• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1================================
2Source Level Debugging with LLVM
3================================
4
5.. contents::
6   :local:
7
8Introduction
9============
10
11This document is the central repository for all information pertaining to debug
12information in LLVM.  It describes the :ref:`actual format that the LLVM debug
13information takes <format>`, which is useful for those interested in creating
14front-ends or dealing directly with the information.  Further, this document
15provides specific examples of what debug information for C/C++ looks like.
16
17Philosophy behind LLVM debugging information
18--------------------------------------------
19
20The idea of the LLVM debugging information is to capture how the important
21pieces of the source-language's Abstract Syntax Tree map onto LLVM code.
22Several design aspects have shaped the solution that appears here.  The
23important ones are:
24
25* Debugging information should have very little impact on the rest of the
26  compiler.  No transformations, analyses, or code generators should need to
27  be modified because of debugging information.
28
29* LLVM optimizations should interact in :ref:`well-defined and easily described
30  ways <intro_debugopt>` with the debugging information.
31
32* Because LLVM is designed to support arbitrary programming languages,
33  LLVM-to-LLVM tools should not need to know anything about the semantics of
34  the source-level-language.
35
36* Source-level languages are often **widely** different from one another.
37  LLVM should not put any restrictions of the flavor of the source-language,
38  and the debugging information should work with any language.
39
40* With code generator support, it should be possible to use an LLVM compiler
41  to compile a program to native machine code and standard debugging
42  formats.  This allows compatibility with traditional machine-code level
43  debuggers, like GDB or DBX.
44
45The approach used by the LLVM implementation is to use a small set of
46:ref:`intrinsic functions <format_common_intrinsics>` to define a mapping
47between LLVM program objects and the source-level objects.  The description of
48the source-level program is maintained in LLVM metadata in an
49:ref:`implementation-defined format <ccxx_frontend>` (the C/C++ front-end
50currently uses working draft 7 of the `DWARF 3 standard
51<http://www.eagercon.com/dwarf/dwarf3std.htm>`_).
52
53When a program is being debugged, a debugger interacts with the user and turns
54the stored debug information into source-language specific information.  As
55such, a debugger must be aware of the source-language, and is thus tied to a
56specific language or family of languages.
57
58Debug information consumers
59---------------------------
60
61The role of debug information is to provide meta information normally stripped
62away during the compilation process.  This meta information provides an LLVM
63user a relationship between generated code and the original program source
64code.
65
66Currently, debug information is consumed by DwarfDebug to produce dwarf
67information used by the gdb debugger.  Other targets could use the same
68information to produce stabs or other debug forms.
69
70It would also be reasonable to use debug information to feed profiling tools
71for analysis of generated code, or, tools for reconstructing the original
72source from generated code.
73
74TODO - expound a bit more.
75
76.. _intro_debugopt:
77
78Debugging optimized code
79------------------------
80
81An extremely high priority of LLVM debugging information is to make it interact
82well with optimizations and analysis.  In particular, the LLVM debug
83information provides the following guarantees:
84
85* LLVM debug information **always provides information to accurately read
86  the source-level state of the program**, regardless of which LLVM
87  optimizations have been run, and without any modification to the
88  optimizations themselves.  However, some optimizations may impact the
89  ability to modify the current state of the program with a debugger, such
90  as setting program variables, or calling functions that have been
91  deleted.
92
93* As desired, LLVM optimizations can be upgraded to be aware of the LLVM
94  debugging information, allowing them to update the debugging information
95  as they perform aggressive optimizations.  This means that, with effort,
96  the LLVM optimizers could optimize debug code just as well as non-debug
97  code.
98
99* LLVM debug information does not prevent optimizations from
100  happening (for example inlining, basic block reordering/merging/cleanup,
101  tail duplication, etc).
102
103* LLVM debug information is automatically optimized along with the rest of
104  the program, using existing facilities.  For example, duplicate
105  information is automatically merged by the linker, and unused information
106  is automatically removed.
107
108Basically, the debug information allows you to compile a program with
109"``-O0 -g``" and get full debug information, allowing you to arbitrarily modify
110the program as it executes from a debugger.  Compiling a program with
111"``-O3 -g``" gives you full debug information that is always available and
112accurate for reading (e.g., you get accurate stack traces despite tail call
113elimination and inlining), but you might lose the ability to modify the program
114and call functions where were optimized out of the program, or inlined away
115completely.
116
117:ref:`LLVM test suite <test-suite-quickstart>` provides a framework to test
118optimizer's handling of debugging information.  It can be run like this:
119
120.. code-block:: bash
121
122  % cd llvm/projects/test-suite/MultiSource/Benchmarks  # or some other level
123  % make TEST=dbgopt
124
125This will test impact of debugging information on optimization passes.  If
126debugging information influences optimization passes then it will be reported
127as a failure.  See :doc:`TestingGuide` for more information on LLVM test
128infrastructure and how to run various tests.
129
130.. _format:
131
132Debugging information format
133============================
134
135LLVM debugging information has been carefully designed to make it possible for
136the optimizer to optimize the program and debugging information without
137necessarily having to know anything about debugging information.  In
138particular, the use of metadata avoids duplicated debugging information from
139the beginning, and the global dead code elimination pass automatically deletes
140debugging information for a function if it decides to delete the function.
141
142To do this, most of the debugging information (descriptors for types,
143variables, functions, source files, etc) is inserted by the language front-end
144in the form of LLVM metadata.
145
146Debug information is designed to be agnostic about the target debugger and
147debugging information representation (e.g. DWARF/Stabs/etc).  It uses a generic
148pass to decode the information that represents variables, types, functions,
149namespaces, etc: this allows for arbitrary source-language semantics and
150type-systems to be used, as long as there is a module written for the target
151debugger to interpret the information.
152
153To provide basic functionality, the LLVM debugger does have to make some
154assumptions about the source-level language being debugged, though it keeps
155these to a minimum.  The only common features that the LLVM debugger assumes
156exist are `source files <LangRef.html#difile>`_, and `program objects
157<LangRef.html#diglobalvariable>`_.  These abstract objects are used by a
158debugger to form stack traces, show information about local variables, etc.
159
160This section of the documentation first describes the representation aspects
161common to any source-language.  :ref:`ccxx_frontend` describes the data layout
162conventions used by the C and C++ front-ends.
163
164Debug information descriptors are `specialized metadata nodes
165<LangRef.html#specialized-metadata>`_, first-class subclasses of ``Metadata``.
166
167.. _format_common_intrinsics:
168
169Debugger intrinsic functions
170----------------------------
171
172LLVM uses several intrinsic functions (name prefixed with "``llvm.dbg``") to
173provide debug information at various points in generated code.
174
175``llvm.dbg.declare``
176^^^^^^^^^^^^^^^^^^^^
177
178.. code-block:: llvm
179
180  void @llvm.dbg.declare(metadata, metadata, metadata)
181
182This intrinsic provides information about a local element (e.g., variable).
