1 ================================
2 Source Level Debugging with LLVM
3 ================================
11 This document is the central repository for all information pertaining to debug
12 information in LLVM. It describes the :ref:`actual format that the LLVM debug
13 information takes <format>`, which is useful for those interested in creating
14 front-ends or dealing directly with the information. Further, this document
15 provides specific examples of what debug information for C/C++ looks like.
17 Philosophy behind LLVM debugging information
18 --------------------------------------------
20 The idea of the LLVM debugging information is to capture how the important
21 pieces of the source-language's Abstract Syntax Tree map onto LLVM code.
22 Several design aspects have shaped the solution that appears here. The
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.
29 * LLVM optimizations should interact in :ref:`well-defined and easily described
30 ways <intro_debugopt>` with the debugging information.
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.
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.
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.
45 The approach used by the LLVM implementation is to use a small set of
46 :ref:`intrinsic functions <format_common_intrinsics>` to define a mapping
47 between LLVM program objects and the source-level objects. The description of
48 the source-level program is maintained in LLVM metadata in an
49 :ref:`implementation-defined format <ccxx_frontend>` (the C/C++ front-end
50 currently uses working draft 7 of the `DWARF 3 standard
51 <http://www.eagercon.com/dwarf/dwarf3std.htm>`_).
53 When a program is being debugged, a debugger interacts with the user and turns
54 the stored debug information into source-language specific information. As
55 such, a debugger must be aware of the source-language, and is thus tied to a
56 specific language or family of languages.
58 Debug information consumers
59 ---------------------------
61 The role of debug information is to provide meta information normally stripped
62 away during the compilation process. This meta information provides an LLVM
63 user a relationship between generated code and the original program source
66 Currently, debug information is consumed by DwarfDebug to produce dwarf
67 information used by the gdb debugger. Other targets could use the same
68 information to produce stabs or other debug forms.
70 It would also be reasonable to use debug information to feed profiling tools
71 for analysis of generated code, or, tools for reconstructing the original
72 source from generated code.
74 TODO - expound a bit more.
78 Debugging optimized code
79 ------------------------
81 An extremely high priority of LLVM debugging information is to make it interact
82 well with optimizations and analysis. In particular, the LLVM debug
83 information provides the following guarantees:
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
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
99 * LLVM debug information does not prevent optimizations from
100 happening (for example inlining, basic block reordering/merging/cleanup,
101 tail duplication, etc).
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.
108 Basically, the debug information allows you to compile a program with
109 "``-O0 -g``" and get full debug information, allowing you to arbitrarily modify
110 the program as it executes from a debugger. Compiling a program with
111 "``-O3 -g``" gives you full debug information that is always available and
112 accurate for reading (e.g., you get accurate stack traces despite tail call
113 elimination and inlining), but you might lose the ability to modify the program
114 and call functions where were optimized out of the program, or inlined away
117 :ref:`LLVM test suite <test-suite-quickstart>` provides a framework to test
118 optimizer's handling of debugging information. It can be run like this:
122 % cd llvm/projects/test-suite/MultiSource/Benchmarks # or some other level
125 This will test impact of debugging information on optimization passes. If
126 debugging information influences optimization passes then it will be reported
127 as a failure. See :doc:`TestingGuide` for more information on LLVM test
128 infrastructure and how to run various tests.
132 Debugging information format
133 ============================
135 LLVM debugging information has been carefully designed to make it possible for
136 the optimizer to optimize the program and debugging information without
137 necessarily having to know anything about debugging information. In
138 particular, the use of metadata avoids duplicated debugging information from
139 the beginning, and the global dead code elimination pass automatically deletes
140 debugging information for a function if it decides to delete the function.
142 To do this, most of the debugging information (descriptors for types,
143 variables, functions, source files, etc) is inserted by the language front-end
144 in the form of LLVM metadata.
146 Debug information is designed to be agnostic about the target debugger and
147 debugging information representation (e.g. DWARF/Stabs/etc). It uses a generic
148 pass to decode the information that represents variables, types, functions,
149 namespaces, etc: this allows for arbitrary source-language semantics and
150 type-systems to be used, as long as there is a module written for the target
151 debugger to interpret the information.
153 To provide basic functionality, the LLVM debugger does have to make some
154 assumptions about the source-level language being debugged, though it keeps
155 these to a minimum. The only common features that the LLVM debugger assumes
156 exist are :ref:`source files <format_files>`, and :ref:`program objects
157 <format_global_variables>`. These abstract objects are used by a debugger to
158 form stack traces, show information about local variables, etc.
160 This section of the documentation first describes the representation aspects
161 common to any source-language. :ref:`ccxx_frontend` describes the data layout
162 conventions used by the C and C++ front-ends.
164 Debug information descriptors
165 -----------------------------
167 In consideration of the complexity and volume of debug information, LLVM
168 provides a specification for well formed debug descriptors.
170 Consumers of LLVM debug information expect the descriptors for program objects
171 to start in a canonical format, but the descriptors can include additional
172 information appended at the end that is source-language specific. All debugging
173 information objects start with a tag to indicate what type of object it is.
174 The source-language is allowed to define its own objects, by using unreserved
175 tag numbers. We recommend using with tags in the range 0x1000 through 0x2000
176 (there is a defined ``enum DW_TAG_user_base = 0x1000``.)
178 The fields of debug descriptors used internally by LLVM are restricted to only
179 the simple data types ``i32``, ``i1``, ``float``, ``double``, ``mdstring`` and
189 <a name="LLVMDebugVersion">The first field of a descriptor is always an
190 ``i32`` containing a tag value identifying the content of the descriptor.
191 The remaining fields are specific to the descriptor. The values of tags are
192 loosely bound to the tag values of DWARF information entries. However, that
193 does not restrict the use of the information supplied to DWARF targets.
195 The details of the various descriptors follow.
197 Compile unit descriptors
198 ^^^^^^^^^^^^^^^^^^^^^^^^
203 i32, ;; Tag = 17 (DW_TAG_compile_unit)
204 metadata, ;; Source directory (including trailing slash) & file pair
205 i32, ;; DWARF language identifier (ex. DW_LANG_C89)
206 metadata ;; Producer (ex. "4.0.1 LLVM (LLVM research group)")
207 i1, ;; True if this is optimized.
209 i32 ;; Runtime version
210 metadata ;; List of enums types
211 metadata ;; List of retained types
212 metadata ;; List of subprograms
213 metadata ;; List of global variables
214 metadata ;; List of imported entities
215 metadata ;; Split debug filename
216 i32 ;; Debug info emission kind (1 = Full Debug Info, 2 = Line Tables Only)
219 These descriptors contain a source language ID for the file (we use the DWARF
220 3.0 ID numbers, such as ``DW_LANG_C89``, ``DW_LANG_C_plus_plus``,
221 ``DW_LANG_Cobol74``, etc), a reference to a metadata node containing a pair of
222 strings for the source file name and the working directory, as well as an
223 identifier string for the compiler that produced it.
225 Compile unit descriptors provide the root context for objects declared in a
226 specific compilation unit. File descriptors are defined using this context.
227 These descriptors are collected by a named metadata ``!llvm.dbg.cu``. They
228 keep track of subprograms, global variables, type information, and imported
229 entities (declarations and namespaces).
239 i32, ;; Tag = 41 (DW_TAG_file_type)
240 metadata, ;; Source directory (including trailing slash) & file pair
243 These descriptors contain information for a file. Global variables and top
244 level functions would be defined using this context. File descriptors also
245 provide context for source line correspondence.
247 Each input file is encoded as a separate file descriptor in LLVM debugging
250 .. _format_global_variables:
252 Global variable descriptors
253 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
258 i32, ;; Tag = 52 (DW_TAG_variable)
259 i32, ;; Unused field.
260 metadata, ;; Reference to context descriptor
262 metadata, ;; Display name (fully qualified C++ name)
263 metadata, ;; MIPS linkage name (for C++)
264 metadata, ;; Reference to file where defined
265 i32, ;; Line number where defined
266 metadata, ;; Reference to type descriptor
267 i1, ;; True if the global is local to compile unit (static)
268 i1, ;; True if the global is defined in the compile unit (not extern)
269 {}*, ;; Reference to the global variable
270 metadata, ;; The static member declaration, if any
273 These descriptors provide debug information about global variables. They
274 provide details such as name, type and where the variable is defined. All
275 global variables are collected inside the named metadata ``!llvm.dbg.cu``.
277 .. _format_subprograms:
279 Subprogram descriptors
280 ^^^^^^^^^^^^^^^^^^^^^^
285 i32, ;; Tag = 46 (DW_TAG_subprogram)
286 metadata, ;; Source directory (including trailing slash) & file pair
287 metadata, ;; Reference to context descriptor
289 metadata, ;; Display name (fully qualified C++ name)
290 metadata, ;; MIPS linkage name (for C++)
291 i32, ;; Line number where defined
292 metadata, ;; Reference to type descriptor
293 i1, ;; True if the global is local to compile unit (static)
294 i1, ;; True if the global is defined in the compile unit (not extern)
295 i32, ;; Virtuality, e.g. dwarf::DW_VIRTUALITY__virtual
296 i32, ;; Index into a virtual function
297 metadata, ;; indicates which base type contains the vtable pointer for the
299 i32, ;; Flags - Artificial, Private, Protected, Explicit, Prototyped.
