0000465 uint, ;; Size in bits
466 uint, ;; Alignment in bits
467 uint, ;; Offset in bits
468 uint ;; Dwarf type encoding
469 }
470</pre>
471
472<p>These descriptors define primitive types used in the code. Example int, bool
473and float. The context provides the scope of the type, which is usually the top
474level. Since basic types are not usually user defined the compile unit and line
475number can be left as NULL and 0. The size, alignment and offset are expressed
476in bits and can be 64 bit values. The alignment is used to round the offset
477when embedded in a <a href="#format_composite_type">composite type</a>
478(example to keep float doubles on 64 bit boundaries.) The offset is the bit
479offset if embedded in a <a href="#format_composite_type">composite
480type</a>.</p>
481
482<p>The type encoding provides the details of the type. The values are typically
483one of the following;</p>
484
485<pre>
486 DW_ATE_address = 1
487 DW_ATE_boolean = 2
488 DW_ATE_float = 4
489 DW_ATE_signed = 5
490 DW_ATE_signed_char = 6
491 DW_ATE_unsigned = 7
492 DW_ATE_unsigned_char = 8
493</pre>
494
495</div>
496
497<!-- ======================================================================= -->
498<div class="doc_subsubsection">
499 <a name="format_derived_type">Derived type descriptors</a>
500</div>
501
502<div class="doc_text">
503
504<pre>
505 %<a href="#format_derived_type">llvm.dbg.derivedtype.type</a> = type {
506 uint, ;; Tag (see below)
507 { }*, ;; Reference to context
508 sbyte*, ;; Name (may be "" for anonymous types)
509 { }*, ;; Reference to compile unit where defined (may be NULL)
Jim Laskey383e0092006-03-23 17:54:33 +0000510 uint, ;; Line number where defined (may be 0)
Jim Laskeycec12a52006-03-14 18:08:46 +0000511 uint, ;; Size in bits
512 uint, ;; Alignment in bits
513 uint, ;; Offset in bits
514 { }* ;; Reference to type derived from
515 }
516</pre>
517
518<p>These descriptors are used to define types derived from other types. The
519value of the tag varies depending on the meaning. The following are possible
520tag values;</p>
521
522<pre>
Jim Laskey3d11bee2006-03-15 19:10:52 +0000523 DW_TAG_formal_parameter = 5
Jim Laskeycec12a52006-03-14 18:08:46 +0000524 DW_TAG_member = 13
525 DW_TAG_pointer_type = 15
526 DW_TAG_reference_type = 16
527 DW_TAG_typedef = 22
528 DW_TAG_const_type = 38
529 DW_TAG_volatile_type = 53
530 DW_TAG_restrict_type = 55
531</pre>
532
533<p> <tt>DW_TAG_member</tt> is used to define a member of a <a
Jim Laskey3d11bee2006-03-15 19:10:52 +0000534href="#format_composite_type">composite type</a> or <a
535href="#format_subprograms">subprogram</a>. The type of the member is the <a
536href="#format_derived_type">derived type</a>. <tt>DW_TAG_formal_parameter</tt>
537is used to define a member which is a formal argument of a subprogram.</p>
Jim Laskeycec12a52006-03-14 18:08:46 +0000538
539<p><tt>DW_TAG_typedef</tt> is used to
540provide a name for the derived type.</p>
541
542<p><tt>DW_TAG_pointer_type</tt>,
543<tt>DW_TAG_reference_type</tt>, <tt>DW_TAG_const_type</tt>,
544<tt>DW_TAG_volatile_type</tt> and <tt>DW_TAG_restrict_type</tt> are used to
545qualify the <a href="#format_derived_type">derived type</a>. </p>
546
547<p><a href="#format_derived_type">Derived type</a> location can be determined
548from the compile unit and line number. The size, alignment and offset are
549expressed in bits and can be 64 bit values. The alignment is used to round the
550offset when embedded in a <a href="#format_composite_type">composite type</a>
551(example to keep float doubles on 64 bit boundaries.) The offset is the bit
552offset if embedded in a <a href="#format_composite_type">composite
553type</a>.</p>
554
555<p>Note that the <tt>void *</tt> type is expressed as a
556<tt>llvm.dbg.derivedtype.type</tt> with tag of <tt>DW_TAG_pointer_type</tt> and
557NULL derived type.</p>
558
559</div>
560
561<!-- ======================================================================= -->
562<div class="doc_subsubsection">
563 <a name="format_composite_type">Composite type descriptors</a>
564</div>
565
566<div class="doc_text">
567
568<pre>
569 %<a href="#format_composite_type">llvm.dbg.compositetype.type</a> = type {
570 uint, ;; Tag (see below)
571 { }*, ;; Reference to context
572 sbyte*, ;; Name (may be "" for anonymous types)
573 { }*, ;; Reference to compile unit where defined (may be NULL)
Jim Laskey383e0092006-03-23 17:54:33 +0000574 uint, ;; Line number where defined (may be 0)
Jim Laskeycec12a52006-03-14 18:08:46 +0000575 uint, ;; Size in bits
576 uint, ;; Alignment in bits
577 uint, ;; Offset in bits
578 { }* ;; Reference to array of member descriptors
579 }
580</pre>
581
582<p>These descriptors are used to define types that are composed of 0 or more
583elements. The value of the tag varies depending on the meaning. The following
584are possible tag values;</p>
585
586<pre>
587 DW_TAG_array_type = 1
588 DW_TAG_enumeration_type = 4
589 DW_TAG_structure_type = 19
590 DW_TAG_union_type = 23
591</pre>
592
593<p>The members of array types (tag = <tt>DW_TAG_array_type</tt>) are <a
594href="#format_subrange">subrange descriptors</a>, each representing the range of
595subscripts at that level of indexing.</p>
596
597<p>The members of enumeration types (tag = <tt>DW_TAG_enumeration_type</tt>) are
598<a href="#format_enumeration">enumerator descriptors</a>, each representing the
599definition of enumeration value
600for the set.</p>
601
602<p>The members of structure (tag = <tt>DW_TAG_structure_type</tt>) or union (tag
603= <tt>DW_TAG_union_type</tt>) types are any one of the <a
604href="#format_basic_type">basic</a>, <a href="#format_derived_type">derived</a>
605or <a href="#format_composite_type">composite</a> type descriptors, each
606representing a field member of the structure or union.</p>
607
608<p><a href="#format_composite_type">Composite type</a> location can be
609determined from the compile unit and line number. The size, alignment and
610offset are expressed in bits and can be 64 bit values. The alignment is used to
611round the offset when embedded in a <a href="#format_composite_type">composite
612type</a> (as an example, to keep float doubles on 64 bit boundaries.) The offset
613is the bit offset if embedded in a <a href="#format_composite_type">composite
614type</a>.</p>
615
616</div>
617
618<!-- ======================================================================= -->
619<div class="doc_subsubsection">
620 <a name="format_subrange">Subrange descriptors</a>
621</div>
622
623<div class="doc_text">
624
625<pre>
626 %<a href="#format_subrange">llvm.dbg.subrange.type</a> = type {
627 uint, ;; Tag = 33 (DW_TAG_subrange_type)
628 uint, ;; Low value
629 uint ;; High value
630 }
631</pre>
632
633<p>These descriptors are used to define ranges of array subscripts for an array
634<a href="#format_composite_type">composite type</a>. The low value defines the
635lower bounds typically zero for C/C++. The high value is the upper bounds.