183The first argument is metadata holding the alloca for the variable.  The second
184argument is a `local variable <LangRef.html#dilocalvariable>`_ containing a
185description of the variable.  The third argument is a `complex expression
186<LangRef.html#diexpression>`_.
187
188``llvm.dbg.value``
189^^^^^^^^^^^^^^^^^^
190
191.. code-block:: llvm
192
193  void @llvm.dbg.value(metadata, i64, metadata, metadata)
194
195This intrinsic provides information when a user source variable is set to a new
196value.  The first argument is the new value (wrapped as metadata).  The second
197argument is the offset in the user source variable where the new value is
198written.  The third argument is a `local variable
199<LangRef.html#dilocalvariable>`_ containing a description of the variable.  The
200third argument is a `complex expression <LangRef.html#diexpression>`_.
201
202Object lifetimes and scoping
203============================
204
205In many languages, the local variables in functions can have their lifetimes or
206scopes limited to a subset of a function.  In the C family of languages, for
207example, variables are only live (readable and writable) within the source
208block that they are defined in.  In functional languages, values are only
209readable after they have been defined.  Though this is a very obvious concept,
210it is non-trivial to model in LLVM, because it has no notion of scoping in this
211sense, and does not want to be tied to a language's scoping rules.
212
213In order to handle this, the LLVM debug format uses the metadata attached to
214llvm instructions to encode line number and scoping information.  Consider the
215following C fragment, for example:
216
217.. code-block:: c
218
219  1.  void foo() {
220  2.    int X = 21;
221  3.    int Y = 22;
222  4.    {
223  5.      int Z = 23;
224  6.      Z = X;
225  7.    }
226  8.    X = Y;
227  9.  }
228
229Compiled to LLVM, this function would be represented like this:
230
231.. code-block:: llvm
232
233  ; Function Attrs: nounwind ssp uwtable
234  define void @foo() #0 !dbg !4 {
235  entry:
236    %X = alloca i32, align 4
237    %Y = alloca i32, align 4
238    %Z = alloca i32, align 4
239    call void @llvm.dbg.declare(metadata i32* %X, metadata !11, metadata !13), !dbg !14
240    store i32 21, i32* %X, align 4, !dbg !14
241    call void @llvm.dbg.declare(metadata i32* %Y, metadata !15, metadata !13), !dbg !16
242    store i32 22, i32* %Y, align 4, !dbg !16
243    call void @llvm.dbg.declare(metadata i32* %Z, metadata !17, metadata !13), !dbg !19
244    store i32 23, i32* %Z, align 4, !dbg !19
245    %0 = load i32, i32* %X, align 4, !dbg !20
246    store i32 %0, i32* %Z, align 4, !dbg !21
247    %1 = load i32, i32* %Y, align 4, !dbg !22
248    store i32 %1, i32* %X, align 4, !dbg !23
249    ret void, !dbg !24
250  }
251
252  ; Function Attrs: nounwind readnone
253  declare void @llvm.dbg.declare(metadata, metadata, metadata) #1
254
255  attributes #0 = { nounwind ssp uwtable "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "no-infs-fp-math"="false" "no-nans-fp-math"="false" "stack-protector-buffer-size"="8" "unsafe-fp-math"="false" "use-soft-float"="false" }
256  attributes #1 = { nounwind readnone }
257
258  !llvm.dbg.cu = !{!0}
259  !llvm.module.flags = !{!7, !8, !9}
260  !llvm.ident = !{!10}
261
262  !0 = !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 3.7.0 (trunk 231150) (llvm/trunk 231154)", isOptimized: false, runtimeVersion: 0, emissionKind: 1, enums: !2, retainedTypes: !2, subprograms: !3, globals: !2, imports: !2)
263  !1 = !DIFile(filename: "/dev/stdin", directory: "/Users/dexonsmith/data/llvm/debug-info")
264  !2 = !{}
265  !3 = !{!4}
266  !4 = distinct !DISubprogram(name: "foo", scope: !1, file: !1, line: 1, type: !5, isLocal: false, isDefinition: true, scopeLine: 1, isOptimized: false, variables: !2)
267  !5 = !DISubroutineType(types: !6)
268  !6 = !{null}
269  !7 = !{i32 2, !"Dwarf Version", i32 2}
270  !8 = !{i32 2, !"Debug Info Version", i32 3}
271  !9 = !{i32 1, !"PIC Level", i32 2}
272  !10 = !{!"clang version 3.7.0 (trunk 231150) (llvm/trunk 231154)"}
273  !11 = !DILocalVariable(name: "X", scope: !4, file: !1, line: 2, type: !12)
274  !12 = !DIBasicType(name: "int", size: 32, align: 32, encoding: DW_ATE_signed)
275  !13 = !DIExpression()
276  !14 = !DILocation(line: 2, column: 9, scope: !4)
277  !15 = !DILocalVariable(name: "Y", scope: !4, file: !1, line: 3, type: !12)
278  !16 = !DILocation(line: 3, column: 9, scope: !4)
279  !17 = !DILocalVariable(name: "Z", scope: !18, file: !1, line: 5, type: !12)
280  !18 = distinct !DILexicalBlock(scope: !4, file: !1, line: 4, column: 5)
281  !19 = !DILocation(line: 5, column: 11, scope: !18)
282  !20 = !DILocation(line: 6, column: 11, scope: !18)
283  !21 = !DILocation(line: 6, column: 9, scope: !18)
284  !22 = !DILocation(line: 8, column: 9, scope: !4)
285  !23 = !DILocation(line: 8, column: 7, scope: !4)
286  !24 = !DILocation(line: 9, column: 3, scope: !4)
287
288
289This example illustrates a few important details about LLVM debugging
290information.  In particular, it shows how the ``llvm.dbg.declare`` intrinsic and
291location information, which are attached to an instruction, are applied
292together to allow a debugger to analyze the relationship between statements,
293variable definitions, and the code used to implement the function.
294
295.. code-block:: llvm
296
297  call void @llvm.dbg.declare(metadata i32* %X, metadata !11, metadata !13), !dbg !14
298    ; [debug line = 2:7] [debug variable = X]
299
300The first intrinsic ``%llvm.dbg.declare`` encodes debugging information for the
301variable ``X``.  The metadata ``!dbg !14`` attached to the intrinsic provides
302scope information for the variable ``X``.
303
304.. code-block:: llvm
305
306  !14 = !DILocation(line: 2, column: 9, scope: !4)
307  !4 = distinct !DISubprogram(name: "foo", scope: !1, file: !1, line: 1, type: !5,
308                              isLocal: false, isDefinition: true, scopeLine: 1,
309                              isOptimized: false, variables: !2)
310
311Here ``!14`` is metadata providing `location information
312<LangRef.html#dilocation>`_.  In this example, scope is encoded by ``!4``, a
313`subprogram descriptor <LangRef.html#disubprogram>`_.  This way the location
314information attached to the intrinsics indicates that the variable ``X`` is
315declared at line number 2 at a function level scope in function ``foo``.
316
317Now lets take another example.