301 {}*, ;; Reference to the LLVM function
302 metadata, ;; Lists function template parameters
303 metadata, ;; Function declaration descriptor
304 metadata, ;; List of function variables
305 i32 ;; Line number where the scope of the subprogram begins
308 These descriptors provide debug information about functions, methods and
309 subprograms. They provide details such as name, return types and the source
310 location where the subprogram is defined.
318 i32, ;; Tag = 11 (DW_TAG_lexical_block)
319 metadata, ;; Source directory (including trailing slash) & file pair
320 metadata, ;; Reference to context descriptor
322 i32, ;; Column number
323 i32, ;; DWARF path discriminator value
324 i32 ;; Unique ID to identify blocks from a template function
327 This descriptor provides debug information about nested blocks within a
328 subprogram. The line number and column numbers are used to dinstinguish two
329 lexical blocks at same depth.
334 i32, ;; Tag = 11 (DW_TAG_lexical_block)
335 metadata, ;; Source directory (including trailing slash) & file pair
336 metadata ;; Reference to the scope we're annotating with a file change
339 This descriptor provides a wrapper around a lexical scope to handle file
340 changes in the middle of a lexical block.
342 .. _format_basic_type:
344 Basic type descriptors
345 ^^^^^^^^^^^^^^^^^^^^^^
350 i32, ;; Tag = 36 (DW_TAG_base_type)
351 metadata, ;; Source directory (including trailing slash) & file pair (may be null)
352 metadata, ;; Reference to context
353 metadata, ;; Name (may be "" for anonymous types)
354 i32, ;; Line number where defined (may be 0)
356 i64, ;; Alignment in bits
357 i64, ;; Offset in bits
359 i32 ;; DWARF type encoding
362 These descriptors define primitive types used in the code. Example ``int``,
363 ``bool`` and ``float``. The context provides the scope of the type, which is
364 usually the top level. Since basic types are not usually user defined the
365 context and line number can be left as NULL and 0. The size, alignment and
366 offset are expressed in bits and can be 64 bit values. The alignment is used
367 to round the offset when embedded in a :ref:`composite type
368 <format_composite_type>` (example to keep float doubles on 64 bit boundaries).
369 The offset is the bit offset if embedded in a :ref:`composite type
370 <format_composite_type>`.
372 The type encoding provides the details of the type. The values are typically
373 one of the following:
381 DW_ATE_signed_char = 6
383 DW_ATE_unsigned_char = 8
385 .. _format_derived_type:
387 Derived type descriptors
388 ^^^^^^^^^^^^^^^^^^^^^^^^
393 i32, ;; Tag (see below)
394 metadata, ;; Source directory (including trailing slash) & file pair (may be null)
395 metadata, ;; Reference to context
396 metadata, ;; Name (may be "" for anonymous types)
397 i32, ;; Line number where defined (may be 0)
399 i64, ;; Alignment in bits
400 i64, ;; Offset in bits
401 i32, ;; Flags to encode attributes, e.g. private
402 metadata, ;; Reference to type derived from
403 metadata, ;; (optional) Name of the Objective C property associated with
404 ;; Objective-C an ivar, or the type of which this
405 ;; pointer-to-member is pointing to members of.
406 metadata, ;; (optional) Name of the Objective C property getter selector.
407 metadata, ;; (optional) Name of the Objective C property setter selector.
408 i32 ;; (optional) Objective C property attributes.
411 These descriptors are used to define types derived from other types. The value
412 of the tag varies depending on the meaning. The following are possible tag
417 DW_TAG_formal_parameter = 5
419 DW_TAG_pointer_type = 15
420 DW_TAG_reference_type = 16
422 DW_TAG_ptr_to_member_type = 31
423 DW_TAG_const_type = 38
424 DW_TAG_volatile_type = 53
425 DW_TAG_restrict_type = 55
427 ``DW_TAG_member`` is used to define a member of a :ref:`composite type
428 <format_composite_type>` or :ref:`subprogram <format_subprograms>`. The type
429 of the member is the :ref:`derived type <format_derived_type>`.
430 ``DW_TAG_formal_parameter`` is used to define a member which is a formal
431 argument of a subprogram.
433 ``DW_TAG_typedef`` is used to provide a name for the derived type.
435 ``DW_TAG_pointer_type``, ``DW_TAG_reference_type``, ``DW_TAG_const_type``,
436 ``DW_TAG_volatile_type`` and ``DW_TAG_restrict_type`` are used to qualify the
437 :ref:`derived type <format_derived_type>`.
439 :ref:`Derived type <format_derived_type>` location can be determined from the
440 context and line number. The size, alignment and offset are expressed in bits
441 and can be 64 bit values. The alignment is used to round the offset when
442 embedded in a :ref:`composite type <format_composite_type>` (example to keep
443 float doubles on 64 bit boundaries.) The offset is the bit offset if embedded
444 in a :ref:`composite type <format_composite_type>`.
446 Note that the ``void *`` type is expressed as a type derived from NULL.
448 .. _format_composite_type:
450 Composite type descriptors
451 ^^^^^^^^^^^^^^^^^^^^^^^^^^
456 i32, ;; Tag (see below)
457 metadata, ;; Source directory (including trailing slash) & file pair (may be null)
458 metadata, ;; Reference to context
459 metadata, ;; Name (may be "" for anonymous types)
460 i32, ;; Line number where defined (may be 0)
462 i64, ;; Alignment in bits
463 i64, ;; Offset in bits
465 metadata, ;; Reference to type derived from
466 metadata, ;; Reference to array of member descriptors
467 i32, ;; Runtime languages
468 metadata, ;; Base type containing the vtable pointer for this type
469 metadata, ;; Template parameters
470 metadata ;; A unique identifier for type uniquing purpose (may be null)
473 These descriptors are used to define types that are composed of 0 or more
474 elements. The value of the tag varies depending on the meaning. The following
475 are possible tag values:
479 DW_TAG_array_type = 1
480 DW_TAG_enumeration_type = 4
481 DW_TAG_structure_type = 19
482 DW_TAG_union_type = 23
483 DW_TAG_subroutine_type = 21
484 DW_TAG_inheritance = 28
486 The vector flag indicates that an array type is a native packed vector.
488 The members of array types (tag = ``DW_TAG_array_type``) are
489 :ref:`subrange descriptors <format_subrange>`, each
490 representing the range of subscripts at that level of indexing.
492 The members of enumeration types (tag = ``DW_TAG_enumeration_type``) are
493 :ref:`enumerator descriptors <format_enumerator>`, each representing the
494 definition of enumeration value for the set. All enumeration type descriptors
495 are collected inside the named metadata ``!llvm.dbg.cu``.
497 The members of structure (tag = ``DW_TAG_structure_type``) or union (tag =
498 ``DW_TAG_union_type``) types are any one of the :ref:`basic
499 <format_basic_type>`, :ref:`derived <format_derived_type>` or :ref:`composite
500 <format_composite_type>` type descriptors, each representing a field member of
501 the structure or union.
503 For C++ classes (tag = ``DW_TAG_structure_type``), member descriptors provide
504 information about base classes, static members and member functions. If a
505 member is a :ref:`derived type descriptor <format_derived_type>` and has a tag
506 of ``DW_TAG_inheritance``, then the type represents a base class. If the member
507 of is a :ref:`global variable descriptor <format_global_variables>` then it
508 represents a static member. And, if the member is a :ref:`subprogram
509 descriptor <format_subprograms>` then it represents a member function. For
510 static members and member functions, ``getName()`` returns the members link or
511 the C++ mangled name. ``getDisplayName()`` the simplied version of the name.
513 The first member of subroutine (tag = ``DW_TAG_subroutine_type``) type elements
514 is the return type for the subroutine. The remaining elements are the formal
515 arguments to the subroutine.
517 :ref:`Composite type <format_composite_type>` location can be determined from
518 the context and line number. The size, alignment and offset are expressed in
519 bits and can be 64 bit values. The alignment is used to round the offset when
520 embedded in a :ref:`composite type <format_composite_type>` (as an example, to
521 keep float doubles on 64 bit boundaries). The offset is the bit offset if
522 embedded in a :ref:`composite type <format_composite_type>`.
532 i32, ;; Tag = 33 (DW_TAG_subrange_type)
537 These descriptors are used to define ranges of array subscripts for an array
538 :ref:`composite type <format_composite_type>`. The low value defines the lower
539 bounds typically zero for C/C++. The high value is the upper bounds. Values
540 are 64 bit. ``High - Low + 1`` is the size of the array. If ``Low > High``
541 the array bounds are not included in generated debugging information.
543 .. _format_enumerator:
545 Enumerator descriptors
546 ^^^^^^^^^^^^^^^^^^^^^^
551 i32, ;; Tag = 40 (DW_TAG_enumerator)
556 These descriptors are used to define members of an enumeration :ref:`composite
557 type <format_composite_type>`, it associates the name to the value.
565 i32, ;; Tag (see below)
568 metadata, ;; Reference to file where defined
569 i32, ;; 24 bit - Line number where defined
570 ;; 8 bit - Argument number. 1 indicates 1st argument.