636Values are 64 bit. High - low + 1 is the size of the array. If
637low == high the array will be unbounded.</p>
638
639</div>
640
641<!-- ======================================================================= -->
642<div class="doc_subsubsection">
643 <a name="format_enumeration">Enumerator descriptors</a>
644</div>
645
646<div class="doc_text">
647
648<pre>
649 %<a href="#format_enumeration">llvm.dbg.enumerator.type</a> = type {
650 uint, ;; Tag = 40 (DW_TAG_enumerator)
651 sbyte*, ;; Name
652 uint ;; Value
653 }
654</pre>
655
656<p>These descriptors are used to define members of an enumeration <a
657href="#format_composite_type">composite type</a>, it associates the name to the
658value.</p>
659
660</div>
661
662<!-- ======================================================================= -->
Jim Laskey383e0092006-03-23 17:54:33 +0000663<div class="doc_subsubsection">
664 <a name="format_variables">Local variables</a>
665</div>
666
667<div class="doc_text">
668<pre>
669 %<a href="#format_variables">llvm.dbg.variable.type</a> = type {
670 uint, ;; Tag (see below)
671 { }*, ;; Context
672 sbyte*, ;; Name
673 { }*, ;; Reference to compile unit where defined
674 uint, ;; Line number where defined
675 { }* ;; Type descriptor
676 }
677</pre>
678
679<p>These descriptors are used to define variables local to a sub program. The
680value of the tag depends on the usage of the variable;</p>
681
682<pre>
683 DW_TAG_auto_variable = 256
684 DW_TAG_arg_variable = 257
685 DW_TAG_return_variable = 258
686</pre>
687
688<p>An auto variable is any variable declared in the body of the function. An
689argument variable is any variable that appears as a formal argument to the
690function. A return variable is used to track the result of a function and has
691no source correspondent.</p>
692
693<p>The context is eitehr the subprogram or block where the variable is defined.
694Name the source variable name. Compile unit and line indicate where the
695variable was defined. Type descriptor defines the declared type of the
696variable.</p>
697
698</div>
699
700<!-- ======================================================================= -->
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000701<div class="doc_subsection">
Jim Laskeycec12a52006-03-14 18:08:46 +0000702 <a name="format_common_intrinsics">Debugger intrinsic functions</a>
703</div>
704
705<div class="doc_text">
706
707<p>LLVM uses several intrinsic functions (name prefixed with "llvm.dbg") to
708provide debug information at various points in generated code.</p>
709
710</div>
711
712<!-- ======================================================================= -->
713<div class="doc_subsubsection">
714 <a name="format_common_stoppoint">llvm.dbg.stoppoint</a>
715</div>
716
717<div class="doc_text">
718<pre>
719 void %<a href="#format_common_stoppoint">llvm.dbg.stoppoint</a>( uint, uint, %<a href="#format_compile_units">llvm.dbg.compile_unit</a>* )
720</pre>
721
722<p>This intrinsic is used to provide correspondence between the source file and
723the generated code. The first argument is the line number (base 1), second
724argument si the column number (0 if unknown) and the third argument the source
725compile unit. Code following a call to this intrinsic will have been defined in
726close proximity of the line, column and file. This information holds until the
Jim Laskey383e0092006-03-23 17:54:33 +0000727next call to <tt>%<a
728href="#format_common_stoppoint">lvm.dbg.stoppoint</a></tt>.</p>
Jim Laskeycec12a52006-03-14 18:08:46 +0000729
730</div>
731
732<!-- ======================================================================= -->
733<div class="doc_subsubsection">
734 <a name="format_common_func_start">llvm.dbg.func.start</a>
735</div>
736
737<div class="doc_text">
738<pre>
Jim Laskey383e0092006-03-23 17:54:33 +0000739 void %<a href="#format_common_func_start">llvm.dbg.func.start</a>( { }* )
Jim Laskeycec12a52006-03-14 18:08:46 +0000740</pre>
741
742<p>This intrinsic is used to link the debug information in <tt>%<a
743href="#format_subprograms">llvm.dbg.subprogram</a></tt> to the function. It also
744defines the beginning of the function's declarative region (scope.) The
745intrinsic should be called early in the function after the all the alloca
Jim Laskey383e0092006-03-23 17:54:33 +0000746instructions. It should be paired off with a closing <tt>%<a
747href="#format_common_region_end">llvm.dbg.region.end</a></tt>. The function's
748single argument is the <tt>%<a
749href="#format_subprograms">llvm.dbg.subprogram.type</a></tt>.</p>
Jim Laskeycec12a52006-03-14 18:08:46 +0000750
751</div>
752
753<!-- ======================================================================= -->
754<div class="doc_subsubsection">
755 <a name="format_common_region_start">llvm.dbg.region.start</a>
756</div>
757
758<div class="doc_text">
759<pre>
Jim Laskey383e0092006-03-23 17:54:33 +0000760 void %<a href="#format_common_region_start">llvm.dbg.region.start</a>( { }* )
Jim Laskeycec12a52006-03-14 18:08:46 +0000761</pre>
762
763<p>This intrinsic is used to define the beginning of a declarative scope (ex.
764block) for local language elements. It should be paired off with a closing
Jim Laskey383e0092006-03-23 17:54:33 +0000765<tt>%<a href="#format_common_region_end">llvm.dbg.region.end</a></tt>. The
766function's single argument is the <tt>%<a
767href="#format_blocks">llvm.dbg.block</a></tt> which is starting.</p>
768
Jim Laskeycec12a52006-03-14 18:08:46 +0000769
770</div>
771
772<!-- ======================================================================= -->
773<div class="doc_subsubsection">
774 <a name="format_common_region_end">llvm.dbg.region.end</a>
775</div>
776
777<div class="doc_text">
778<pre>
Jim Laskey383e0092006-03-23 17:54:33 +0000779 void %<a href="#format_common_region_end">llvm.dbg.region.end</a>( { }* )
Jim Laskeycec12a52006-03-14 18:08:46 +0000780</pre>
781
782<p>This intrinsic is used to define the end of a declarative scope (ex. block)
783for local language elements. It should be paired off with an opening <tt>%<a
784href="#format_common_region_start">llvm.dbg.region.start</a></tt> or <tt>%<a
Jim Laskey383e0092006-03-23 17:54:33 +0000785href="#format_common_func_start">llvm.dbg.func.start</a></tt>. The function's
786single argument is either the <tt>%<a
787href="#format_blocks">llvm.dbg.block</a></tt> or the <tt>%<a
788href="#format_subprograms">llvm.dbg.subprogram.type</a></tt> which is
789ending.</p>
Jim Laskeycec12a52006-03-14 18:08:46 +0000790
791</div>
792
793<!-- ======================================================================= -->
794<div class="doc_subsubsection">
795 <a name="format_common_declare">llvm.dbg.declare</a>
796</div>
797
798<div class="doc_text">
799<pre>
Jim Laskey383e0092006-03-23 17:54:33 +0000800 void %<a href="#format_common_declare">llvm.dbg.declare</a>( { } *, { }* )
Jim Laskeycec12a52006-03-14 18:08:46 +0000801</pre>
802
803<p>This intrinsic provides information about a local element (ex. variable.)