318
319.. code-block:: llvm
320
321  call void @llvm.dbg.declare(metadata i32* %Z, metadata !17, metadata !13), !dbg !19
322    ; [debug line = 5:9] [debug variable = Z]
323
324The third intrinsic ``%llvm.dbg.declare`` encodes debugging information for
325variable ``Z``.  The metadata ``!dbg !19`` attached to the intrinsic provides
326scope information for the variable ``Z``.
327
328.. code-block:: llvm
329
330  !18 = distinct !DILexicalBlock(scope: !4, file: !1, line: 4, column: 5)
331  !19 = !DILocation(line: 5, column: 11, scope: !18)
332
333Here ``!19`` indicates that ``Z`` is declared at line number 5 and column
334number 0 inside of lexical scope ``!18``.  The lexical scope itself resides
335inside of subprogram ``!4`` described above.
336
337The scope information attached with each instruction provides a straightforward
338way to find instructions covered by a scope.
339
340.. _ccxx_frontend:
341
342C/C++ front-end specific debug information
343==========================================
344
345The C and C++ front-ends represent information about the program in a format
346that is effectively identical to `DWARF 3.0
347<http://www.eagercon.com/dwarf/dwarf3std.htm>`_ in terms of information
348content.  This allows code generators to trivially support native debuggers by
349generating standard dwarf information, and contains enough information for
350non-dwarf targets to translate it as needed.
351
352This section describes the forms used to represent C and C++ programs.  Other
353languages could pattern themselves after this (which itself is tuned to
354representing programs in the same way that DWARF 3 does), or they could choose
355to provide completely different forms if they don't fit into the DWARF model.
356As support for debugging information gets added to the various LLVM
357source-language front-ends, the information used should be documented here.
358
359The following sections provide examples of a few C/C++ constructs and the debug
360information that would best describe those constructs.  The canonical
361references are the ``DIDescriptor`` classes defined in
362``include/llvm/IR/DebugInfo.h`` and the implementations of the helper functions
363in ``lib/IR/DIBuilder.cpp``.
364
365C/C++ source file information
366-----------------------------
367
368``llvm::Instruction`` provides easy access to metadata attached with an
369instruction.  One can extract line number information encoded in LLVM IR using
370``Instruction::getDebugLoc()`` and ``DILocation::getLine()``.
371
372.. code-block:: c++
373
374  if (DILocation *Loc = I->getDebugLoc()) { // Here I is an LLVM instruction
375    unsigned Line = Loc->getLine();
376    StringRef File = Loc->getFilename();
377    StringRef Dir = Loc->getDirectory();
378  }
379
380C/C++ global variable information
381---------------------------------
382
383Given an integer global variable declared as follows:
384
385.. code-block:: c
386
387  int MyGlobal = 100;
388
389a C/C++ front-end would generate the following descriptors:
390
391.. code-block:: llvm
392
393  ;;
394  ;; Define the global itself.
395  ;;
396  @MyGlobal = global i32 100, align 4
397
398  ;;
399  ;; List of debug info of globals
400  ;;
401  !llvm.dbg.cu = !{!0}
402
403  ;; Some unrelated metadata.
404  !llvm.module.flags = !{!6, !7}
405
406  ;; Define the compile unit.
407  !0 = !DICompileUnit(language: DW_LANG_C99, file: !1,
408                      producer:
409                      "clang version 3.7.0 (trunk 231150) (llvm/trunk 231154)",
410                      isOptimized: false, runtimeVersion: 0, emissionKind: 1,
411                      enums: !2, retainedTypes: !2, subprograms: !2, globals:
412                      !3, imports: !2)
413
414  ;;
415  ;; Define the file
416  ;;
417  !1 = !DIFile(filename: "/dev/stdin",
418               directory: "/Users/dexonsmith/data/llvm/debug-info")
419
420  ;; An empty array.
421  !2 = !{}
422
423  ;; The Array of Global Variables
424  !3 = !{!4}
425
426  ;;
427  ;; Define the global variable itself.
428  ;;
429  !4 = !DIGlobalVariable(name: "MyGlobal", scope: !0, file: !1, line: 1,
430                         type: !5, isLocal: false, isDefinition: true,
431                         variable: i32* @MyGlobal)
432
433  ;;
434  ;; Define the type
435  ;;
436  !5 = !DIBasicType(name: "int", size: 32, align: 32, encoding: DW_ATE_signed)
437
438  ;; Dwarf version to output.
439  !6 = !{i32 2, !"Dwarf Version", i32 2}
440
441  ;; Debug info schema version.
442  !7 = !{i32 2, !"Debug Info Version", i32 3}
443
444C/C++ function information
445--------------------------
446
447Given a function declared as follows:
448
449.. code-block:: c
450
451  int main(int argc, char *argv[]) {
452    return 0;
453  }
454
455a C/C++ front-end would generate the following descriptors:
456
457.. code-block:: llvm
458
459  ;;
460  ;; Define the anchor for subprograms.
461  ;;
462  !4 = !DISubprogram(name: "main", scope: !1, file: !1, line: 1, type: !5,
463                     isLocal: false, isDefinition: true, scopeLine: 1,
464                     flags: DIFlagPrototyped, isOptimized: false,
465                     variables: !2)
466
467  ;;
468  ;; Define the subprogram itself.
469  ;;
470  define i32 @main(i32 %argc, i8** %argv) !dbg !4 {
471  ...
472  }
473
474Debugging information format
475============================
476
477Debugging Information Extension for Objective C Properties
478----------------------------------------------------------
479
480Introduction
481^^^^^^^^^^^^
482
483Objective C provides a simpler way to declare and define accessor methods using
484declared properties.  The language provides features to declare a property and
485to let compiler synthesize accessor methods.
486
487The debugger lets developer inspect Objective C interfaces and their instance
488variables and class variables.  However, the debugger does not know anything
489about the properties defined in Objective C interfaces.  The debugger consumes
490information generated by compiler in DWARF format.  The format does not support
491encoding of Objective C properties.  This proposal describes DWARF extensions to
492encode Objective C properties, which the debugger can use to let developers
493inspect Objective C properties.
494
495Proposal
496^^^^^^^^
497
498Objective C properties exist separately from class members.  A property can be
499defined only by "setter" and "getter" selectors, and be calculated anew on each
500access.  Or a property can just be a direct access to some declared ivar.
501Finally it can have an ivar "automatically synthesized" for it by the compiler,
502in which case the property can be referred to in user code directly using the
503standard C dereference syntax as well as through the property "dot" syntax, but
504there is no entry in the ``@interface`` declaration corresponding to this ivar.
505
506To facilitate debugging, these properties we will add a new DWARF TAG into the
507``DW_TAG_structure_type`` definition for the class to hold the description of a
508given property, and a set of DWARF attributes that provide said description.
509The property tag will also contain the name and declared type of the property.
510
511If there is a related ivar, there will also be a DWARF property attribute placed
512in the ``DW_TAG_member`` DIE for that ivar referring back to the property TAG
513for that property.  And in the case where the compiler synthesizes the ivar
514directly, the compiler is expected to generate a ``DW_TAG_member`` for that
515ivar (with the ``DW_AT_artificial`` set to 1), whose name will be the name used
516to access this ivar directly in code, and with the property attribute pointing
517back to the property it is backing.