571 metadata, ;; Reference to the type descriptor
573 metadata ;; (optional) Reference to inline location
574 metadata ;; (optional) Reference to a complex expression (see below)
577 These descriptors are used to define variables local to a sub program. The
578 value of the tag depends on the usage of the variable:
582 DW_TAG_auto_variable = 256
583 DW_TAG_arg_variable = 257
585 An auto variable is any variable declared in the body of the function. An
586 argument variable is any variable that appears as a formal argument to the
589 The context is either the subprogram or block where the variable is defined.
590 Name the source variable name. Context and line indicate where the variable
591 was defined. Type descriptor defines the declared type of the variable.
593 The ``OpPiece`` operator is used for (typically larger aggregate)
594 variables that are fragmented across several locations. It takes two
595 i32 arguments, an offset and a size in bytes to describe which piece
596 of the variable is at this location.
599 .. _format_common_intrinsics:
601 Debugger intrinsic functions
602 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
604 LLVM uses several intrinsic functions (name prefixed with "``llvm.dbg``") to
605 provide debug information at various points in generated code.
612 void %llvm.dbg.declare(metadata, metadata)
614 This intrinsic provides information about a local element (e.g., variable).
615 The first argument is metadata holding the alloca for the variable. The second
616 argument is metadata containing a description of the variable.
623 void %llvm.dbg.value(metadata, i64, metadata)
625 This intrinsic provides information when a user source variable is set to a new
626 value. The first argument is the new value (wrapped as metadata). The second
627 argument is the offset in the user source variable where the new value is
628 written. The third argument is metadata containing a description of the user
631 Object lifetimes and scoping
632 ============================
634 In many languages, the local variables in functions can have their lifetimes or
635 scopes limited to a subset of a function. In the C family of languages, for
636 example, variables are only live (readable and writable) within the source
637 block that they are defined in. In functional languages, values are only
638 readable after they have been defined. Though this is a very obvious concept,
639 it is non-trivial to model in LLVM, because it has no notion of scoping in this
640 sense, and does not want to be tied to a language's scoping rules.
642 In order to handle this, the LLVM debug format uses the metadata attached to
643 llvm instructions to encode line number and scoping information. Consider the
644 following C fragment, for example:
658 Compiled to LLVM, this function would be represented like this:
662 define void @foo() #0 {
664 %X = alloca i32, align 4
665 %Y = alloca i32, align 4
666 %Z = alloca i32, align 4
667 call void @llvm.dbg.declare(metadata !{i32* %X}, metadata !10), !dbg !12
668 ; [debug line = 2:7] [debug variable = X]
669 store i32 21, i32* %X, align 4, !dbg !12
670 call void @llvm.dbg.declare(metadata !{i32* %Y}, metadata !13), !dbg !14
671 ; [debug line = 3:7] [debug variable = Y]
672 store i32 22, i32* %Y, align 4, !dbg !14
673 call void @llvm.dbg.declare(metadata !{i32* %Z}, metadata !15), !dbg !17
674 ; [debug line = 5:9] [debug variable = Z]
675 store i32 23, i32* %Z, align 4, !dbg !17
676 %0 = load i32* %X, align 4, !dbg !18
678 store i32 %0, i32* %Z, align 4, !dbg !18
679 %1 = load i32* %Y, align 4, !dbg !19
681 store i32 %1, i32* %X, align 4, !dbg !19
685 ; Function Attrs: nounwind readnone
686 declare void @llvm.dbg.declare(metadata, metadata) #1
688 attributes #0 = { nounwind ssp uwtable "less-precise-fpmad"="false"
689 "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf"
690 "no-infs-fp-math"="false" "no-nans-fp-math"="false"
691 "stack-protector-buffer-size"="8" "unsafe-fp-math"="false"
692 "use-soft-float"="false" }
693 attributes #1 = { nounwind readnone }
696 !llvm.module.flags = !{!8}
699 !0 = metadata !{i32 786449, metadata !1, i32 12,
700 metadata !"clang version 3.4 (trunk 193128) (llvm/trunk 193139)",
701 i1 false, metadata !"", i32 0, metadata !2, metadata !2, metadata !3,
702 metadata !2, metadata !2, metadata !""} ; [ DW_TAG_compile_unit ] \
703 [/private/tmp/foo.c] \
705 !1 = metadata !{metadata !"t.c", metadata !"/private/tmp"}
706 !2 = metadata !{i32 0}
707 !3 = metadata !{metadata !4}
708 !4 = metadata !{i32 786478, metadata !1, metadata !5, metadata !"foo",
709 metadata !"foo", metadata !"", i32 1, metadata !6,
710 i1 false, i1 true, i32 0, i32 0, null, i32 0, i1 false,
711 void ()* @foo, null, null, metadata !2, i32 1}
712 ; [ DW_TAG_subprogram ] [line 1] [def] [foo]
713 !5 = metadata !{i32 786473, metadata !1} ; [ DW_TAG_file_type ] \
715 !6 = metadata !{i32 786453, i32 0, null, metadata !"", i32 0, i64 0, i64 0,
716 i64 0, i32 0, null, metadata !7, i32 0, null, null, null}
717 ; [ DW_TAG_subroutine_type ] \
718 [line 0, size 0, align 0, offset 0] [from ]
719 !7 = metadata !{null}
720 !8 = metadata !{i32 2, metadata !"Dwarf Version", i32 2}
721 !9 = metadata !{metadata !"clang version 3.4 (trunk 193128) (llvm/trunk 193139)"}
722 !10 = metadata !{i32 786688, metadata !4, metadata !"X", metadata !5, i32 2,
723 metadata !11, i32 0, i32 0} ; [ DW_TAG_auto_variable ] [X] \
725 !11 = metadata !{i32 786468, null, null, metadata !"int", i32 0, i64 32,
726 i64 32, i64 0, i32 0, i32 5} ; [ DW_TAG_base_type ] [int] \
727 [line 0, size 32, align 32, offset 0, enc DW_ATE_signed]
728 !12 = metadata !{i32 2, i32 0, metadata !4, null}
729 !13 = metadata !{i32 786688, metadata !4, metadata !"Y", metadata !5, i32 3,
730 metadata !11, i32 0, i32 0} ; [ DW_TAG_auto_variable ] [Y] \
732 !14 = metadata !{i32 3, i32 0, metadata !4, null}
733 !15 = metadata !{i32 786688, metadata !16, metadata !"Z", metadata !5, i32 5,
734 metadata !11, i32 0, i32 0} ; [ DW_TAG_auto_variable ] [Z] \
736 !16 = metadata !{i32 786443, metadata !1, metadata !4, i32 4, i32 0, i32 0,
738 ; [ DW_TAG_lexical_block ] [/private/tmp/t.c]
739 !17 = metadata !{i32 5, i32 0, metadata !16, null}
740 !18 = metadata !{i32 6, i32 0, metadata !16, null}
741 !19 = metadata !{i32 8, i32 0, metadata !4, null} ; [ DW_TAG_imported_declaration ]
742 !20 = metadata !{i32 9, i32 0, metadata !4, null}
744 This example illustrates a few important details about LLVM debugging
745 information. In particular, it shows how the ``llvm.dbg.declare`` intrinsic and
746 location information, which are attached to an instruction, are applied
747 together to allow a debugger to analyze the relationship between statements,
748 variable definitions, and the code used to implement the function.
752 call void @llvm.dbg.declare(metadata !{i32* %X}, metadata !10), !dbg !12
753 ; [debug line = 2:7] [debug variable = X]
755 The first intrinsic ``%llvm.dbg.declare`` encodes debugging information for the
756 variable ``X``. The metadata ``!dbg !12`` attached to the intrinsic provides
757 scope information for the variable ``X``.
761 !12 = metadata !{i32 2, i32 0, metadata !4, null}
762 !4 = metadata !{i32 786478, metadata !1, metadata !5, metadata !"foo",
763 metadata !"foo", metadata !"", i32 1, metadata !6,
764 i1 false, i1 true, i32 0, i32 0, null, i32 0, i1 false,
765 void ()* @foo, null, null, metadata !2, i32 1}
766 ; [ DW_TAG_subprogram ] [line 1] [def] [foo]
768 Here ``!12`` is metadata providing location information. It has four fields:
769 line number, column number, scope, and original scope. The original scope
770 represents inline location if this instruction is inlined inside a caller, and
771 is null otherwise. In this example, scope is encoded by ``!4``, a
772 :ref:`subprogram descriptor <format_subprograms>`. This way the location
773 information attached to the intrinsics indicates that the variable ``X`` is
774 declared at line number 2 at a function level scope in function ``foo``.
776 Now lets take another example.
780 call void @llvm.dbg.declare(metadata !{i32* %Z}, metadata !15), !dbg !17
781 ; [debug line = 5:9] [debug variable = Z]
783 The third intrinsic ``%llvm.dbg.declare`` encodes debugging information for
784 variable ``Z``. The metadata ``!dbg !17`` attached to the intrinsic provides
785 scope information for the variable ``Z``.