Jim Laskey383e0092006-03-23 17:54:33 +0000804The first argument as a AllocA for the variable cast to a <tt>{ }*</tt>. The second
805argument is the <tt>%<a href="#format_variables">llvm.dbg.variable</a></tt>
806containing the description of the variable, also cast to a <tt>{ }*</tt>.</p>
Jim Laskeycec12a52006-03-14 18:08:46 +0000807
808</div>
809
810<!-- ======================================================================= -->
811<div class="doc_subsection">
812 <a name="format_common_stoppoints">
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000813 Representing stopping points in the source program
814 </a>
815</div>
816
817<div class="doc_text">
818
819<p>LLVM debugger "stop points" are a key part of the debugging representation
820that allows the LLVM to maintain simple semantics for <a
821href="#debugopt">debugging optimized code</a>. The basic idea is that the
Jim Laskeycec12a52006-03-14 18:08:46 +0000822front-end inserts calls to the <a
823href="#format_common_stoppoint">%<tt>llvm.dbg.stoppoint</tt></a> intrinsic
Jim Laskey3d11bee2006-03-15 19:10:52 +0000824function at every point in the program where a debugger should be able to
825inspect the program (these correspond to places a debugger stops when you
Jim Laskeycec12a52006-03-14 18:08:46 +0000826"<tt>step</tt>" through it). The front-end can choose to place these as
827fine-grained as it would like (for example, before every subexpression
828evaluated), but it is recommended to only put them after every source statement
829that includes executable code.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000830
Misha Brukman82873732004-05-12 19:21:57 +0000831<p>Using calls to this intrinsic function to demark legal points for the
832debugger to inspect the program automatically disables any optimizations that
833could potentially confuse debugging information. To non-debug-information-aware
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000834transformations, these calls simply look like calls to an external function,
835which they must assume to do anything (including reading or writing to any part
836of reachable memory). On the other hand, it does not impact many optimizations,
837such as code motion of non-trapping instructions, nor does it impact
Chris Lattner8ff75902004-01-06 05:31:32 +0000838optimization of subexpressions, code duplication transformations, or basic-block
839reordering transformations.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000840
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000841</div>
842
843
844<!-- ======================================================================= -->
845<div class="doc_subsection">
Chris Lattner8ff75902004-01-06 05:31:32 +0000846 <a name="format_common_lifetime">Object lifetimes and scoping</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000847</div>
848
849<div class="doc_text">
Misha Brukman82873732004-05-12 19:21:57 +0000850<p>In many languages, the local variables in functions can have their lifetime
851or scope limited to a subset of a function. In the C family of languages, for
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000852example, variables are only live (readable and writable) within the source block
853that they are defined in. In functional languages, values are only readable
854after they have been defined. Though this is a very obvious concept, it is also
855non-trivial to model in LLVM, because it has no notion of scoping in this sense,
Misha Brukman82873732004-05-12 19:21:57 +0000856and does not want to be tied to a language's scoping rules.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000857
Misha Brukman82873732004-05-12 19:21:57 +0000858<p>In order to handle this, the LLVM debug format uses the notion of "regions"
859of a function, delineated by calls to intrinsic functions. These intrinsic
860functions define new regions of the program and indicate when the region
861lifetime expires. Consider the following C fragment, for example:</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000862
Misha Brukman82873732004-05-12 19:21:57 +0000863<pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +00008641. void foo() {
8652. int X = ...;
8663. int Y = ...;
8674. {
8685. int Z = ...;
8696. ...
8707. }
8718. ...
8729. }
Misha Brukman82873732004-05-12 19:21:57 +0000873</pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000874
Jim Laskeycec12a52006-03-14 18:08:46 +0000875<p>Compiled to LLVM, this function would be represented like this:</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000876
Misha Brukman82873732004-05-12 19:21:57 +0000877<pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000878void %foo() {
Jim Laskeycec12a52006-03-14 18:08:46 +0000879entry:
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000880 %X = alloca int
881 %Y = alloca int
882 %Z = alloca int
Jim Laskeycec12a52006-03-14 18:08:46 +0000883
884 ...
885
886 call void %<a href="#format_common_func_start">llvm.dbg.func.start</a>( %<a href="#format_subprograms">llvm.dbg.subprogram.type</a>* %llvm.dbg.subprogram )
887
888 call void %<a href="#format_common_stoppoint">llvm.dbg.stoppoint</a>( uint 2, uint 2, %<a href="#format_compile_units">llvm.dbg.compile_unit</a>* %llvm.dbg.compile_unit )
889
890 call void %<a href="#format_common_declare">llvm.dbg.declare</a>({}* %X, ...)
891 call void %<a href="#format_common_declare">llvm.dbg.declare</a>({}* %Y, ...)
892
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000893 <i>;; Evaluate expression on line 2, assigning to X.</i>
Jim Laskeycec12a52006-03-14 18:08:46 +0000894
895 call void %<a href="#format_common_stoppoint">llvm.dbg.stoppoint</a>( uint 3, uint 2, %<a href="#format_compile_units">llvm.dbg.compile_unit</a>* %llvm.dbg.compile_unit )
896
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000897 <i>;; Evaluate expression on line 3, assigning to Y.</i>
Jim Laskeycec12a52006-03-14 18:08:46 +0000898
899 call void %<a href="#format_common_stoppoint">llvm.region.start</a>()
900 call void %<a href="#format_common_stoppoint">llvm.dbg.stoppoint</a>( uint 5, uint 4, %<a href="#format_compile_units">llvm.dbg.compile_unit</a>* %llvm.dbg.compile_unit )
901 call void %<a href="#format_common_declare">llvm.dbg.declare</a>({}* %X, ...)