518
519The following examples will serve as illustration for our discussion:
520
521.. code-block:: objc
522
523  @interface I1 {
524    int n2;
525  }
526
527  @property int p1;
528  @property int p2;
529  @end
530
531  @implementation I1
532  @synthesize p1;
533  @synthesize p2 = n2;
534  @end
535
536This produces the following DWARF (this is a "pseudo dwarfdump" output):
537
538.. code-block:: none
539
540  0x00000100:  TAG_structure_type [7] *
541                 AT_APPLE_runtime_class( 0x10 )
542                 AT_name( "I1" )
543                 AT_decl_file( "Objc_Property.m" )
544                 AT_decl_line( 3 )
545
546  0x00000110    TAG_APPLE_property
547                  AT_name ( "p1" )
548                  AT_type ( {0x00000150} ( int ) )
549
550  0x00000120:   TAG_APPLE_property
551                  AT_name ( "p2" )
552                  AT_type ( {0x00000150} ( int ) )
553
554  0x00000130:   TAG_member [8]
555                  AT_name( "_p1" )
556                  AT_APPLE_property ( {0x00000110} "p1" )
557                  AT_type( {0x00000150} ( int ) )
558                  AT_artificial ( 0x1 )
559
560  0x00000140:    TAG_member [8]
561                   AT_name( "n2" )
562                   AT_APPLE_property ( {0x00000120} "p2" )
563                   AT_type( {0x00000150} ( int ) )
564
565  0x00000150:  AT_type( ( int ) )
566
567Note, the current convention is that the name of the ivar for an
568auto-synthesized property is the name of the property from which it derives
569with an underscore prepended, as is shown in the example.  But we actually
570don't need to know this convention, since we are given the name of the ivar
571directly.
572
573Also, it is common practice in ObjC to have different property declarations in
574the @interface and @implementation - e.g. to provide a read-only property in
575the interface,and a read-write interface in the implementation.  In that case,
576the compiler should emit whichever property declaration will be in force in the
577current translation unit.
578
579Developers can decorate a property with attributes which are encoded using
580``DW_AT_APPLE_property_attribute``.
581
582.. code-block:: objc
583
584  @property (readonly, nonatomic) int pr;
585
586.. code-block:: none
587
588  TAG_APPLE_property [8]
589    AT_name( "pr" )
590    AT_type ( {0x00000147} (int) )
591    AT_APPLE_property_attribute (DW_APPLE_PROPERTY_readonly, DW_APPLE_PROPERTY_nonatomic)
592
593The setter and getter method names are attached to the property using
594``DW_AT_APPLE_property_setter`` and ``DW_AT_APPLE_property_getter`` attributes.
595
596.. code-block:: objc
597
598  @interface I1
599  @property (setter=myOwnP3Setter:) int p3;
600  -(void)myOwnP3Setter:(int)a;
601  @end
602
603  @implementation I1
604  @synthesize p3;
605  -(void)myOwnP3Setter:(int)a{ }
606  @end
607
608The DWARF for this would be:
609
610.. code-block:: none
611
612  0x000003bd: TAG_structure_type [7] *
613                AT_APPLE_runtime_class( 0x10 )
614                AT_name( "I1" )
615                AT_decl_file( "Objc_Property.m" )
616                AT_decl_line( 3 )
617
618  0x000003cd      TAG_APPLE_property
619                    AT_name ( "p3" )
620                    AT_APPLE_property_setter ( "myOwnP3Setter:" )
621                    AT_type( {0x00000147} ( int ) )
622
623  0x000003f3:     TAG_member [8]
624                    AT_name( "_p3" )
625                    AT_type ( {0x00000147} ( int ) )
626                    AT_APPLE_property ( {0x000003cd} )
627                    AT_artificial ( 0x1 )
628
629New DWARF Tags
630^^^^^^^^^^^^^^
631
632+-----------------------+--------+
633| TAG                   | Value  |
634+=======================+========+
635| DW_TAG_APPLE_property | 0x4200 |
636+-----------------------+--------+
637
638New DWARF Attributes
639^^^^^^^^^^^^^^^^^^^^
640
641+--------------------------------+--------+-----------+
642| Attribute                      | Value  | Classes   |
643+================================+========+===========+
644| DW_AT_APPLE_property           | 0x3fed | Reference |
645+--------------------------------+--------+-----------+
646| DW_AT_APPLE_property_getter    | 0x3fe9 | String    |
647+--------------------------------+--------+-----------+
648| DW_AT_APPLE_property_setter    | 0x3fea | String    |
649+--------------------------------+--------+-----------+
650| DW_AT_APPLE_property_attribute | 0x3feb | Constant  |
651+--------------------------------+--------+-----------+
652
653New DWARF Constants
654^^^^^^^^^^^^^^^^^^^
655
656+--------------------------------------+-------+
657| Name                                 | Value |
658+======================================+=======+
659| DW_APPLE_PROPERTY_readonly           | 0x01  |
660+--------------------------------------+-------+
661| DW_APPLE_PROPERTY_getter             | 0x02  |
662+--------------------------------------+-------+
663| DW_APPLE_PROPERTY_assign             | 0x04  |
664+--------------------------------------+-------+
665| DW_APPLE_PROPERTY_readwrite          | 0x08  |
666+--------------------------------------+-------+
667| DW_APPLE_PROPERTY_retain             | 0x10  |
668+--------------------------------------+-------+
669| DW_APPLE_PROPERTY_copy               | 0x20  |
670+--------------------------------------+-------+
671| DW_APPLE_PROPERTY_nonatomic          | 0x40  |
672+--------------------------------------+-------+
673| DW_APPLE_PROPERTY_setter             | 0x80  |
674+--------------------------------------+-------+
675| DW_APPLE_PROPERTY_atomic             | 0x100 |
676+--------------------------------------+-------+
677| DW_APPLE_PROPERTY_weak               | 0x200 |
678+--------------------------------------+-------+
679| DW_APPLE_PROPERTY_strong             | 0x400 |
680+--------------------------------------+-------+
681| DW_APPLE_PROPERTY_unsafe_unretained  | 0x800 |
682+--------------------------------+-----+-------+
683
684Name Accelerator Tables
685-----------------------
686
687Introduction
688^^^^^^^^^^^^
689
690The "``.debug_pubnames``" and "``.debug_pubtypes``" formats are not what a
691debugger needs.  The "``pub``" in the section name indicates that the entries
692in the table are publicly visible names only.  This means no static or hidden
693functions show up in the "``.debug_pubnames``".  No static variables or private
694class variables are in the "``.debug_pubtypes``".  Many compilers add different
695things to these tables, so we can't rely upon the contents between gcc, icc, or
696clang.