789 !16 = metadata !{i32 786443, metadata !1, metadata !4, i32 4, i32 0, i32 0,
791 ; [ DW_TAG_lexical_block ] [/private/tmp/t.c]
792 !17 = metadata !{i32 5, i32 0, metadata !16, null}
794 Here ``!15`` indicates that ``Z`` is declared at line number 5 and
795 column number 0 inside of lexical scope ``!16``. The lexical scope itself
796 resides inside of subprogram ``!4`` described above.
798 The scope information attached with each instruction provides a straightforward
799 way to find instructions covered by a scope.
803 C/C++ front-end specific debug information
804 ==========================================
806 The C and C++ front-ends represent information about the program in a format
807 that is effectively identical to `DWARF 3.0
808 <http://www.eagercon.com/dwarf/dwarf3std.htm>`_ in terms of information
809 content. This allows code generators to trivially support native debuggers by
810 generating standard dwarf information, and contains enough information for
811 non-dwarf targets to translate it as needed.
813 This section describes the forms used to represent C and C++ programs. Other
814 languages could pattern themselves after this (which itself is tuned to
815 representing programs in the same way that DWARF 3 does), or they could choose
816 to provide completely different forms if they don't fit into the DWARF model.
817 As support for debugging information gets added to the various LLVM
818 source-language front-ends, the information used should be documented here.
820 The following sections provide examples of various C/C++ constructs and the
821 debug information that would best describe those constructs.
823 C/C++ source file information
824 -----------------------------
826 Given the source files ``MySource.cpp`` and ``MyHeader.h`` located in the
827 directory ``/Users/mine/sources``, the following code:
831 #include "MyHeader.h"
833 int main(int argc, char *argv[]) {
837 a C/C++ front-end would generate the following descriptors:
843 ;; Define the compile unit for the main source file "/Users/mine/sources/MySource.cpp".
847 metadata !1, ;; File/directory name
848 i32 4, ;; Language Id
849 metadata !"clang version 3.4 ",
850 i1 false, ;; Optimized compile unit
851 metadata !"", ;; Compiler flags
852 i32 0, ;; Runtime version
853 metadata !2, ;; Enumeration types
854 metadata !2, ;; Retained types
855 metadata !3, ;; Subprograms
856 metadata !2, ;; Global variables
857 metadata !2, ;; Imported entities (declarations and namespaces)
858 metadata !"" ;; Split debug filename
859 1, ;; Full debug info
863 ;; Define the file for the file "/Users/mine/sources/MySource.cpp".
866 metadata !"MySource.cpp",
867 metadata !"/Users/mine/sources"
875 ;; Define the file for the file "/Users/mine/sources/Myheader.h"
882 metadata !"./MyHeader.h",
883 metadata !"/Users/mine/sources",
888 ``llvm::Instruction`` provides easy access to metadata attached with an
889 instruction. One can extract line number information encoded in LLVM IR using
890 ``Instruction::getMetadata()`` and ``DILocation::getLineNumber()``.
894 if (MDNode *N = I->getMetadata("dbg")) { // Here I is an LLVM instruction
895 DILocation Loc(N); // DILocation is in DebugInfo.h
896 unsigned Line = Loc.getLineNumber();
897 StringRef File = Loc.getFilename();
898 StringRef Dir = Loc.getDirectory();
901 C/C++ global variable information
902 ---------------------------------
904 Given an integer global variable declared as follows:
910 a C/C++ front-end would generate the following descriptors:
915 ;; Define the global itself.
917 %MyGlobal = global int 100
920 ;; List of debug info of globals
924 ;; Define the compile unit.
929 metadata !"foo.cpp", ;; File
930 metadata !"/Volumes/Data/tmp", ;; Directory
931 metadata !"clang version 3.1 ", ;; Producer
932 i1 true, ;; Deprecated field
933 i1 false, ;; "isOptimized"?
934 metadata !"", ;; Flags
935 i32 0, ;; Runtime Version
936 metadata !1, ;; Enum Types
937 metadata !1, ;; Retained Types
938 metadata !1, ;; Subprograms
939 metadata !3, ;; Global Variables
940 metadata !1, ;; Imported entities
941 "", ;; Split debug filename
942 1, ;; Full debug info
943 } ; [ DW_TAG_compile_unit ]
945 ;; The Array of Global Variables
951 ;; Define the global variable itself.
957 metadata !"MyGlobal", ;; Name
958 metadata !"MyGlobal", ;; Display Name
959 metadata !"", ;; Linkage Name
963 i32 0, ;; IsLocalToUnit
964 i32 1, ;; IsDefinition
965 i32* @MyGlobal, ;; LLVM-IR Value
966 null ;; Static member declaration
967 } ; [ DW_TAG_variable ]
973 metadata !"foo.cpp", ;; File
974 metadata !"/Volumes/Data/tmp", ;; Directory
978 metadata !5 ;; Unused
979 } ; [ DW_TAG_file_type ]
988 metadata !"int", ;; Name
990 i64 32, ;; Size in Bits
991 i64 32, ;; Align in Bits
995 } ; [ DW_TAG_base_type ]
997 C/C++ function information
998 --------------------------
1000 Given a function declared as follows:
1004 int main(int argc, char *argv[]) {
1008 a C/C++ front-end would generate the following descriptors:
1010 .. code-block:: llvm
1013 ;; Define the anchor for subprograms.
1017 metadata !1, ;; File
1018 metadata !1, ;; Context
1019 metadata !"main", ;; Name
1020 metadata !"main", ;; Display name
1021 metadata !"main", ;; Linkage name
1022 i32 1, ;; Line number
1023 metadata !4, ;; Type
1024 i1 false, ;; Is local
1025 i1 true, ;; Is definition
1026 i32 0, ;; Virtuality attribute, e.g. pure virtual function
1027 i32 0, ;; Index into virtual table for C++ methods
1028 i32 0, ;; Type that holds virtual table.
1030 i1 false, ;; True if this function is optimized
1031 Function *, ;; Pointer to llvm::Function
1032 null, ;; Function template parameters
1033 null, ;; List of function variables (emitted when optimizing)
1034 1 ;; Line number of the opening '{' of the function
1037 ;; Define the subprogram itself.
1039 define i32 @main(i32 %argc, i8** %argv) {
1046 The following are the basic type descriptors for C/C++ core types:
1051 .. code-block:: llvm
1057 metadata !"bool", ;; Name
1058 i32 0, ;; Line number
1059 i64 8, ;; Size in Bits
1060 i64 8, ;; Align in Bits
1061 i64 0, ;; Offset in Bits
1069 .. code-block:: llvm
1075 metadata !"char", ;; Name
1076 i32 0, ;; Line number
1077 i64 8, ;; Size in Bits
1078 i64 8, ;; Align in Bits
1079 i64 0, ;; Offset in Bits
1087 .. code-block:: llvm
1093 metadata !"unsigned char",
1094 i32 0, ;; Line number
1095 i64 8, ;; Size in Bits
1096 i64 8, ;; Align in Bits
1097 i64 0, ;; Offset in Bits
1105 .. code-block:: llvm
1111 metadata !"short int",
1112 i32 0, ;; Line number
1113 i64 16, ;; Size in Bits
1114 i64 16, ;; Align in Bits
1115 i64 0, ;; Offset in Bits
1123 .. code-block:: llvm
1129 metadata !"short unsigned int",
1130 i32 0, ;; Line number
1131 i64 16, ;; Size in Bits
1132 i64 16, ;; Align in Bits
1133 i64 0, ;; Offset in Bits
1141 .. code-block:: llvm
1147 metadata !"int", ;; Name
1148 i32 0, ;; Line number
1149 i64 32, ;; Size in Bits
1150 i64 32, ;; Align in Bits
1151 i64 0, ;; Offset in Bits
1159 .. code-block:: llvm
1165 metadata !"unsigned int",
1166 i32 0, ;; Line number
1167 i64 32, ;; Size in Bits
1168 i64 32, ;; Align in Bits
1169 i64 0, ;; Offset in Bits
1177 .. code-block:: llvm
1183 metadata !"long long int",
1184 i32 0, ;; Line number
1185 i64 64, ;; Size in Bits
1186 i64 64, ;; Align in Bits
1187 i64 0, ;; Offset in Bits
1195 .. code-block:: llvm
1201 metadata !"long long unsigned int",
1202 i32 0, ;; Line number
1203 i64 64, ;; Size in Bits
1204 i64 64, ;; Align in Bits
1205 i64 0, ;; Offset in Bits
1213 .. code-block:: llvm
1220 i32 0, ;; Line number
1221 i64 32, ;; Size in Bits
1222 i64 32, ;; Align in Bits
1223 i64 0, ;; Offset in Bits
1231 .. code-block:: llvm
1237 metadata !"double",;; Name
1238 i32 0, ;; Line number
1239 i64 64, ;; Size in Bits
1240 i64 64, ;; Align in Bits
1241 i64 0, ;; Offset in Bits
1249 Given the following as an example of C/C++ derived type:
1253 typedef const int *IntPtr;
1255 a C/C++ front-end would generate the following descriptors:
1257 .. code-block:: llvm
1260 ;; Define the typedef "IntPtr".