902
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000903 <i>;; Evaluate expression on line 5, assigning to Z.</i>
Jim Laskeycec12a52006-03-14 18:08:46 +0000904
905 call void %<a href="#format_common_stoppoint">llvm.dbg.stoppoint</a>( uint 7, uint 2, %<a href="#format_compile_units">llvm.dbg.compile_unit</a>* %llvm.dbg.compile_unit )
906 call void %<a href="#format_common_region_end">llvm.region.end</a>()
907
908 call void %<a href="#format_common_stoppoint">llvm.dbg.stoppoint</a>( uint 9, uint 2, %<a href="#format_compile_units">llvm.dbg.compile_unit</a>* %llvm.dbg.compile_unit )
909
910 call void %<a href="#format_common_region_end">llvm.region.end</a>()
911
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000912 ret void
913}
Misha Brukman82873732004-05-12 19:21:57 +0000914</pre>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000915
Misha Brukman82873732004-05-12 19:21:57 +0000916<p>This example illustrates a few important details about the LLVM debugging
Jim Laskeycec12a52006-03-14 18:08:46 +0000917information. In particular, it shows how the various intrinsics are applied
918together to allow a debugger to analyze the relationship between statements,
919variable definitions, and the code used to implement the function.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000920
Jim Laskeycec12a52006-03-14 18:08:46 +0000921<p>The first intrinsic <tt>%<a
922href="#format_common_func_start">llvm.dbg.func.start</a></tt> provides
923a link with the <a href="#format_subprograms">subprogram descriptor</a>
924containing the details of this function. This call also defines the beginning
925of the function region, bounded by the <tt>%<a
926href="#format_common_region_end">llvm.region.end</a></tt> at the end of
927the function. This region is used to bracket the lifetime of variables declared
928within. For a function, this outer region defines a new stack frame whose
929lifetime ends when the region is ended.</p>
930
931<p>It is possible to define inner regions for short term variables by using the
932%<a href="#format_common_stoppoint"><tt>llvm.region.start</tt></a> and <a
933href="#format_common_region_end"><tt>%llvm.region.end</tt></a> to bound a
934region. The inner region in this example would be for the block containing the
935declaration of Z.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000936
Misha Brukman82873732004-05-12 19:21:57 +0000937<p>Using regions to represent the boundaries of source-level functions allow
938LLVM interprocedural optimizations to arbitrarily modify LLVM functions without
Chris Lattner8ff75902004-01-06 05:31:32 +0000939having to worry about breaking mapping information between the LLVM code and the
940and source-level program. In particular, the inliner requires no modification
941to support inlining with debugging information: there is no explicit correlation
942drawn between LLVM functions and their source-level counterparts (note however,
943that if the inliner inlines all instances of a non-strong-linkage function into
944its caller that it will not be possible for the user to manually invoke the
Jim Laskey3d11bee2006-03-15 19:10:52 +0000945inlined function from a debugger).</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000946
Misha Brukman82873732004-05-12 19:21:57 +0000947<p>Once the function has been defined, the <a
Jim Laskeycec12a52006-03-14 18:08:46 +0000948href="#format_common_stoppoint"><tt>stopping point</tt></a> corresponding to
949line #2 (column #2) of the function is encountered. At this point in the
950function, <b>no</b> local variables are live. As lines 2 and 3 of the example
951are executed, their variable definitions are introduced into the program using
952%<a href="#format_common_declare"><tt>llvm.dbg.declare</tt></a>, without the
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000953need to specify a new region. These variables do not require new regions to be
954introduced because they go out of scope at the same point in the program: line
Misha Brukman82873732004-05-12 19:21:57 +00009559.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000956
Misha Brukman82873732004-05-12 19:21:57 +0000957<p>In contrast, the <tt>Z</tt> variable goes out of scope at a different time,
Jim Laskeycec12a52006-03-14 18:08:46 +0000958on line 7. For this reason, it is defined within the inner region, which kills
959the availability of <tt>Z</tt> before the code for line 8 is executed. In this
960way, regions can support arbitrary source-language scoping rules, as long as
961they can only be nested (ie, one scope cannot partially overlap with a part of
962another scope).</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000963
Misha Brukman82873732004-05-12 19:21:57 +0000964<p>It is worth noting that this scoping mechanism is used to control scoping of
965all declarations, not just variable declarations. For example, the scope of a
Jim Laskeycec12a52006-03-14 18:08:46 +0000966C++ using declaration is controlled with this couldchange how name lookup is
967performed.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000968
969</div>
970
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000971
972
973<!-- *********************************************************************** -->
974<div class="doc_section">
Chris Lattner8ff75902004-01-06 05:31:32 +0000975 <a name="ccxx_frontend">C/C++ front-end specific debug information</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000976</div>
Misha Brukman94218a72004-12-09 20:27:37 +0000977<!-- *********************************************************************** -->
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000978
979<div class="doc_text">
980
Misha Brukman82873732004-05-12 19:21:57 +0000981<p>The C and C++ front-ends represent information about the program in a format
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000982that is effectively identical to <a
983href="http://www.eagercon.com/dwarf/dwarf3std.htm">Dwarf 3.0</a> in terms of
984information content. This allows code generators to trivially support native
985debuggers by generating standard dwarf information, and contains enough
Chris Lattner8ff75902004-01-06 05:31:32 +0000986information for non-dwarf targets to translate it as needed.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000987
Jim Laskeycec12a52006-03-14 18:08:46 +0000988<p>This section describes the forms used to represent C and C++ programs. Other
989languages could pattern themselves after this (which itself is tuned to
Chris Lattner8ff75902004-01-06 05:31:32 +0000990representing programs in the same way that Dwarf 3 does), or they could choose
Jim Laskeycec12a52006-03-14 18:08:46 +0000991to provide completely different forms if they don't fit into the Dwarf model.
992As support for debugging information gets added to the various LLVM
Misha Brukman82873732004-05-12 19:21:57 +0000993source-language front-ends, the information used should be documented here.</p>
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000994
Jim Laskeycec12a52006-03-14 18:08:46 +0000995<p>The following sections provide examples of various C/C++ constructs and the
996debug information that would best describe those constructs.</p>
997
Chris Lattnerbdfb3392004-01-05 05:06:33 +0000998</div>
999
1000<!-- ======================================================================= -->
1001<div class="doc_subsection">
Jim Laskeycec12a52006-03-14 18:08:46 +00001002 <a name="ccxx_compile_units">C/C++ source file information</a>
Chris Lattnerbdfb3392004-01-05 05:06:33 +00001003</div>
1004
1005<div class="doc_text">
Chris Lattnerbdfb3392004-01-05 05:06:33 +00001006
Jim Laskeycec12a52006-03-14 18:08:46 +00001007<p>Given the source files "MySource.cpp" and "MyHeader.h" located in the
1008directory "/Users/mine/sources", the following code;</p>
Chris Lattner8ff75902004-01-06 05:31:32 +00001009
Jim Laskeycec12a52006-03-14 18:08:46 +00001010<pre>
1011#include "MyHeader.h"
Chris Lattner8ff75902004-01-06 05:31:32 +00001012
Jim Laskeycec12a52006-03-14 18:08:46 +00001013int main(int argc, char *argv[]) {
1014 return 0;
1015}
1016</pre>
Chris Lattner8ff75902004-01-06 05:31:32 +00001017
Jim Laskeycec12a52006-03-14 18:08:46 +00001018<p>a C/C++ front-end would generate the following descriptors;</p>
1019
1020<pre>
1021...
1022;;
1023;; Define types used. In this case we need one for compile unit anchors and one
1024;; for compile units.
1025;;
1026%<a href="#format_anchors">llvm.dbg.anchor.type</a> = type { uint, uint }
1027%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a> = type { uint, { }*, uint, uint, sbyte*, sbyte*, sbyte* }
1028...
1029;;
1030;; Define the anchor for compile units. Note that the second field of the
1031;; anchor is 17, which is the same as the tag for compile units
1032;; (17 = DW_TAG_compile_unit.)
1033;;
1034%<a href="#format_compile_units">llvm.dbg.compile_units</a> = linkonce constant %<a href="#format_anchors">llvm.dbg.anchor.type</a> { uint 0, uint 17 }, section "llvm.metadata"
1035
1036;;
1037;; Define the compile unit for the source file "/Users/mine/sources/MySource.cpp".
1038;;
1039%<a href="#format_compile_units">llvm.dbg.compile_unit1</a> = internal constant %<a href="#format_compile_units">llvm.dbg.compile_unit.type</a> {
1040 uint 17,
1041 { }* cast (%<a href="#format_anchors">llvm.dbg.anchor.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_units</a> to { }*),
1042 uint 1,
1043 uint 1,
1044 sbyte* getelementptr ([13 x sbyte]* %str1, int 0, int 0),
1045 sbyte* getelementptr ([21 x sbyte]* %str2, int 0, int 0),
1046 sbyte* getelementptr ([33 x sbyte]* %str3, int 0, int 0) }, section "llvm.metadata"
1047
1048;;
1049;; Define the compile unit for the header file "/Users/mine/sources/MyHeader.h".