697
698The typical query given by users tends not to match up with the contents of
699these tables.  For example, the DWARF spec states that "In the case of the name
700of a function member or static data member of a C++ structure, class or union,
701the name presented in the "``.debug_pubnames``" section is not the simple name
702given by the ``DW_AT_name attribute`` of the referenced debugging information
703entry, but rather the fully qualified name of the data or function member."
704So the only names in these tables for complex C++ entries is a fully
705qualified name.  Debugger users tend not to enter their search strings as
706"``a::b::c(int,const Foo&) const``", but rather as "``c``", "``b::c``" , or
707"``a::b::c``".  So the name entered in the name table must be demangled in
708order to chop it up appropriately and additional names must be manually entered
709into the table to make it effective as a name lookup table for debuggers to
710use.
711
712All debuggers currently ignore the "``.debug_pubnames``" table as a result of
713its inconsistent and useless public-only name content making it a waste of
714space in the object file.  These tables, when they are written to disk, are not
715sorted in any way, leaving every debugger to do its own parsing and sorting.
716These tables also include an inlined copy of the string values in the table
717itself making the tables much larger than they need to be on disk, especially
718for large C++ programs.
719
720Can't we just fix the sections by adding all of the names we need to this
721table? No, because that is not what the tables are defined to contain and we
722won't know the difference between the old bad tables and the new good tables.
723At best we could make our own renamed sections that contain all of the data we
724need.
725
726These tables are also insufficient for what a debugger like LLDB needs.  LLDB
727uses clang for its expression parsing where LLDB acts as a PCH.  LLDB is then
728often asked to look for type "``foo``" or namespace "``bar``", or list items in
729namespace "``baz``".  Namespaces are not included in the pubnames or pubtypes
730tables.  Since clang asks a lot of questions when it is parsing an expression,
731we need to be very fast when looking up names, as it happens a lot.  Having new
732accelerator tables that are optimized for very quick lookups will benefit this
733type of debugging experience greatly.
734
735We would like to generate name lookup tables that can be mapped into memory
736from disk, and used as is, with little or no up-front parsing.  We would also
737be able to control the exact content of these different tables so they contain
738exactly what we need.  The Name Accelerator Tables were designed to fix these
739issues.  In order to solve these issues we need to:
740
741* Have a format that can be mapped into memory from disk and used as is
742* Lookups should be very fast
743* Extensible table format so these tables can be made by many producers
744* Contain all of the names needed for typical lookups out of the box
745* Strict rules for the contents of tables
746
747Table size is important and the accelerator table format should allow the reuse
748of strings from common string tables so the strings for the names are not
749duplicated.  We also want to make sure the table is ready to be used as-is by
750simply mapping the table into memory with minimal header parsing.
751
752The name lookups need to be fast and optimized for the kinds of lookups that
753debuggers tend to do.  Optimally we would like to touch as few parts of the
754mapped table as possible when doing a name lookup and be able to quickly find
755the name entry we are looking for, or discover there are no matches.  In the
756case of debuggers we optimized for lookups that fail most of the time.
757
758Each table that is defined should have strict rules on exactly what is in the
759accelerator tables and documented so clients can rely on the content.
760
761Hash Tables
762^^^^^^^^^^^
763
764Standard Hash Tables
765""""""""""""""""""""
766
767Typical hash tables have a header, buckets, and each bucket points to the
768bucket contents:
769
770.. code-block:: none
771
772  .------------.
773  |  HEADER    |
774  |------------|
775  |  BUCKETS   |
776  |------------|
777  |  DATA      |
778  `------------'
779
780The BUCKETS are an array of offsets to DATA for each hash:
781
782.. code-block:: none
783
784  .------------.
785  | 0x00001000 | BUCKETS[0]
786  | 0x00002000 | BUCKETS[1]
787  | 0x00002200 | BUCKETS[2]
788  | 0x000034f0 | BUCKETS[3]
789  |            | ...
790  | 0xXXXXXXXX | BUCKETS[n_buckets]
791  '------------'
792
793So for ``bucket[3]`` in the example above, we have an offset into the table
7940x000034f0 which points to a chain of entries for the bucket.  Each bucket must
795contain a next pointer, full 32 bit hash value, the string itself, and the data
796for the current string value.
797
798.. code-block:: none
799
800              .------------.
801  0x000034f0: | 0x00003500 | next pointer
802              | 0x12345678 | 32 bit hash
803              | "erase"    | string value
804              | data[n]    | HashData for this bucket
805              |------------|
806  0x00003500: | 0x00003550 | next pointer
807              | 0x29273623 | 32 bit hash
808              | "dump"     | string value
809              | data[n]    | HashData for this bucket
810              |------------|
811  0x00003550: | 0x00000000 | next pointer
812              | 0x82638293 | 32 bit hash
813              | "main"     | string value
814              | data[n]    | HashData for this bucket
815              `------------'
816
817The problem with this layout for debuggers is that we need to optimize for the
818negative lookup case where the symbol we're searching for is not present.  So
819if we were to lookup "``printf``" in the table above, we would make a 32 hash
820for "``printf``", it might match ``bucket[3]``.  We would need to go to the
821offset 0x000034f0 and start looking to see if our 32 bit hash matches.  To do
822so, we need to read the next pointer, then read the hash, compare it, and skip
823to the next bucket.  Each time we are skipping many bytes in memory and
824touching new cache pages just to do the compare on the full 32 bit hash.  All
825of these accesses then tell us that we didn't have a match.
826
827Name Hash Tables
828""""""""""""""""
829
830To solve the issues mentioned above we have structured the hash tables a bit
831differently: a header, buckets, an array of all unique 32 bit hash values,
832followed by an array of hash value data offsets, one for each hash value, then
833the data for all hash values:
834
835.. code-block:: none
836
837  .-------------.
838  |  HEADER     |
839  |-------------|
840  |  BUCKETS    |
841  |-------------|
842  |  HASHES     |
843  |-------------|
844  |  OFFSETS    |
845  |-------------|
846  |  DATA       |
847  `-------------'
848
849The ``BUCKETS`` in the name tables are an index into the ``HASHES`` array.  By
850making all of the full 32 bit hash values contiguous in memory, we allow
851ourselves to efficiently check for a match while touching as little memory as
852possible.  Most often checking the 32 bit hash values is as far as the lookup
853goes.  If it does match, it usually is a match with no collisions.  So for a
854table with "``n_buckets``" buckets, and "``n_hashes``" unique 32 bit hash
855values, we can clarify the contents of the ``BUCKETS``, ``HASHES`` and
856``OFFSETS`` as:
857
858.. code-block:: none
859
860  .-------------------------.
861  |  HEADER.magic           | uint32_t
862  |  HEADER.version         | uint16_t
863  |  HEADER.hash_function   | uint16_t
864  |  HEADER.bucket_count    | uint32_t
865  |  HEADER.hashes_count    | uint32_t
866  |  HEADER.header_data_len | uint32_t
867  |  HEADER_DATA            | HeaderData
868  |-------------------------|
869  |  BUCKETS                | uint32_t[n_buckets] // 32 bit hash indexes
870  |-------------------------|
871  |  HASHES                 | uint32_t[n_hashes] // 32 bit hash values
872  |-------------------------|
873  |  OFFSETS                | uint32_t[n_hashes] // 32 bit offsets to hash value data
874  |-------------------------|
875  |  ALL HASH DATA          |
876  `-------------------------'
877
878So taking the exact same data from the standard hash example above we end up
879with:
880
881.. code-block:: none
882
883              .------------.