1264 metadata !3, ;; File
1265 metadata !1, ;; Context
1266 metadata !"IntPtr", ;; Name
1267 i32 0, ;; Line number
1268 i64 0, ;; Size in bits
1269 i64 0, ;; Align in bits
1270 i64 0, ;; Offset in bits
1272 metadata !4 ;; Derived From type
1275 ;; Define the pointer type.
1281 metadata !"", ;; Name
1282 i32 0, ;; Line number
1283 i64 64, ;; Size in bits
1284 i64 64, ;; Align in bits
1285 i64 0, ;; Offset in bits
1287 metadata !5 ;; Derived From type
1290 ;; Define the const type.
1296 metadata !"", ;; Name
1297 i32 0, ;; Line number
1298 i64 0, ;; Size in bits
1299 i64 0, ;; Align in bits
1300 i64 0, ;; Offset in bits
1302 metadata !6 ;; Derived From type
1305 ;; Define the int type.
1311 metadata !"int", ;; Name
1312 i32 0, ;; Line number
1313 i64 32, ;; Size in bits
1314 i64 32, ;; Align in bits
1315 i64 0, ;; Offset in bits
1320 C/C++ struct/union types
1321 ------------------------
1323 Given the following as an example of C/C++ struct type:
1333 a C/C++ front-end would generate the following descriptors:
1335 .. code-block:: llvm
1338 ;; Define basic type for unsigned int.
1344 metadata !"unsigned int",
1345 i32 0, ;; Line number
1346 i64 32, ;; Size in Bits
1347 i64 32, ;; Align in Bits
1348 i64 0, ;; Offset in Bits
1353 ;; Define composite type for struct Color.
1357 metadata !1, ;; Compile unit
1359 metadata !"Color", ;; Name
1360 i32 1, ;; Line number
1361 i64 96, ;; Size in bits
1362 i64 32, ;; Align in bits
1363 i64 0, ;; Offset in bits
1365 null, ;; Derived From
1366 metadata !3, ;; Elements
1367 i32 0, ;; Runtime Language
1368 null, ;; Base type containing the vtable pointer for this type
1369 null ;; Template parameters
1373 ;; Define the Red field.
1377 metadata !1, ;; File
1378 metadata !1, ;; Context
1379 metadata !"Red", ;; Name
1380 i32 2, ;; Line number
1381 i64 32, ;; Size in bits
1382 i64 32, ;; Align in bits
1383 i64 0, ;; Offset in bits
1385 metadata !5 ;; Derived From type
1389 ;; Define the Green field.
1393 metadata !1, ;; File
1394 metadata !1, ;; Context
1395 metadata !"Green", ;; Name
1396 i32 3, ;; Line number
1397 i64 32, ;; Size in bits
1398 i64 32, ;; Align in bits
1399 i64 32, ;; Offset in bits
1401 metadata !5 ;; Derived From type
1405 ;; Define the Blue field.
1409 metadata !1, ;; File
1410 metadata !1, ;; Context
1411 metadata !"Blue", ;; Name
1412 i32 4, ;; Line number
1413 i64 32, ;; Size in bits
1414 i64 32, ;; Align in bits
1415 i64 64, ;; Offset in bits
1417 metadata !5 ;; Derived From type
1421 ;; Define the array of fields used by the composite type Color.
1423 !3 = metadata !{metadata !4, metadata !6, metadata !7}
1425 C/C++ enumeration types
1426 -----------------------
1428 Given the following as an example of C/C++ enumeration type:
1438 a C/C++ front-end would generate the following descriptors:
1440 .. code-block:: llvm
1443 ;; Define composite type for enum Trees
1447 metadata !1, ;; File
1448 metadata !1, ;; Context
1449 metadata !"Trees", ;; Name
1450 i32 1, ;; Line number
1451 i64 32, ;; Size in bits
1452 i64 32, ;; Align in bits
1453 i64 0, ;; Offset in bits
1455 null, ;; Derived From type
1456 metadata !3, ;; Elements
1457 i32 0 ;; Runtime language
1461 ;; Define the array of enumerators used by composite type Trees.
1463 !3 = metadata !{metadata !4, metadata !5, metadata !6}
1466 ;; Define Spruce enumerator.
1468 !4 = metadata !{i32 786472, metadata !"Spruce", i64 100}
1471 ;; Define Oak enumerator.
1473 !5 = metadata !{i32 786472, metadata !"Oak", i64 200}
1476 ;; Define Maple enumerator.
1478 !6 = metadata !{i32 786472, metadata !"Maple", i64 300}
1480 Debugging information format
1481 ============================
1483 Debugging Information Extension for Objective C Properties
1484 ----------------------------------------------------------
1489 Objective C provides a simpler way to declare and define accessor methods using
1490 declared properties. The language provides features to declare a property and
1491 to let compiler synthesize accessor methods.
1493 The debugger lets developer inspect Objective C interfaces and their instance
1494 variables and class variables. However, the debugger does not know anything
1495 about the properties defined in Objective C interfaces. The debugger consumes
1496 information generated by compiler in DWARF format. The format does not support
1497 encoding of Objective C properties. This proposal describes DWARF extensions to
1498 encode Objective C properties, which the debugger can use to let developers
1499 inspect Objective C properties.
1504 Objective C properties exist separately from class members. A property can be
1505 defined only by "setter" and "getter" selectors, and be calculated anew on each
1506 access. Or a property can just be a direct access to some declared ivar.
1507 Finally it can have an ivar "automatically synthesized" for it by the compiler,
1508 in which case the property can be referred to in user code directly using the
1509 standard C dereference syntax as well as through the property "dot" syntax, but
1510 there is no entry in the ``@interface`` declaration corresponding to this ivar.
1512 To facilitate debugging, these properties we will add a new DWARF TAG into the
1513 ``DW_TAG_structure_type`` definition for the class to hold the description of a
1514 given property, and a set of DWARF attributes that provide said description.
1515 The property tag will also contain the name and declared type of the property.
1517 If there is a related ivar, there will also be a DWARF property attribute placed
1518 in the ``DW_TAG_member`` DIE for that ivar referring back to the property TAG
1519 for that property. And in the case where the compiler synthesizes the ivar
1520 directly, the compiler is expected to generate a ``DW_TAG_member`` for that
1521 ivar (with the ``DW_AT_artificial`` set to 1), whose name will be the name used
1522 to access this ivar directly in code, and with the property attribute pointing
1523 back to the property it is backing.
1525 The following examples will serve as illustration for our discussion:
1527 .. code-block:: objc
1539 @synthesize p2 = n2;
1542 This produces the following DWARF (this is a "pseudo dwarfdump" output):
1544 .. code-block:: none
1546 0x00000100: TAG_structure_type [7] *
1547 AT_APPLE_runtime_class( 0x10 )
1549 AT_decl_file( "Objc_Property.m" )
1552 0x00000110 TAG_APPLE_property
1554 AT_type ( {0x00000150} ( int ) )
1556 0x00000120: TAG_APPLE_property
1558 AT_type ( {0x00000150} ( int ) )
1560 0x00000130: TAG_member [8]
1562 AT_APPLE_property ( {0x00000110} "p1" )
1563 AT_type( {0x00000150} ( int ) )
1564 AT_artificial ( 0x1 )
1566 0x00000140: TAG_member [8]
1568 AT_APPLE_property ( {0x00000120} "p2" )
1569 AT_type( {0x00000150} ( int ) )
1571 0x00000150: AT_type( ( int ) )
1573 Note, the current convention is that the name of the ivar for an
1574 auto-synthesized property is the name of the property from which it derives
1575 with an underscore prepended, as is shown in the example. But we actually
1576 don't need to know this convention, since we are given the name of the ivar
1579 Also, it is common practice in ObjC to have different property declarations in
1580 the @interface and @implementation - e.g. to provide a read-only property in
1581 the interface,and a read-write interface in the implementation. In that case,
1582 the compiler should emit whichever property declaration will be in force in the
1583 current translation unit.
1585 Developers can decorate a property with attributes which are encoded using
1586 ``DW_AT_APPLE_property_attribute``.
1588 .. code-block:: objc
1590 @property (readonly, nonatomic) int pr;
1592 .. code-block:: none
1594 TAG_APPLE_property [8]
1596 AT_type ( {0x00000147} (int) )
1597 AT_APPLE_property_attribute (DW_APPLE_PROPERTY_readonly, DW_APPLE_PROPERTY_nonatomic)
1599 The setter and getter method names are attached to the property using
1600 ``DW_AT_APPLE_property_setter`` and ``DW_AT_APPLE_property_getter`` attributes.