1050;;
1051%<a href="#format_compile_units">llvm.dbg.compile_unit2</a> = internal constant %<a href="#format_compile_units">llvm.dbg.compile_unit.type</a> {
1052 uint 17,
1053 { }* cast (%<a href="#format_anchors">llvm.dbg.anchor.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_units</a> to { }*),
1054 uint 1,
1055 uint 1,
1056 sbyte* getelementptr ([11 x sbyte]* %str4, int 0, int 0),
1057 sbyte* getelementptr ([21 x sbyte]* %str2, int 0, int 0),
1058 sbyte* getelementptr ([33 x sbyte]* %str3, int 0, int 0) }, section "llvm.metadata"
1059
1060;;
1061;; Define each of the strings used in the compile units.
1062;;
1063%str1 = internal constant [13 x sbyte] c"MySource.cpp\00", section "llvm.metadata";
1064%str2 = internal constant [21 x sbyte] c"/Users/mine/sources/\00", section "llvm.metadata";
1065%str3 = internal constant [33 x sbyte] c"4.0.1 LLVM (LLVM research group)\00", section "llvm.metadata";
1066%str4 = internal constant [11 x sbyte] c"MyHeader.h\00", section "llvm.metadata";
1067...
1068</pre>
1069
Chris Lattner8ff75902004-01-06 05:31:32 +00001070</div>
1071
1072<!-- ======================================================================= -->
1073<div class="doc_subsection">
Jim Laskeycec12a52006-03-14 18:08:46 +00001074 <a name="ccxx_global_variable">C/C++ global variable information</a>
Chris Lattner8ff75902004-01-06 05:31:32 +00001075</div>
1076
1077<div class="doc_text">
Jim Laskeycec12a52006-03-14 18:08:46 +00001078
1079<p>Given an integer global variable declared as follows;</p>
1080
1081<pre>
1082int MyGlobal = 100;
1083</pre>
1084
1085<p>a C/C++ front-end would generate the following descriptors;</p>
1086
1087<pre>
1088;;
1089;; Define types used. One for global variable anchors, one for the global
1090;; variable descriptor, one for the global's basic type and one for the global's
1091;; compile unit.
1092;;
1093%<a href="#format_anchors">llvm.dbg.anchor.type</a> = type { uint, uint }
Jim Laskey3d11bee2006-03-15 19:10:52 +00001094%<a href="#format_global_variables">llvm.dbg.global_variable.type</a> = type { uint, { }*, { }*, sbyte*, { }*, uint, { }*, bool, bool, { }*, uint }
Jim Laskeycec12a52006-03-14 18:08:46 +00001095%<a href="#format_basic_type">llvm.dbg.basictype.type</a> = type { uint, { }*, sbyte*, { }*, int, uint, uint, uint, uint }
1096%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a> = ...
1097...
1098;;
1099;; Define the global itself.
1100;;
1101%MyGlobal = global int 100
1102...
1103;;
1104;; Define the anchor for global variables. Note that the second field of the
1105;; anchor is 52, which is the same as the tag for global variables
1106;; (52 = DW_TAG_variable.)
1107;;
1108%<a href="#format_global_variables">llvm.dbg.global_variables</a> = linkonce constant %<a href="#format_anchors">llvm.dbg.anchor.type</a> { uint 0, uint 52 }, section "llvm.metadata"
1109
1110;;
1111;; Define the global variable descriptor. Note the reference to the global
1112;; variable anchor and the global variable itself.
1113;;
1114%<a href="#format_global_variables">llvm.dbg.global_variable</a> = internal constant %<a href="#format_global_variables">llvm.dbg.global_variable.type</a> {
1115 uint 52,
1116 { }* cast (%<a href="#format_anchors">llvm.dbg.anchor.type</a>* %<a href="#format_global_variables">llvm.dbg.global_variables</a> to { }*),
1117 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1118 sbyte* getelementptr ([9 x sbyte]* %str1, int 0, int 0),
Jim Laskey3d11bee2006-03-15 19:10:52 +00001119 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1120 uint 1,
Jim Laskeycec12a52006-03-14 18:08:46 +00001121 { }* cast (%<a href="#format_basic_type">llvm.dbg.basictype.type</a>* %<a href="#format_basic_type">llvm.dbg.basictype</a> to { }*),
1122 bool false,
1123 bool true,
Jim Laskey3d11bee2006-03-15 19:10:52 +00001124 { }* cast (int* %MyGlobal to { }*) }, section "llvm.metadata"
Jim Laskeycec12a52006-03-14 18:08:46 +00001125
1126;;
1127;; Define the basic type of 32 bit signed integer. Note that since int is an
1128;; intrinsic type the source file is NULL and line 0.
1129;;
1130%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1131 uint 36,
1132 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1133 sbyte* getelementptr ([4 x sbyte]* %str2, int 0, int 0),
1134 { }* null,
1135 int 0,
1136 uint 32,
1137 uint 32,
1138 uint 0,
1139 uint 5 }, section "llvm.metadata"
1140
1141;;
1142;; Define the names of the global variable and basic type.
1143;;
1144%str1 = internal constant [9 x sbyte] c"MyGlobal\00", section "llvm.metadata"
1145%str2 = internal constant [4 x sbyte] c"int\00", section "llvm.metadata"
1146</pre>
1147
Chris Lattner8ff75902004-01-06 05:31:32 +00001148</div>
Chris Lattnerbdfb3392004-01-05 05:06:33 +00001149
Jim Laskeycec12a52006-03-14 18:08:46 +00001150<!-- ======================================================================= -->
1151<div class="doc_subsection">
1152 <a name="ccxx_subprogram">C/C++ function information</a>
1153</div>
1154
1155<div class="doc_text">
1156
1157<p>Given a function declared as follows;</p>
1158
1159<pre>
1160int main(int argc, char *argv[]) {
1161 return 0;
1162}
1163</pre>
1164
1165<p>a C/C++ front-end would generate the following descriptors;</p>
1166
1167<pre>
1168;;
1169;; Define types used. One for subprogram anchors, one for the subprogram
1170;; descriptor, one for the global's basic type and one for the subprogram's
1171;; compile unit.
1172;;
Jim Laskey383e0092006-03-23 17:54:33 +00001173%<a href="#format_subprograms">llvm.dbg.subprogram.type</a> = type { uint, { }*, { }*, sbyte*, { }*, bool, bool }
Jim Laskeycec12a52006-03-14 18:08:46 +00001174%<a href="#format_anchors">llvm.dbg.anchor.type</a> = type { uint, uint }
1175%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a> = ...
1176
1177;;
1178;; Define the anchor for subprograms. Note that the second field of the
1179;; anchor is 46, which is the same as the tag for subprograms
1180;; (46 = DW_TAG_subprogram.)
1181;;
1182%<a href="#format_subprograms">llvm.dbg.subprograms</a> = linkonce constant %<a href="#format_anchors">llvm.dbg.anchor.type</a> { uint 0, uint 46 }, section "llvm.metadata"
1183
1184;;
1185;; Define the descriptor for the subprogram. TODO - more details.