884              | HEADER     |
885              |------------|
886              |          0 | BUCKETS[0]
887              |          2 | BUCKETS[1]
888              |          5 | BUCKETS[2]
889              |          6 | BUCKETS[3]
890              |            | ...
891              |        ... | BUCKETS[n_buckets]
892              |------------|
893              | 0x........ | HASHES[0]
894              | 0x........ | HASHES[1]
895              | 0x........ | HASHES[2]
896              | 0x........ | HASHES[3]
897              | 0x........ | HASHES[4]
898              | 0x........ | HASHES[5]
899              | 0x12345678 | HASHES[6]    hash for BUCKETS[3]
900              | 0x29273623 | HASHES[7]    hash for BUCKETS[3]
901              | 0x82638293 | HASHES[8]    hash for BUCKETS[3]
902              | 0x........ | HASHES[9]
903              | 0x........ | HASHES[10]
904              | 0x........ | HASHES[11]
905              | 0x........ | HASHES[12]
906              | 0x........ | HASHES[13]
907              | 0x........ | HASHES[n_hashes]
908              |------------|
909              | 0x........ | OFFSETS[0]
910              | 0x........ | OFFSETS[1]
911              | 0x........ | OFFSETS[2]
912              | 0x........ | OFFSETS[3]
913              | 0x........ | OFFSETS[4]
914              | 0x........ | OFFSETS[5]
915              | 0x000034f0 | OFFSETS[6]   offset for BUCKETS[3]
916              | 0x00003500 | OFFSETS[7]   offset for BUCKETS[3]
917              | 0x00003550 | OFFSETS[8]   offset for BUCKETS[3]
918              | 0x........ | OFFSETS[9]
919              | 0x........ | OFFSETS[10]
920              | 0x........ | OFFSETS[11]
921              | 0x........ | OFFSETS[12]
922              | 0x........ | OFFSETS[13]
923              | 0x........ | OFFSETS[n_hashes]
924              |------------|
925              |            |
926              |            |
927              |            |
928              |            |
929              |            |
930              |------------|
931  0x000034f0: | 0x00001203 | .debug_str ("erase")
932              | 0x00000004 | A 32 bit array count - number of HashData with name "erase"
933              | 0x........ | HashData[0]
934              | 0x........ | HashData[1]
935              | 0x........ | HashData[2]
936              | 0x........ | HashData[3]
937              | 0x00000000 | String offset into .debug_str (terminate data for hash)
938              |------------|
939  0x00003500: | 0x00001203 | String offset into .debug_str ("collision")
940              | 0x00000002 | A 32 bit array count - number of HashData with name "collision"
941              | 0x........ | HashData[0]
942              | 0x........ | HashData[1]
943              | 0x00001203 | String offset into .debug_str ("dump")
944              | 0x00000003 | A 32 bit array count - number of HashData with name "dump"
945              | 0x........ | HashData[0]
946              | 0x........ | HashData[1]
947              | 0x........ | HashData[2]
948              | 0x00000000 | String offset into .debug_str (terminate data for hash)
949              |------------|
950  0x00003550: | 0x00001203 | String offset into .debug_str ("main")
951              | 0x00000009 | A 32 bit array count - number of HashData with name "main"
952              | 0x........ | HashData[0]
953              | 0x........ | HashData[1]
954              | 0x........ | HashData[2]
955              | 0x........ | HashData[3]
956              | 0x........ | HashData[4]
957              | 0x........ | HashData[5]
958              | 0x........ | HashData[6]
959              | 0x........ | HashData[7]
960              | 0x........ | HashData[8]
961              | 0x00000000 | String offset into .debug_str (terminate data for hash)
962              `------------'
963
964So we still have all of the same data, we just organize it more efficiently for
965debugger lookup.  If we repeat the same "``printf``" lookup from above, we
966would hash "``printf``" and find it matches ``BUCKETS[3]`` by taking the 32 bit
967hash value and modulo it by ``n_buckets``.  ``BUCKETS[3]`` contains "6" which
968is the index into the ``HASHES`` table.  We would then compare any consecutive
96932 bit hashes values in the ``HASHES`` array as long as the hashes would be in
970``BUCKETS[3]``.  We do this by verifying that each subsequent hash value modulo
971``n_buckets`` is still 3.  In the case of a failed lookup we would access the
972memory for ``BUCKETS[3]``, and then compare a few consecutive 32 bit hashes
973before we know that we have no match.  We don't end up marching through
974multiple words of memory and we really keep the number of processor data cache
975lines being accessed as small as possible.
976
977The string hash that is used for these lookup tables is the Daniel J.
978Bernstein hash which is also used in the ELF ``GNU_HASH`` sections.  It is a
979very good hash for all kinds of names in programs with very few hash
980collisions.
981
982Empty buckets are designated by using an invalid hash index of ``UINT32_MAX``.
983
984Details
985^^^^^^^
986
987These name hash tables are designed to be generic where specializations of the
988table get to define additional data that goes into the header ("``HeaderData``"),
989how the string value is stored ("``KeyType``") and the content of the data for each
990hash value.
991
992Header Layout
993"""""""""""""
994
995The header has a fixed part, and the specialized part.  The exact format of the
996header is:
997
998.. code-block:: c
999
1000  struct Header
1001  {
1002    uint32_t   magic;           // 'HASH' magic value to allow endian detection
1003    uint16_t   version;         // Version number
1004    uint16_t   hash_function;   // The hash function enumeration that was used
1005    uint32_t   bucket_count;    // The number of buckets in this hash table
1006    uint32_t   hashes_count;    // The total number of unique hash values and hash data offsets in this table
1007    uint32_t   header_data_len; // The bytes to skip to get to the hash indexes (buckets) for correct alignment
1008                                // Specifically the length of the following HeaderData field - this does not
1009                                // include the size of the preceding fields
1010    HeaderData header_data;     // Implementation specific header data
1011  };
1012
1013The header starts with a 32 bit "``magic``" value which must be ``'HASH'``
1014encoded as an ASCII integer.  This allows the detection of the start of the
1015hash table and also allows the table's byte order to be determined so the table
1016can be correctly extracted.  The "``magic``" value is followed by a 16 bit
1017``version`` number which allows the table to be revised and modified in the
1018future.  The current version number is 1. ``hash_function`` is a ``uint16_t``
1019enumeration that specifies which hash function was used to produce this table.