1602 .. code-block:: objc
1605 @property (setter=myOwnP3Setter:) int p3;
1606 -(void)myOwnP3Setter:(int)a;
1611 -(void)myOwnP3Setter:(int)a{ }
1614 The DWARF for this would be:
1616 .. code-block:: none
1618 0x000003bd: TAG_structure_type [7] *
1619 AT_APPLE_runtime_class( 0x10 )
1621 AT_decl_file( "Objc_Property.m" )
1624 0x000003cd TAG_APPLE_property
1626 AT_APPLE_property_setter ( "myOwnP3Setter:" )
1627 AT_type( {0x00000147} ( int ) )
1629 0x000003f3: TAG_member [8]
1631 AT_type ( {0x00000147} ( int ) )
1632 AT_APPLE_property ( {0x000003cd} )
1633 AT_artificial ( 0x1 )
1638 +-----------------------+--------+
1640 +=======================+========+
1641 | DW_TAG_APPLE_property | 0x4200 |
1642 +-----------------------+--------+
1644 New DWARF Attributes
1645 ^^^^^^^^^^^^^^^^^^^^
1647 +--------------------------------+--------+-----------+
1648 | Attribute | Value | Classes |
1649 +================================+========+===========+
1650 | DW_AT_APPLE_property | 0x3fed | Reference |
1651 +--------------------------------+--------+-----------+
1652 | DW_AT_APPLE_property_getter | 0x3fe9 | String |
1653 +--------------------------------+--------+-----------+
1654 | DW_AT_APPLE_property_setter | 0x3fea | String |
1655 +--------------------------------+--------+-----------+
1656 | DW_AT_APPLE_property_attribute | 0x3feb | Constant |
1657 +--------------------------------+--------+-----------+
1662 +--------------------------------+-------+
1664 +================================+=======+
1665 | DW_AT_APPLE_PROPERTY_readonly | 0x1 |
1666 +--------------------------------+-------+
1667 | DW_AT_APPLE_PROPERTY_readwrite | 0x2 |
1668 +--------------------------------+-------+
1669 | DW_AT_APPLE_PROPERTY_assign | 0x4 |
1670 +--------------------------------+-------+
1671 | DW_AT_APPLE_PROPERTY_retain | 0x8 |
1672 +--------------------------------+-------+
1673 | DW_AT_APPLE_PROPERTY_copy | 0x10 |
1674 +--------------------------------+-------+
1675 | DW_AT_APPLE_PROPERTY_nonatomic | 0x20 |
1676 +--------------------------------+-------+
1678 Name Accelerator Tables
1679 -----------------------
1684 The "``.debug_pubnames``" and "``.debug_pubtypes``" formats are not what a
1685 debugger needs. The "``pub``" in the section name indicates that the entries
1686 in the table are publicly visible names only. This means no static or hidden
1687 functions show up in the "``.debug_pubnames``". No static variables or private
1688 class variables are in the "``.debug_pubtypes``". Many compilers add different
1689 things to these tables, so we can't rely upon the contents between gcc, icc, or
1692 The typical query given by users tends not to match up with the contents of
1693 these tables. For example, the DWARF spec states that "In the case of the name
1694 of a function member or static data member of a C++ structure, class or union,
1695 the name presented in the "``.debug_pubnames``" section is not the simple name
1696 given by the ``DW_AT_name attribute`` of the referenced debugging information
1697 entry, but rather the fully qualified name of the data or function member."
1698 So the only names in these tables for complex C++ entries is a fully
1699 qualified name. Debugger users tend not to enter their search strings as
1700 "``a::b::c(int,const Foo&) const``", but rather as "``c``", "``b::c``" , or
1701 "``a::b::c``". So the name entered in the name table must be demangled in
1702 order to chop it up appropriately and additional names must be manually entered
1703 into the table to make it effective as a name lookup table for debuggers to
1706 All debuggers currently ignore the "``.debug_pubnames``" table as a result of
1707 its inconsistent and useless public-only name content making it a waste of
1708 space in the object file. These tables, when they are written to disk, are not
1709 sorted in any way, leaving every debugger to do its own parsing and sorting.
1710 These tables also include an inlined copy of the string values in the table
1711 itself making the tables much larger than they need to be on disk, especially
1712 for large C++ programs.
1714 Can't we just fix the sections by adding all of the names we need to this
1715 table? No, because that is not what the tables are defined to contain and we
1716 won't know the difference between the old bad tables and the new good tables.
1717 At best we could make our own renamed sections that contain all of the data we
1720 These tables are also insufficient for what a debugger like LLDB needs. LLDB
1721 uses clang for its expression parsing where LLDB acts as a PCH. LLDB is then
1722 often asked to look for type "``foo``" or namespace "``bar``", or list items in
1723 namespace "``baz``". Namespaces are not included in the pubnames or pubtypes
1724 tables. Since clang asks a lot of questions when it is parsing an expression,
1725 we need to be very fast when looking up names, as it happens a lot. Having new
1726 accelerator tables that are optimized for very quick lookups will benefit this
1727 type of debugging experience greatly.
1729 We would like to generate name lookup tables that can be mapped into memory
1730 from disk, and used as is, with little or no up-front parsing. We would also
1731 be able to control the exact content of these different tables so they contain
1732 exactly what we need. The Name Accelerator Tables were designed to fix these
1733 issues. In order to solve these issues we need to:
1735 * Have a format that can be mapped into memory from disk and used as is
1736 * Lookups should be very fast
1737 * Extensible table format so these tables can be made by many producers
1738 * Contain all of the names needed for typical lookups out of the box
1739 * Strict rules for the contents of tables
1741 Table size is important and the accelerator table format should allow the reuse
1742 of strings from common string tables so the strings for the names are not
1743 duplicated. We also want to make sure the table is ready to be used as-is by
1744 simply mapping the table into memory with minimal header parsing.
1746 The name lookups need to be fast and optimized for the kinds of lookups that
1747 debuggers tend to do. Optimally we would like to touch as few parts of the
1748 mapped table as possible when doing a name lookup and be able to quickly find
1749 the name entry we are looking for, or discover there are no matches. In the
1750 case of debuggers we optimized for lookups that fail most of the time.
1752 Each table that is defined should have strict rules on exactly what is in the
1753 accelerator tables and documented so clients can rely on the content.
1758 Standard Hash Tables
1759 """"""""""""""""""""
1761 Typical hash tables have a header, buckets, and each bucket points to the
1764 .. code-block:: none
1774 The BUCKETS are an array of offsets to DATA for each hash:
1776 .. code-block:: none
1779 | 0x00001000 | BUCKETS[0]
1780 | 0x00002000 | BUCKETS[1]
1781 | 0x00002200 | BUCKETS[2]
1782 | 0x000034f0 | BUCKETS[3]
1784 | 0xXXXXXXXX | BUCKETS[n_buckets]
1787 So for ``bucket[3]`` in the example above, we have an offset into the table
1788 0x000034f0 which points to a chain of entries for the bucket. Each bucket must
1789 contain a next pointer, full 32 bit hash value, the string itself, and the data
1790 for the current string value.
1792 .. code-block:: none
1795 0x000034f0: | 0x00003500 | next pointer
1796 | 0x12345678 | 32 bit hash
1797 | "erase" | string value
1798 | data[n] | HashData for this bucket
1800 0x00003500: | 0x00003550 | next pointer
1801 | 0x29273623 | 32 bit hash
1802 | "dump" | string value
1803 | data[n] | HashData for this bucket
1805 0x00003550: | 0x00000000 | next pointer
1806 | 0x82638293 | 32 bit hash
1807 | "main" | string value
1808 | data[n] | HashData for this bucket
1811 The problem with this layout for debuggers is that we need to optimize for the
1812 negative lookup case where the symbol we're searching for is not present. So
1813 if we were to lookup "``printf``" in the table above, we would make a 32 hash
1814 for "``printf``", it might match ``bucket[3]``. We would need to go to the
1815 offset 0x000034f0 and start looking to see if our 32 bit hash matches. To do
1816 so, we need to read the next pointer, then read the hash, compare it, and skip
1817 to the next bucket. Each time we are skipping many bytes in memory and
1818 touching new cache pages just to do the compare on the full 32 bit hash. All
1819 of these accesses then tell us that we didn't have a match.
1824 To solve the issues mentioned above we have structured the hash tables a bit
1825 differently: a header, buckets, an array of all unique 32 bit hash values,
1826 followed by an array of hash value data offsets, one for each hash value, then
1827 the data for all hash values:
1829 .. code-block:: none
1843 The ``BUCKETS`` in the name tables are an index into the ``HASHES`` array. By
1844 making all of the full 32 bit hash values contiguous in memory, we allow
1845 ourselves to efficiently check for a match while touching as little memory as
1846 possible. Most often checking the 32 bit hash values is as far as the lookup
1847 goes. If it does match, it usually is a match with no collisions. So for a
1848 table with "``n_buckets``" buckets, and "``n_hashes``" unique 32 bit hash
1849 values, we can clarify the contents of the ``BUCKETS``, ``HASHES`` and
1852 .. code-block:: none
1854 .-------------------------.