1186;;
1187%<a href="#format_subprograms">llvm.dbg.subprogram</a> = internal constant %<a href="#format_subprograms">llvm.dbg.subprogram.type</a> {
1188 uint 46,
1189 { }* cast (%<a href="#format_anchors">llvm.dbg.anchor.type</a>* %<a href="#format_subprograms">llvm.dbg.subprograms</a> to { }*),
1190 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1191 sbyte* getelementptr ([5 x sbyte]* %str1, int 0, int 0),
Jim Laskey3d11bee2006-03-15 19:10:52 +00001192 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1193 uint 1,
Jim Laskeycec12a52006-03-14 18:08:46 +00001194 { }* null,
1195 bool false,
Jim Laskey383e0092006-03-23 17:54:33 +00001196 bool true }, section "llvm.metadata"
Jim Laskeycec12a52006-03-14 18:08:46 +00001197
1198;;
1199;; Define the name of the subprogram.
1200;;
1201%str1 = internal constant [5 x sbyte] c"main\00", section "llvm.metadata"
1202
1203;;
1204;; Define the subprogram itself.
1205;;
1206int %main(int %argc, sbyte** %argv) {
1207...
1208}
1209</pre>
1210
1211</div>
1212
1213<!-- ======================================================================= -->
1214<div class="doc_subsection">
1215 <a name="ccxx_basic_types">C/C++ basic types</a>
1216</div>
1217
1218<div class="doc_text">
1219
1220<p>The following are the basic type descriptors for C/C++ core types;</p>
1221
1222</div>
1223
1224<!-- ======================================================================= -->
1225<div class="doc_subsubsection">
1226 <a name="ccxx_basic_type_bool">bool</a>
1227</div>
1228
1229<div class="doc_text">
1230
1231<pre>
1232%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1233 uint 36,
1234 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1235 sbyte* getelementptr ([5 x sbyte]* %str1, int 0, int 0),
1236 { }* null,
1237 int 0,
1238 uint 32,
1239 uint 32,
1240 uint 0,
1241 uint 2 }, section "llvm.metadata"
1242%str1 = internal constant [5 x sbyte] c"bool\00", section "llvm.metadata"
1243</pre>
1244
1245</div>
1246
1247<!-- ======================================================================= -->
1248<div class="doc_subsubsection">
1249 <a name="ccxx_basic_char">char</a>
1250</div>
1251
1252<div class="doc_text">
1253
1254<pre>
1255%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1256 uint 36,
1257 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1258 sbyte* getelementptr ([5 x sbyte]* %str1, int 0, int 0),
1259 { }* null,
1260 int 0,
1261 uint 8,
1262 uint 8,
1263 uint 0,
1264 uint 6 }, section "llvm.metadata"
1265%str1 = internal constant [5 x sbyte] c"char\00", section "llvm.metadata"
1266</pre>
1267
1268</div>
1269
1270<!-- ======================================================================= -->
1271<div class="doc_subsubsection">
1272 <a name="ccxx_basic_unsigned_char">unsigned char</a>
1273</div>
1274
1275<div class="doc_text">
1276
1277<pre>
1278%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1279 uint 36,
1280 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1281 sbyte* getelementptr ([14 x sbyte]* %str1, int 0, int 0),
1282 { }* null,
1283 int 0,
1284 uint 8,
1285 uint 8,
1286 uint 0,
1287 uint 8 }, section "llvm.metadata"
1288%str1 = internal constant [14 x sbyte] c"unsigned char\00", section "llvm.metadata"
1289</pre>
1290
1291</div>
1292
1293<!-- ======================================================================= -->
1294<div class="doc_subsubsection">
1295 <a name="ccxx_basic_short">short</a>
1296</div>
1297
1298<div class="doc_text">
1299
1300<pre>
1301%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1302 uint 36,
1303 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1304 sbyte* getelementptr ([10 x sbyte]* %str1, int 0, int 0),
1305 { }* null,
1306 int 0,
1307 uint 16,
1308 uint 16,
1309 uint 0,
1310 uint 5 }, section "llvm.metadata"
1311%str1 = internal constant [10 x sbyte] c"short int\00", section "llvm.metadata"
1312</pre>
1313
1314</div>
1315
1316<!-- ======================================================================= -->
1317<div class="doc_subsubsection">
1318 <a name="ccxx_basic_unsigned_short">unsigned short</a>
1319</div>
1320
1321<div class="doc_text">
1322
1323<pre>
1324%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1325 uint 36,
1326 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1327 sbyte* getelementptr ([19 x sbyte]* %str1, int 0, int 0),
1328 { }* null,
1329 int 0,
1330 uint 16,
1331 uint 16,
1332 uint 0,
1333 uint 7 }, section "llvm.metadata"
1334%str1 = internal constant [19 x sbyte] c"short unsigned int\00", section "llvm.metadata"
1335</pre>
1336
1337</div>
1338
1339<!-- ======================================================================= -->
1340<div class="doc_subsubsection">
1341 <a name="ccxx_basic_int">int</a>
1342</div>
1343
1344<div class="doc_text">
1345
1346<pre>
1347%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1348 uint 36,
1349 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1350 sbyte* getelementptr ([4 x sbyte]* %str1, int 0, int 0),
1351 { }* null,
1352 int 0,
1353 uint 32,
1354 uint 32,
1355 uint 0,
1356 uint 5 }, section "llvm.metadata"
1357%str1 = internal constant [4 x sbyte] c"int\00", section "llvm.metadata"
1358</pre>
1359
1360</div>
1361
1362<!-- ======================================================================= -->
1363<div class="doc_subsubsection">
1364 <a name="ccxx_basic_unsigned_int">unsigned int</a>
1365</div>
1366
1367<div class="doc_text">
1368
1369<pre>
1370%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1371 uint 36,
1372 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1373 sbyte* getelementptr ([13 x sbyte]* %str1, int 0, int 0),
1374 { }* null,
1375 int 0,
1376 uint 32,
1377 uint 32,
1378 uint 0,
1379 uint 7 }, section "llvm.metadata"
1380%str1 = internal constant [13 x sbyte] c"unsigned int\00", section "llvm.metadata"
1381</pre>
1382
1383</div>
1384
1385<!-- ======================================================================= -->
1386<div class="doc_subsubsection">
1387 <a name="ccxx_basic_long_long">long long</a>
1388</div>
1389
1390<div class="doc_text">
1391
1392<pre>
1393%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1394 uint 36,
1395 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1396 sbyte* getelementptr ([14 x sbyte]* %str1, int 0, int 0),
1397 { }* null,
1398 int 0,
1399 uint 64,
1400 uint 64,
1401 uint 0,
1402 uint 5 }, section "llvm.metadata"
1403%str1 = internal constant [14 x sbyte] c"long long int\00", section "llvm.metadata"
1404</pre>
1405
1406</div>
1407
1408<!-- ======================================================================= -->
1409<div class="doc_subsubsection">
1410 <a name="ccxx_basic_unsigned_long_long">unsigned long long</a>
1411</div>
1412
1413<div class="doc_text">
1414
1415<pre>
1416%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1417 uint 36,
1418 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1419 sbyte* getelementptr ([23 x sbyte]* %str1, int 0, int 0),
1420 { }* null,
1421 int 0,
1422 uint 64,
1423 uint 64,
1424 uint 0,
1425 uint 7 }, section "llvm.metadata"
1426%str1 = internal constant [23 x sbyte] c"long long unsigned int\00", section "llvm.metadata"
1427</pre>
1428
1429</div>
1430
1431<!-- ======================================================================= -->
1432<div class="doc_subsubsection">
1433 <a name="ccxx_basic_float">float</a>
1434</div>
1435
1436<div class="doc_text">
1437
1438<pre>
1439%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1440 uint 36,
1441 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1442 sbyte* getelementptr ([6 x sbyte]* %str1, int 0, int 0),
1443 { }* null,
1444 int 0,
1445 uint 32,
1446 uint 32,
1447 uint 0,
1448 uint 4 }, section "llvm.metadata"
1449%str1 = internal constant [6 x sbyte] c"float\00", section "llvm.metadata"
1450</pre>
1451
1452</div>
1453
1454<!-- ======================================================================= -->
1455<div class="doc_subsubsection">
1456 <a name="ccxx_basic_double">double</a>
1457</div>
1458
1459<div class="doc_text">
1460
1461<pre>
1462%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1463 uint 36,
1464 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1465 sbyte* getelementptr ([7 x sbyte]* %str1, int 0, int 0),
1466 { }* null,
1467 int 0,
1468 uint 64,
1469 uint 64,
1470 uint 0,
1471 uint 4 }, section "llvm.metadata"
1472%str1 = internal constant [7 x sbyte] c"double\00", section "llvm.metadata"
1473</pre>
1474
1475</div>
1476
1477<!-- ======================================================================= -->
1478<div class="doc_subsection">
1479 <a name="ccxx_derived_types">C/C++ derived types</a>
1480</div>
1481
1482<div class="doc_text">
1483
1484<p>Given the following as an example of C/C++ derived type;</p>
1485
1486<pre>
1487typedef const int *IntPtr;
1488</pre>
1489
1490<p>a C/C++ front-end would generate the following descriptors;</p>
1491
1492<pre>
1493;;
1494;; Define the typedef "IntPtr".