1020The current values for the hash function enumerations include:
1021
1022.. code-block:: c
1023
1024  enum HashFunctionType
1025  {
1026    eHashFunctionDJB = 0u, // Daniel J Bernstein hash function
1027  };
1028
1029``bucket_count`` is a 32 bit unsigned integer that represents how many buckets
1030are in the ``BUCKETS`` array.  ``hashes_count`` is the number of unique 32 bit
1031hash values that are in the ``HASHES`` array, and is the same number of offsets
1032are contained in the ``OFFSETS`` array.  ``header_data_len`` specifies the size
1033in bytes of the ``HeaderData`` that is filled in by specialized versions of
1034this table.
1035
1036Fixed Lookup
1037""""""""""""
1038
1039The header is followed by the buckets, hashes, offsets, and hash value data.
1040
1041.. code-block:: c
1042
1043  struct FixedTable
1044  {
1045    uint32_t buckets[Header.bucket_count];  // An array of hash indexes into the "hashes[]" array below
1046    uint32_t hashes [Header.hashes_count];  // Every unique 32 bit hash for the entire table is in this table
1047    uint32_t offsets[Header.hashes_count];  // An offset that corresponds to each item in the "hashes[]" array above
1048  };
1049
1050``buckets`` is an array of 32 bit indexes into the ``hashes`` array.  The
1051``hashes`` array contains all of the 32 bit hash values for all names in the
1052hash table.  Each hash in the ``hashes`` table has an offset in the ``offsets``
1053array that points to the data for the hash value.
1054
1055This table setup makes it very easy to repurpose these tables to contain
1056different data, while keeping the lookup mechanism the same for all tables.
1057This layout also makes it possible to save the table to disk and map it in
1058later and do very efficient name lookups with little or no parsing.
1059
1060DWARF lookup tables can be implemented in a variety of ways and can store a lot
1061of information for each name.  We want to make the DWARF tables extensible and
1062able to store the data efficiently so we have used some of the DWARF features
1063that enable efficient data storage to define exactly what kind of data we store
1064for each name.
1065
1066The ``HeaderData`` contains a definition of the contents of each HashData chunk.
1067We might want to store an offset to all of the debug information entries (DIEs)
1068for each name.  To keep things extensible, we create a list of items, or
1069Atoms, that are contained in the data for each name.  First comes the type of
1070the data in each atom:
1071
1072.. code-block:: c
1073
1074  enum AtomType
1075  {
1076    eAtomTypeNULL       = 0u,
1077    eAtomTypeDIEOffset  = 1u,   // DIE offset, check form for encoding
1078    eAtomTypeCUOffset   = 2u,   // DIE offset of the compiler unit header that contains the item in question
1079    eAtomTypeTag        = 3u,   // DW_TAG_xxx value, should be encoded as DW_FORM_data1 (if no tags exceed 255) or DW_FORM_data2
1080    eAtomTypeNameFlags  = 4u,   // Flags from enum NameFlags
1081    eAtomTypeTypeFlags  = 5u,   // Flags from enum TypeFlags
1082  };
1083
1084The enumeration values and their meanings are:
1085
1086.. code-block:: none
1087
1088  eAtomTypeNULL       - a termination atom that specifies the end of the atom list
1089  eAtomTypeDIEOffset  - an offset into the .debug_info section for the DWARF DIE for this name
1090  eAtomTypeCUOffset   - an offset into the .debug_info section for the CU that contains the DIE
1091  eAtomTypeDIETag     - The DW_TAG_XXX enumeration value so you don't have to parse the DWARF to see what it is
1092  eAtomTypeNameFlags  - Flags for functions and global variables (isFunction, isInlined, isExternal...)
1093  eAtomTypeTypeFlags  - Flags for types (isCXXClass, isObjCClass, ...)
1094
1095Then we allow each atom type to define the atom type and how the data for each
1096atom type data is encoded:
1097
1098.. code-block:: c
1099
1100  struct Atom
1101  {
1102    uint16_t type;  // AtomType enum value
1103    uint16_t form;  // DWARF DW_FORM_XXX defines
1104  };
1105
1106The ``form`` type above is from the DWARF specification and defines the exact
1107encoding of the data for the Atom type.  See the DWARF specification for the
1108``DW_FORM_`` definitions.
1109
1110.. code-block:: c
1111
1112  struct HeaderData
1113  {
1114    uint32_t die_offset_base;
1115    uint32_t atom_count;
1116    Atoms    atoms[atom_count0];
1117  };
1118
1119``HeaderData`` defines the base DIE offset that should be added to any atoms
1120that are encoded using the ``DW_FORM_ref1``, ``DW_FORM_ref2``,
1121``DW_FORM_ref4``, ``DW_FORM_ref8`` or ``DW_FORM_ref_udata``.  It also defines
1122what is contained in each ``HashData`` object -- ``Atom.form`` tells us how large
1123each field will be in the ``HashData`` and the ``Atom.type`` tells us how this data
1124should be interpreted.
1125
1126For the current implementations of the "``.apple_names``" (all functions +
1127globals), the "``.apple_types``" (names of all types that are defined), and
1128the "``.apple_namespaces``" (all namespaces), we currently set the ``Atom``
1129array to be:
1130
1131.. code-block:: c
1132
1133  HeaderData.atom_count = 1;
1134  HeaderData.atoms[0].type = eAtomTypeDIEOffset;
1135  HeaderData.atoms[0].form = DW_FORM_data4;
1136
1137This defines the contents to be the DIE offset (eAtomTypeDIEOffset) that is
1138encoded as a 32 bit value (DW_FORM_data4).  This allows a single name to have
1139multiple matching DIEs in a single file, which could come up with an inlined
1140function for instance.  Future tables could include more information about the
1141DIE such as flags indicating if the DIE is a function, method, block,
1142or inlined.
1143
1144The KeyType for the DWARF table is a 32 bit string table offset into the
1145".debug_str" table.  The ".debug_str" is the string table for the DWARF which
1146may already contain copies of all of the strings.  This helps make sure, with
1147help from the compiler, that we reuse the strings between all of the DWARF
1148sections and keeps the hash table size down.  Another benefit to having the
1149compiler generate all strings as DW_FORM_strp in the debug info, is that
1150DWARF parsing can be made much faster.
1151
1152After a lookup is made, we get an offset into the hash data.  The hash data
1153needs to be able to deal with 32 bit hash collisions, so the chunk of data
1154at the offset in the hash data consists of a triple:
1155
1156.. code-block:: c
1157
1158  uint32_t str_offset
1159  uint32_t hash_data_count
1160  HashData[hash_data_count]
1161
1162If "str_offset" is zero, then the bucket contents are done. 99.9% of the
1163hash data chunks contain a single item (no 32 bit hash collision):
1164
1165.. code-block:: none
1166
1167  .------------.
1168  | 0x00001023 | uint32_t KeyType (.debug_str[0x0001023] => "main")
1169  | 0x00000004 | uint32_t HashData count
1170  | 0x........ | uint32_t HashData[0] DIE offset
1171  | 0x........ | uint32_t HashData[1] DIE offset
1172  | 0x........ | uint32_t HashData[2] DIE offset
1173  | 0x........ | uint32_t HashData[3] DIE offset
1174  | 0x00000000 | uint32_t KeyType (end of hash chain)
1175  `------------'
1176
1177If there are collisions, you will have multiple valid string offsets:
1178
1179.. code-block:: none
1180
1181  .------------.