1855 | HEADER.magic | uint32_t
1856 | HEADER.version | uint16_t
1857 | HEADER.hash_function | uint16_t
1858 | HEADER.bucket_count | uint32_t
1859 | HEADER.hashes_count | uint32_t
1860 | HEADER.header_data_len | uint32_t
1861 | HEADER_DATA | HeaderData
1862 |-------------------------|
1863 | BUCKETS | uint32_t[n_buckets] // 32 bit hash indexes
1864 |-------------------------|
1865 | HASHES | uint32_t[n_hashes] // 32 bit hash values
1866 |-------------------------|
1867 | OFFSETS | uint32_t[n_hashes] // 32 bit offsets to hash value data
1868 |-------------------------|
1870 `-------------------------'
1872 So taking the exact same data from the standard hash example above we end up
1875 .. code-block:: none
1885 | ... | BUCKETS[n_buckets]
1887 | 0x........ | HASHES[0]
1888 | 0x........ | HASHES[1]
1889 | 0x........ | HASHES[2]
1890 | 0x........ | HASHES[3]
1891 | 0x........ | HASHES[4]
1892 | 0x........ | HASHES[5]
1893 | 0x12345678 | HASHES[6] hash for BUCKETS[3]
1894 | 0x29273623 | HASHES[7] hash for BUCKETS[3]
1895 | 0x82638293 | HASHES[8] hash for BUCKETS[3]
1896 | 0x........ | HASHES[9]
1897 | 0x........ | HASHES[10]
1898 | 0x........ | HASHES[11]
1899 | 0x........ | HASHES[12]
1900 | 0x........ | HASHES[13]
1901 | 0x........ | HASHES[n_hashes]
1903 | 0x........ | OFFSETS[0]
1904 | 0x........ | OFFSETS[1]
1905 | 0x........ | OFFSETS[2]
1906 | 0x........ | OFFSETS[3]
1907 | 0x........ | OFFSETS[4]
1908 | 0x........ | OFFSETS[5]
1909 | 0x000034f0 | OFFSETS[6] offset for BUCKETS[3]
1910 | 0x00003500 | OFFSETS[7] offset for BUCKETS[3]
1911 | 0x00003550 | OFFSETS[8] offset for BUCKETS[3]
1912 | 0x........ | OFFSETS[9]
1913 | 0x........ | OFFSETS[10]
1914 | 0x........ | OFFSETS[11]
1915 | 0x........ | OFFSETS[12]
1916 | 0x........ | OFFSETS[13]
1917 | 0x........ | OFFSETS[n_hashes]
1925 0x000034f0: | 0x00001203 | .debug_str ("erase")
1926 | 0x00000004 | A 32 bit array count - number of HashData with name "erase"
1927 | 0x........ | HashData[0]
1928 | 0x........ | HashData[1]
1929 | 0x........ | HashData[2]
1930 | 0x........ | HashData[3]
1931 | 0x00000000 | String offset into .debug_str (terminate data for hash)
1933 0x00003500: | 0x00001203 | String offset into .debug_str ("collision")
1934 | 0x00000002 | A 32 bit array count - number of HashData with name "collision"
1935 | 0x........ | HashData[0]
1936 | 0x........ | HashData[1]
1937 | 0x00001203 | String offset into .debug_str ("dump")
1938 | 0x00000003 | A 32 bit array count - number of HashData with name "dump"
1939 | 0x........ | HashData[0]
1940 | 0x........ | HashData[1]
1941 | 0x........ | HashData[2]
1942 | 0x00000000 | String offset into .debug_str (terminate data for hash)
1944 0x00003550: | 0x00001203 | String offset into .debug_str ("main")
1945 | 0x00000009 | A 32 bit array count - number of HashData with name "main"
1946 | 0x........ | HashData[0]
1947 | 0x........ | HashData[1]
1948 | 0x........ | HashData[2]
1949 | 0x........ | HashData[3]
1950 | 0x........ | HashData[4]
1951 | 0x........ | HashData[5]
1952 | 0x........ | HashData[6]
1953 | 0x........ | HashData[7]
1954 | 0x........ | HashData[8]
1955 | 0x00000000 | String offset into .debug_str (terminate data for hash)
1958 So we still have all of the same data, we just organize it more efficiently for
1959 debugger lookup. If we repeat the same "``printf``" lookup from above, we
1960 would hash "``printf``" and find it matches ``BUCKETS[3]`` by taking the 32 bit
1961 hash value and modulo it by ``n_buckets``. ``BUCKETS[3]`` contains "6" which
1962 is the index into the ``HASHES`` table. We would then compare any consecutive
1963 32 bit hashes values in the ``HASHES`` array as long as the hashes would be in
1964 ``BUCKETS[3]``. We do this by verifying that each subsequent hash value modulo
1965 ``n_buckets`` is still 3. In the case of a failed lookup we would access the
1966 memory for ``BUCKETS[3]``, and then compare a few consecutive 32 bit hashes
1967 before we know that we have no match. We don't end up marching through
1968 multiple words of memory and we really keep the number of processor data cache
1969 lines being accessed as small as possible.
1971 The string hash that is used for these lookup tables is the Daniel J.
1972 Bernstein hash which is also used in the ELF ``GNU_HASH`` sections. It is a
1973 very good hash for all kinds of names in programs with very few hash
1976 Empty buckets are designated by using an invalid hash index of ``UINT32_MAX``.
1981 These name hash tables are designed to be generic where specializations of the
1982 table get to define additional data that goes into the header ("``HeaderData``"),
1983 how the string value is stored ("``KeyType``") and the content of the data for each
1989 The header has a fixed part, and the specialized part. The exact format of the
1996 uint32_t magic; // 'HASH' magic value to allow endian detection
1997 uint16_t version; // Version number
1998 uint16_t hash_function; // The hash function enumeration that was used
1999 uint32_t bucket_count; // The number of buckets in this hash table
2000 uint32_t hashes_count; // The total number of unique hash values and hash data offsets in this table
2001 uint32_t header_data_len; // The bytes to skip to get to the hash indexes (buckets) for correct alignment
2002 // Specifically the length of the following HeaderData field - this does not
2003 // include the size of the preceding fields
2004 HeaderData header_data; // Implementation specific header data
2007 The header starts with a 32 bit "``magic``" value which must be ``'HASH'``
2008 encoded as an ASCII integer. This allows the detection of the start of the
2009 hash table and also allows the table's byte order to be determined so the table
2010 can be correctly extracted. The "``magic``" value is followed by a 16 bit
2011 ``version`` number which allows the table to be revised and modified in the
2012 future. The current version number is 1. ``hash_function`` is a ``uint16_t``
2013 enumeration that specifies which hash function was used to produce this table.
2014 The current values for the hash function enumerations include:
2018 enum HashFunctionType
2020 eHashFunctionDJB = 0u, // Daniel J Bernstein hash function
2023 ``bucket_count`` is a 32 bit unsigned integer that represents how many buckets
2024 are in the ``BUCKETS`` array. ``hashes_count`` is the number of unique 32 bit
2025 hash values that are in the ``HASHES`` array, and is the same number of offsets
2026 are contained in the ``OFFSETS`` array. ``header_data_len`` specifies the size
2027 in bytes of the ``HeaderData`` that is filled in by specialized versions of
2033 The header is followed by the buckets, hashes, offsets, and hash value data.
2039 uint32_t buckets[Header.bucket_count]; // An array of hash indexes into the "hashes[]" array below
2040 uint32_t hashes [Header.hashes_count]; // Every unique 32 bit hash for the entire table is in this table
2041 uint32_t offsets[Header.hashes_count]; // An offset that corresponds to each item in the "hashes[]" array above
2044 ``buckets`` is an array of 32 bit indexes into the ``hashes`` array. The
2045 ``hashes`` array contains all of the 32 bit hash values for all names in the
2046 hash table. Each hash in the ``hashes`` table has an offset in the ``offsets``
2047 array that points to the data for the hash value.
2049 This table setup makes it very easy to repurpose these tables to contain
2050 different data, while keeping the lookup mechanism the same for all tables.
2051 This layout also makes it possible to save the table to disk and map it in
2052 later and do very efficient name lookups with little or no parsing.
2054 DWARF lookup tables can be implemented in a variety of ways and can store a lot
2055 of information for each name. We want to make the DWARF tables extensible and
2056 able to store the data efficiently so we have used some of the DWARF features
2057 that enable efficient data storage to define exactly what kind of data we store
2060 The ``HeaderData`` contains a definition of the contents of each HashData chunk.
2061 We might want to store an offset to all of the debug information entries (DIEs)
2062 for each name. To keep things extensible, we create a list of items, or
2063 Atoms, that are contained in the data for each name. First comes the type of
2064 the data in each atom:
2071 eAtomTypeDIEOffset = 1u, // DIE offset, check form for encoding
2072 eAtomTypeCUOffset = 2u, // DIE offset of the compiler unit header that contains the item in question
2073 eAtomTypeTag = 3u, // DW_TAG_xxx value, should be encoded as DW_FORM_data1 (if no tags exceed 255) or DW_FORM_data2
2074 eAtomTypeNameFlags = 4u, // Flags from enum NameFlags
2075 eAtomTypeTypeFlags = 5u, // Flags from enum TypeFlags
2078 The enumeration values and their meanings are:
2080 .. code-block:: none
2082 eAtomTypeNULL - a termination atom that specifies the end of the atom list
2083 eAtomTypeDIEOffset - an offset into the .debug_info section for the DWARF DIE for this name
2084 eAtomTypeCUOffset - an offset into the .debug_info section for the CU that contains the DIE
2085 eAtomTypeDIETag - The DW_TAG_XXX enumeration value so you don't have to parse the DWARF to see what it is
2086 eAtomTypeNameFlags - Flags for functions and global variables (isFunction, isInlined, isExternal...)
2087 eAtomTypeTypeFlags - Flags for types (isCXXClass, isObjCClass, ...)