1495;;
1496%<a href="#format_derived_type">llvm.dbg.derivedtype1</a> = internal constant %<a href="#format_derived_type">llvm.dbg.derivedtype.type</a> {
1497 uint 22,
1498 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1499 sbyte* getelementptr ([7 x sbyte]* %str1, int 0, int 0),
1500 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1501 int 1,
1502 uint 0,
1503 uint 0,
1504 uint 0,
1505 { }* cast (%<a href="#format_derived_type">llvm.dbg.derivedtype.type</a>* %<a href="#format_derived_type">llvm.dbg.derivedtype2</a> to { }*) }, section "llvm.metadata"
1506%str1 = internal constant [7 x sbyte] c"IntPtr\00", section "llvm.metadata"
1507
1508;;
1509;; Define the pointer type.
1510;;
1511%<a href="#format_derived_type">llvm.dbg.derivedtype2</a> = internal constant %<a href="#format_derived_type">llvm.dbg.derivedtype.type</a> {
1512 uint 15,
1513 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
Jim Laskey4d9b10b2006-03-14 18:50:50 +00001514 sbyte* null,
Jim Laskeycec12a52006-03-14 18:08:46 +00001515 { }* null,
1516 int 0,
1517 uint 32,
1518 uint 32,
1519 uint 0,
1520 { }* cast (%<a href="#format_derived_type">llvm.dbg.derivedtype.type</a>* %<a href="#format_derived_type">llvm.dbg.derivedtype3</a> to { }*) }, section "llvm.metadata"
Jim Laskeycec12a52006-03-14 18:08:46 +00001521
1522;;
1523;; Define the const type.
1524;;
1525%<a href="#format_derived_type">llvm.dbg.derivedtype3</a> = internal constant %<a href="#format_derived_type">llvm.dbg.derivedtype.type</a> {
1526 uint 38,
1527 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
Jim Laskey4d9b10b2006-03-14 18:50:50 +00001528 sbyte* null,
Jim Laskeycec12a52006-03-14 18:08:46 +00001529 { }* null,
1530 int 0,
1531 uint 0,
1532 uint 0,
1533 uint 0,
1534 { }* cast (%<a href="#format_basic_type">llvm.dbg.basictype.type</a>* %<a href="#format_basic_type">llvm.dbg.basictype1</a> to { }*) }, section "llvm.metadata"
1535
1536;;
1537;; Define the int type.
1538;;
1539%<a href="#format_basic_type">llvm.dbg.basictype1</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1540 uint 36,
1541 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
Jim Laskey4d9b10b2006-03-14 18:50:50 +00001542 sbyte* getelementptr ([4 x sbyte]* %str2, int 0, int 0),
Jim Laskeycec12a52006-03-14 18:08:46 +00001543 { }* null,
1544 int 0,
1545 uint 32,
1546 uint 32,
1547 uint 0,
1548 uint 5 }, section "llvm.metadata"
Jim Laskey4d9b10b2006-03-14 18:50:50 +00001549%str2 = internal constant [4 x sbyte] c"int\00", section "llvm.metadata"
Jim Laskeycec12a52006-03-14 18:08:46 +00001550</pre>
1551
1552</div>
1553
1554<!-- ======================================================================= -->
1555<div class="doc_subsection">
1556 <a name="ccxx_composite_types">C/C++ struct/union types</a>
1557</div>
1558
1559<div class="doc_text">
1560
1561<p>Given the following as an example of C/C++ struct type;</p>
1562
1563<pre>
1564struct Color {
1565 unsigned Red;
1566 unsigned Green;
1567 unsigned Blue;
1568};
1569</pre>
1570
1571<p>a C/C++ front-end would generate the following descriptors;</p>
1572
1573<pre>
1574;;
1575;; Define basic type for unsigned int.
1576;;
1577%<a href="#format_basic_type">llvm.dbg.basictype</a> = internal constant %<a href="#format_basic_type">llvm.dbg.basictype.type</a> {
1578 uint 36,
1579 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1580 sbyte* getelementptr ([13 x sbyte]* %str1, int 0, int 0),
1581 { }* null,
1582 int 0,
1583 uint 32,
1584 uint 32,
1585 uint 0,
1586 uint 7 }, section "llvm.metadata"
1587%str1 = internal constant [13 x sbyte] c"unsigned int\00", section "llvm.metadata"
1588
1589;;
1590;; Define composite type for struct Color.
1591;;
1592%<a href="#format_composite_type">llvm.dbg.compositetype</a> = internal constant %<a href="#format_composite_type">llvm.dbg.compositetype.type</a> {
1593 uint 19,
1594 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1595 sbyte* getelementptr ([6 x sbyte]* %str2, int 0, int 0),
1596 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1597 int 1,
1598 uint 96,
1599 uint 32,
1600 uint 0,
1601 { }* null,
1602 { }* cast ([3 x { }*]* %llvm.dbg.array to { }*) }, section "llvm.metadata"
1603%str2 = internal constant [6 x sbyte] c"Color\00", section "llvm.metadata"
1604
1605;;
1606;; Define the Red field.