1182  | 0x00001023 | uint32_t KeyType (.debug_str[0x0001023] => "main")
1183  | 0x00000004 | uint32_t HashData count
1184  | 0x........ | uint32_t HashData[0] DIE offset
1185  | 0x........ | uint32_t HashData[1] DIE offset
1186  | 0x........ | uint32_t HashData[2] DIE offset
1187  | 0x........ | uint32_t HashData[3] DIE offset
1188  | 0x00002023 | uint32_t KeyType (.debug_str[0x0002023] => "print")
1189  | 0x00000002 | uint32_t HashData count
1190  | 0x........ | uint32_t HashData[0] DIE offset
1191  | 0x........ | uint32_t HashData[1] DIE offset
1192  | 0x00000000 | uint32_t KeyType (end of hash chain)
1193  `------------'
1194
1195Current testing with real world C++ binaries has shown that there is around 1
119632 bit hash collision per 100,000 name entries.
1197
1198Contents
1199^^^^^^^^
1200
1201As we said, we want to strictly define exactly what is included in the
1202different tables.  For DWARF, we have 3 tables: "``.apple_names``",
1203"``.apple_types``", and "``.apple_namespaces``".
1204
1205"``.apple_names``" sections should contain an entry for each DWARF DIE whose
1206``DW_TAG`` is a ``DW_TAG_label``, ``DW_TAG_inlined_subroutine``, or
1207``DW_TAG_subprogram`` that has address attributes: ``DW_AT_low_pc``,
1208``DW_AT_high_pc``, ``DW_AT_ranges`` or ``DW_AT_entry_pc``.  It also contains
1209``DW_TAG_variable`` DIEs that have a ``DW_OP_addr`` in the location (global and
1210static variables).  All global and static variables should be included,
1211including those scoped within functions and classes.  For example using the
1212following code:
1213
1214.. code-block:: c
1215
1216  static int var = 0;
1217
1218  void f ()
1219  {
1220    static int var = 0;
1221  }
1222
1223Both of the static ``var`` variables would be included in the table.  All
1224functions should emit both their full names and their basenames.  For C or C++,
1225the full name is the mangled name (if available) which is usually in the
1226``DW_AT_MIPS_linkage_name`` attribute, and the ``DW_AT_name`` contains the
1227function basename.  If global or static variables have a mangled name in a
1228``DW_AT_MIPS_linkage_name`` attribute, this should be emitted along with the
1229simple name found in the ``DW_AT_name`` attribute.
1230
1231"``.apple_types``" sections should contain an entry for each DWARF DIE whose
1232tag is one of:
1233
1234* DW_TAG_array_type
1235* DW_TAG_class_type
1236* DW_TAG_enumeration_type
1237* DW_TAG_pointer_type
1238* DW_TAG_reference_type
1239* DW_TAG_string_type
1240* DW_TAG_structure_type
1241* DW_TAG_subroutine_type
1242* DW_TAG_typedef
1243* DW_TAG_union_type
1244* DW_TAG_ptr_to_member_type
1245* DW_TAG_set_type
1246* DW_TAG_subrange_type
1247* DW_TAG_base_type
1248* DW_TAG_const_type
1249* DW_TAG_file_type
1250* DW_TAG_namelist
1251* DW_TAG_packed_type
1252* DW_TAG_volatile_type
1253* DW_TAG_restrict_type
1254* DW_TAG_interface_type
1255* DW_TAG_unspecified_type
1256* DW_TAG_shared_type
1257
1258Only entries with a ``DW_AT_name`` attribute are included, and the entry must
1259not be a forward declaration (``DW_AT_declaration`` attribute with a non-zero
1260value).  For example, using the following code:
1261
1262.. code-block:: c
1263
1264  int main ()
1265  {
1266    int *b = 0;
1267    return *b;
1268  }
1269
1270We get a few type DIEs:
1271
1272.. code-block:: none
1273
1274  0x00000067:     TAG_base_type [5]
1275                  AT_encoding( DW_ATE_signed )
1276                  AT_name( "int" )
1277                  AT_byte_size( 0x04 )
1278
1279  0x0000006e:     TAG_pointer_type [6]
1280                  AT_type( {0x00000067} ( int ) )
1281                  AT_byte_size( 0x08 )
1282
1283The DW_TAG_pointer_type is not included because it does not have a ``DW_AT_name``.
1284
1285"``.apple_namespaces``" section should contain all ``DW_TAG_namespace`` DIEs.
1286If we run into a namespace that has no name this is an anonymous namespace, and
1287the name should be output as "``(anonymous namespace)``" (without the quotes).
1288Why?  This matches the output of the ``abi::cxa_demangle()`` that is in the
1289standard C++ library that demangles mangled names.
1290
1291
1292Language Extensions and File Format Changes
1293^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1294
1295Objective-C Extensions
1296""""""""""""""""""""""
1297
1298"``.apple_objc``" section should contain all ``DW_TAG_subprogram`` DIEs for an
1299Objective-C class.  The name used in the hash table is the name of the
1300Objective-C class itself.  If the Objective-C class has a category, then an
1301entry is made for both the class name without the category, and for the class
1302name with the category.  So if we have a DIE at offset 0x1234 with a name of
1303method "``-[NSString(my_additions) stringWithSpecialString:]``", we would add
1304an entry for "``NSString``" that points to DIE 0x1234, and an entry for
1305"``NSString(my_additions)``" that points to 0x1234.  This allows us to quickly
1306track down all Objective-C methods for an Objective-C class when doing
1307expressions.  It is needed because of the dynamic nature of Objective-C where
1308anyone can add methods to a class.  The DWARF for Objective-C methods is also
1309emitted differently from C++ classes where the methods are not usually
1310contained in the class definition, they are scattered about across one or more
1311compile units.  Categories can also be defined in different shared libraries.
1312So we need to be able to quickly find all of the methods and class functions
1313given the Objective-C class name, or quickly find all methods and class
1314functions for a class + category name.  This table does not contain any
1315selector names, it just maps Objective-C class names (or class names +
1316category) to all of the methods and class functions.  The selectors are added
1317as function basenames in the "``.debug_names``" section.
1318
1319In the "``.apple_names``" section for Objective-C functions, the full name is
1320the entire function name with the brackets ("``-[NSString
1321stringWithCString:]``") and the basename is the selector only
1322("``stringWithCString:``").
1323
1324Mach-O Changes
1325""""""""""""""
1326
1327The sections names for the apple hash tables are for non-mach-o files.  For
1328mach-o files, the sections should be contained in the ``__DWARF`` segment with
1329names as follows:
1330
1331* "``.apple_names``" -> "``__apple_names``"
1332* "``.apple_types``" -> "``__apple_types``"
1333* "``.apple_namespaces``" -> "``__apple_namespac``" (16 character limit)
1334* "``.apple_objc``" -> "``__apple_objc``"
1335
1336