2089 Then we allow each atom type to define the atom type and how the data for each
2090 atom type data is encoded:
2096 uint16_t type; // AtomType enum value
2097 uint16_t form; // DWARF DW_FORM_XXX defines
2100 The ``form`` type above is from the DWARF specification and defines the exact
2101 encoding of the data for the Atom type. See the DWARF specification for the
2102 ``DW_FORM_`` definitions.
2108 uint32_t die_offset_base;
2109 uint32_t atom_count;
2110 Atoms atoms[atom_count0];
2113 ``HeaderData`` defines the base DIE offset that should be added to any atoms
2114 that are encoded using the ``DW_FORM_ref1``, ``DW_FORM_ref2``,
2115 ``DW_FORM_ref4``, ``DW_FORM_ref8`` or ``DW_FORM_ref_udata``. It also defines
2116 what is contained in each ``HashData`` object -- ``Atom.form`` tells us how large
2117 each field will be in the ``HashData`` and the ``Atom.type`` tells us how this data
2118 should be interpreted.
2120 For the current implementations of the "``.apple_names``" (all functions +
2121 globals), the "``.apple_types``" (names of all types that are defined), and
2122 the "``.apple_namespaces``" (all namespaces), we currently set the ``Atom``
2127 HeaderData.atom_count = 1;
2128 HeaderData.atoms[0].type = eAtomTypeDIEOffset;
2129 HeaderData.atoms[0].form = DW_FORM_data4;
2131 This defines the contents to be the DIE offset (eAtomTypeDIEOffset) that is
2132 encoded as a 32 bit value (DW_FORM_data4). This allows a single name to have
2133 multiple matching DIEs in a single file, which could come up with an inlined
2134 function for instance. Future tables could include more information about the
2135 DIE such as flags indicating if the DIE is a function, method, block,
2138 The KeyType for the DWARF table is a 32 bit string table offset into the
2139 ".debug_str" table. The ".debug_str" is the string table for the DWARF which
2140 may already contain copies of all of the strings. This helps make sure, with
2141 help from the compiler, that we reuse the strings between all of the DWARF
2142 sections and keeps the hash table size down. Another benefit to having the
2143 compiler generate all strings as DW_FORM_strp in the debug info, is that
2144 DWARF parsing can be made much faster.
2146 After a lookup is made, we get an offset into the hash data. The hash data
2147 needs to be able to deal with 32 bit hash collisions, so the chunk of data
2148 at the offset in the hash data consists of a triple:
2153 uint32_t hash_data_count
2154 HashData[hash_data_count]
2156 If "str_offset" is zero, then the bucket contents are done. 99.9% of the
2157 hash data chunks contain a single item (no 32 bit hash collision):
2159 .. code-block:: none
2162 | 0x00001023 | uint32_t KeyType (.debug_str[0x0001023] => "main")
2163 | 0x00000004 | uint32_t HashData count
2164 | 0x........ | uint32_t HashData[0] DIE offset
2165 | 0x........ | uint32_t HashData[1] DIE offset
2166 | 0x........ | uint32_t HashData[2] DIE offset
2167 | 0x........ | uint32_t HashData[3] DIE offset
2168 | 0x00000000 | uint32_t KeyType (end of hash chain)
2171 If there are collisions, you will have multiple valid string offsets:
2173 .. code-block:: none
2176 | 0x00001023 | uint32_t KeyType (.debug_str[0x0001023] => "main")
2177 | 0x00000004 | uint32_t HashData count
2178 | 0x........ | uint32_t HashData[0] DIE offset
2179 | 0x........ | uint32_t HashData[1] DIE offset
2180 | 0x........ | uint32_t HashData[2] DIE offset
2181 | 0x........ | uint32_t HashData[3] DIE offset
2182 | 0x00002023 | uint32_t KeyType (.debug_str[0x0002023] => "print")
2183 | 0x00000002 | uint32_t HashData count
2184 | 0x........ | uint32_t HashData[0] DIE offset
2185 | 0x........ | uint32_t HashData[1] DIE offset
2186 | 0x00000000 | uint32_t KeyType (end of hash chain)
2189 Current testing with real world C++ binaries has shown that there is around 1
2190 32 bit hash collision per 100,000 name entries.
2195 As we said, we want to strictly define exactly what is included in the
2196 different tables. For DWARF, we have 3 tables: "``.apple_names``",
2197 "``.apple_types``", and "``.apple_namespaces``".
2199 "``.apple_names``" sections should contain an entry for each DWARF DIE whose
2200 ``DW_TAG`` is a ``DW_TAG_label``, ``DW_TAG_inlined_subroutine``, or
2201 ``DW_TAG_subprogram`` that has address attributes: ``DW_AT_low_pc``,
2202 ``DW_AT_high_pc``, ``DW_AT_ranges`` or ``DW_AT_entry_pc``. It also contains
2203 ``DW_TAG_variable`` DIEs that have a ``DW_OP_addr`` in the location (global and
2204 static variables). All global and static variables should be included,
2205 including those scoped within functions and classes. For example using the
2217 Both of the static ``var`` variables would be included in the table. All
2218 functions should emit both their full names and their basenames. For C or C++,
2219 the full name is the mangled name (if available) which is usually in the
2220 ``DW_AT_MIPS_linkage_name`` attribute, and the ``DW_AT_name`` contains the
2221 function basename. If global or static variables have a mangled name in a
2222 ``DW_AT_MIPS_linkage_name`` attribute, this should be emitted along with the
2223 simple name found in the ``DW_AT_name`` attribute.
2225 "``.apple_types``" sections should contain an entry for each DWARF DIE whose
2230 * DW_TAG_enumeration_type
2231 * DW_TAG_pointer_type
2232 * DW_TAG_reference_type
2233 * DW_TAG_string_type
2234 * DW_TAG_structure_type
2235 * DW_TAG_subroutine_type
2238 * DW_TAG_ptr_to_member_type
2240 * DW_TAG_subrange_type
2246 * DW_TAG_packed_type
2247 * DW_TAG_volatile_type
2248 * DW_TAG_restrict_type
2249 * DW_TAG_interface_type
2250 * DW_TAG_unspecified_type
2251 * DW_TAG_shared_type
2253 Only entries with a ``DW_AT_name`` attribute are included, and the entry must
2254 not be a forward declaration (``DW_AT_declaration`` attribute with a non-zero
2255 value). For example, using the following code:
2265 We get a few type DIEs:
2267 .. code-block:: none
2269 0x00000067: TAG_base_type [5]
2270 AT_encoding( DW_ATE_signed )
2272 AT_byte_size( 0x04 )
2274 0x0000006e: TAG_pointer_type [6]
2275 AT_type( {0x00000067} ( int ) )
2276 AT_byte_size( 0x08 )
2278 The DW_TAG_pointer_type is not included because it does not have a ``DW_AT_name``.
2280 "``.apple_namespaces``" section should contain all ``DW_TAG_namespace`` DIEs.
2281 If we run into a namespace that has no name this is an anonymous namespace, and
2282 the name should be output as "``(anonymous namespace)``" (without the quotes).
2283 Why? This matches the output of the ``abi::cxa_demangle()`` that is in the
2284 standard C++ library that demangles mangled names.
2287 Language Extensions and File Format Changes
2288 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2290 Objective-C Extensions
2291 """"""""""""""""""""""
2293 "``.apple_objc``" section should contain all ``DW_TAG_subprogram`` DIEs for an
2294 Objective-C class. The name used in the hash table is the name of the
2295 Objective-C class itself. If the Objective-C class has a category, then an
2296 entry is made for both the class name without the category, and for the class
2297 name with the category. So if we have a DIE at offset 0x1234 with a name of
2298 method "``-[NSString(my_additions) stringWithSpecialString:]``", we would add
2299 an entry for "``NSString``" that points to DIE 0x1234, and an entry for
2300 "``NSString(my_additions)``" that points to 0x1234. This allows us to quickly
2301 track down all Objective-C methods for an Objective-C class when doing
2302 expressions. It is needed because of the dynamic nature of Objective-C where
2303 anyone can add methods to a class. The DWARF for Objective-C methods is also
2304 emitted differently from C++ classes where the methods are not usually
2305 contained in the class definition, they are scattered about across one or more
2306 compile units. Categories can also be defined in different shared libraries.
2307 So we need to be able to quickly find all of the methods and class functions
2308 given the Objective-C class name, or quickly find all methods and class
2309 functions for a class + category name. This table does not contain any
2310 selector names, it just maps Objective-C class names (or class names +
2311 category) to all of the methods and class functions. The selectors are added
2312 as function basenames in the "``.debug_names``" section.
2314 In the "``.apple_names``" section for Objective-C functions, the full name is
2315 the entire function name with the brackets ("``-[NSString
2316 stringWithCString:]``") and the basename is the selector only
2317 ("``stringWithCString:``").
2322 The sections names for the apple hash tables are for non-mach-o files. For
2323 mach-o files, the sections should be contained in the ``__DWARF`` segment with
2326 * "``.apple_names``" -> "``__apple_names``"
2327 * "``.apple_types``" -> "``__apple_types``"
2328 * "``.apple_namespaces``" -> "``__apple_namespac``" (16 character limit)
2329 * "``.apple_objc``" -> "``__apple_objc``"