1607;;
1608%<a href="#format_derived_type">llvm.dbg.derivedtype1</a> = internal constant %<a href="#format_derived_type">llvm.dbg.derivedtype.type</a> {
1609 uint 13,
1610 { }* null,
1611 sbyte* getelementptr ([4 x sbyte]* %str3, int 0, int 0),
1612 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1613 int 2,
1614 uint 32,
1615 uint 32,
1616 uint 0,
1617 { }* cast (%<a href="#format_basic_type">llvm.dbg.basictype.type</a>* %<a href="#format_basic_type">llvm.dbg.basictype</a> to { }*) }, section "llvm.metadata"
1618%str3 = internal constant [4 x sbyte] c"Red\00", section "llvm.metadata"
1619
1620;;
1621;; Define the Green field.
1622;;
1623%<a href="#format_derived_type">llvm.dbg.derivedtype2</a> = internal constant %<a href="#format_derived_type">llvm.dbg.derivedtype.type</a> {
1624 uint 13,
1625 { }* null,
1626 sbyte* getelementptr ([6 x sbyte]* %str4, int 0, int 0),
1627 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1628 int 3,
1629 uint 32,
1630 uint 32,
1631 uint 32,
1632 { }* cast (%<a href="#format_basic_type">llvm.dbg.basictype.type</a>* %<a href="#format_basic_type">llvm.dbg.basictype</a> to { }*) }, section "llvm.metadata"
1633%str4 = internal constant [6 x sbyte] c"Green\00", section "llvm.metadata"
1634
1635;;
1636;; Define the Blue field.
1637;;
1638%<a href="#format_derived_type">llvm.dbg.derivedtype3</a> = internal constant %<a href="#format_derived_type">llvm.dbg.derivedtype.type</a> {
1639 uint 13,
1640 { }* null,
1641 sbyte* getelementptr ([5 x sbyte]* %str5, int 0, int 0),
1642 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1643 int 4,
1644 uint 32,
1645 uint 32,
1646 uint 64,
1647 { }* cast (%<a href="#format_basic_type">llvm.dbg.basictype.type</a>* %<a href="#format_basic_type">llvm.dbg.basictype</a> to { }*) }, section "llvm.metadata"
1648%str5 = internal constant [5 x sbyte] c"Blue\00", section "llvm.metadata"
1649
1650;;
1651;; Define the array of fields used by the composite type Color.
1652;;
1653%llvm.dbg.array = internal constant [3 x { }*] [
1654 { }* cast (%<a href="#format_derived_type">llvm.dbg.derivedtype.type</a>* %<a href="#format_derived_type">llvm.dbg.derivedtype1</a> to { }*),
1655 { }* cast (%<a href="#format_derived_type">llvm.dbg.derivedtype.type</a>* %<a href="#format_derived_type">llvm.dbg.derivedtype2</a> to { }*),
1656 { }* cast (%<a href="#format_derived_type">llvm.dbg.derivedtype.type</a>* %<a href="#format_derived_type">llvm.dbg.derivedtype3</a> to { }*) ], section "llvm.metadata"
1657</pre>
1658
1659</div>
1660
1661<!-- ======================================================================= -->
1662<div class="doc_subsection">
1663 <a name="ccxx_enumeration_types">C/C++ enumeration types</a>
1664</div>
1665
1666<div class="doc_text">
1667
1668<p>Given the following as an example of C/C++ enumeration type;</p>
1669
1670<pre>
1671enum Trees {
1672 Spruce = 100,
1673 Oak = 200,
1674 Maple = 300
1675};
1676</pre>
1677
1678<p>a C/C++ front-end would generate the following descriptors;</p>
1679
1680<pre>
1681;;
1682;; Define composite type for enum Trees
1683;;
1684%<a href="#format_composite_type">llvm.dbg.compositetype</a> = internal constant %<a href="#format_composite_type">llvm.dbg.compositetype.type</a> {
1685 uint 4,
1686 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1687 sbyte* getelementptr ([6 x sbyte]* %str1, int 0, int 0),
1688 { }* cast (%<a href="#format_compile_units">llvm.dbg.compile_unit.type</a>* %<a href="#format_compile_units">llvm.dbg.compile_unit</a> to { }*),
1689 int 1,
1690 uint 32,
1691 uint 32,
1692 uint 0,
1693 { }* null,
1694 { }* cast ([3 x { }*]* %llvm.dbg.array to { }*) }, section "llvm.metadata"
1695%str1 = internal constant [6 x sbyte] c"Trees\00", section "llvm.metadata"
1696
1697;;
1698;; Define Spruce enumerator.
1699;;
1700%<a href="#format_enumeration">llvm.dbg.enumerator1</a> = internal constant %<a href="#format_enumeration">llvm.dbg.enumerator.type</a> {
1701 uint 40,
1702 sbyte* getelementptr ([7 x sbyte]* %str2, int 0, int 0),
1703 int 100 }, section "llvm.metadata"
1704%str2 = internal constant [7 x sbyte] c"Spruce\00", section "llvm.metadata"
1705
1706;;
1707;; Define Oak enumerator.
1708;;
1709%<a href="#format_enumeration">llvm.dbg.enumerator2</a> = internal constant %<a href="#format_enumeration">llvm.dbg.enumerator.type</a> {
1710 uint 40,
1711 sbyte* getelementptr ([4 x sbyte]* %str3, int 0, int 0),
1712 int 200 }, section "llvm.metadata"
1713%str3 = internal constant [4 x sbyte] c"Oak\00", section "llvm.metadata"
1714
1715;;
1716;; Define Maple enumerator.
1717;;
1718%<a href="#format_enumeration">llvm.dbg.enumerator3</a> = internal constant %<a href="#format_enumeration">llvm.dbg.enumerator.type</a> {
1719 uint 40,
1720 sbyte* getelementptr ([6 x sbyte]* %str4, int 0, int 0),
1721 int 300 }, section "llvm.metadata"
1722%str4 = internal constant [6 x sbyte] c"Maple\00", section "llvm.metadata"
1723
1724;;
1725;; Define the array of enumerators used by composite type Trees.
1726;;
1727%llvm.dbg.array = internal constant [3 x { }*] [
1728 { }* cast (%<a href="#format_enumeration">llvm.dbg.enumerator.type</a>* %<a href="#format_enumeration">llvm.dbg.enumerator1</a> to { }*),
1729 { }* cast (%<a href="#format_enumeration">llvm.dbg.enumerator.type</a>* %<a href="#format_enumeration">llvm.dbg.enumerator2</a> to { }*),
1730 { }* cast (%<a href="#format_enumeration">llvm.dbg.enumerator.type</a>* %<a href="#format_enumeration">llvm.dbg.enumerator3</a> to { }*) ], section "llvm.metadata"
1731</pre>
1732
1733</div>
Chris Lattnerbdfb3392004-01-05 05:06:33 +00001734
1735<!-- *********************************************************************** -->
Misha Brukman82873732004-05-12 19:21:57 +00001736
Chris Lattnerbdfb3392004-01-05 05:06:33 +00001737<hr>
Misha Brukman82873732004-05-12 19:21:57 +00001738<address>
1739 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
1740 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
1741 <a href="http://validator.w3.org/check/referer"><img
1742 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
1743
1744 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
Reid Spencer05fe4b02006-03-14 05:39:39 +00001745 <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br>
Chris Lattnerbdfb3392004-01-05 05:06:33 +00001746 Last modified: $Date$
Misha Brukman82873732004-05-12 19:21:57 +00001747</address>
Chris Lattnerbdfb3392004-01-05 05:06:33 +00001748
1749</body>
1750</html>