Source Level Debugging with LLVM
A leafy and green bug eater

Written by Chris Lattner and Jim Laskey

Introduction

This document is the central repository for all information pertaining to debug information in LLVM. It describes the actual format that the LLVM debug information takes, which is useful for those interested in creating front-ends or dealing directly with the information. Further, this document provides specifc examples of what debug information for C/C++.

Philosophy behind LLVM debugging information

The idea of the LLVM debugging information is to capture how the important pieces of the source-language's Abstract Syntax Tree map onto LLVM code. Several design aspects have shaped the solution that appears here. The important ones are:

The approach used by the LLVM implementation is to use a small set of intrinsic functions to define a mapping between LLVM program objects and the source-level objects. The description of the source-level program is maintained in LLVM global variables in an implementation-defined format (the C/C++ front-end currently uses working draft 7 of the Dwarf 3 standard).

When a program is being debugged, a debugger interacts with the user and turns the stored debug information into source-language specific information. As such, a debugger must be aware of the source-language, and is thus tied to a specific language or family of languages.

Debug information consumers

The role of debug information is to provide meta information normally stripped away during the compilation process. This meta information provides an LLVM user a relationship between generated code and the original program source code.

Currently, debug information is consumed by the DwarfWriter to produce dwarf information used by the gdb debugger. Other targets could use the same information to produce stabs or other debug forms.

It would also be reasonable to use debug information to feed profiling tools for analysis of generated code, or, tools for reconstructing the original source from generated code.

TODO - expound a bit more.

Debugging optimized code

An extremely high priority of LLVM debugging information is to make it interact well with optimizations and analysis. In particular, the LLVM debug information provides the following guarantees:

Basically, the debug information allows you to compile a program with "-O0 -g" and get full debug information, allowing you to arbitrarily modify the program as it executes from a debugger. Compiling a program with "-O3 -g" gives you full debug information that is always available and accurate for reading (e.g., you get accurate stack traces despite tail call elimination and inlining), but you might lose the ability to modify the program and call functions where were optimized out of the program, or inlined away completely.

LLVM test suite provides a framework to test optimizer's handling of debugging information. It can be run like this:

% cd llvm/projects/test-suite/MultiSource/Benchmarks  # or some other level
% make TEST=dbgopt

This will test impact of debugging information on optimization passes. If debugging information influences optimization passes then it will be reported as a failure. See TestingGuide for more information on LLVM test infratsture and how to run various tests.

Debugging information format

LLVM debugging information has been carefully designed to make it possible for the optimizer to optimize the program and debugging information without necessarily having to know anything about debugging information. In particular, the global constant merging pass automatically eliminates duplicated debugging information (often caused by header files), the global dead code elimination pass automatically deletes debugging information for a function if it decides to delete the function, and the linker eliminates debug information when it merges linkonce functions.

To do this, most of the debugging information (descriptors for types, variables, functions, source files, etc) is inserted by the language front-end in the form of LLVM global variables. These LLVM global variables are no different from any other global variables, except that they have a web of LLVM intrinsic functions that point to them. If the last references to a particular piece of debugging information are deleted (for example, by the -globaldce pass), the extraneous debug information will automatically become dead and be removed by the optimizer.

Debug information is designed to be agnostic about the target debugger and debugging information representation (e.g. DWARF/Stabs/etc). It uses a generic machine debug information pass to decode the information that represents variables, types, functions, namespaces, etc: this allows for arbitrary source-language semantics and type-systems to be used, as long as there is a module written for the target debugger to interpret the information. In addition, debug global variables are declared in the "llvm.metadata" section. All values declared in this section are stripped away after target debug information is constructed and before the program object is emitted.

To provide basic functionality, the LLVM debugger does have to make some assumptions about the source-level language being debugged, though it keeps these to a minimum. The only common features that the LLVM debugger assumes exist are source files, and program objects. These abstract objects are used by a debugger to form stack traces, show information about local variables, etc.

This section of the documentation first describes the representation aspects common to any source-language. The next section describes the data layout conventions used by the C and C++ front-ends.

Debug information descriptors

In consideration of the complexity and volume of debug information, LLVM provides a specification for well formed debug global variables. The constant value of each of these globals is one of a limited set of structures, known as debug descriptors.

Consumers of LLVM debug information expect the descriptors for program objects to start in a canonical format, but the descriptors can include additional information appended at the end that is source-language specific. All LLVM debugging information is versioned, allowing backwards compatibility in the case that the core structures need to change in some way. Also, all debugging information objects start with a tag to indicate what type of object it is. The source-language is allowed to define its own objects, by using unreserved tag numbers. We recommend using with tags in the range 0x1000 thru 0x2000 (there is a defined enum DW_TAG_user_base = 0x1000.)

The fields of debug descriptors used internally by LLVM (MachineModuleInfo) are restricted to only the simple data types int, uint, bool, float, double, sbyte* and { }* . References to arbitrary values are handled using a { }* and a cast to { }* expression; typically references to other field descriptors, arrays of descriptors or global variables.

  %llvm.dbg.object.type = type {
    uint,   ;; A tag
    ...
  }

The first field of a descriptor is always an uint containing a tag value identifying the content of the descriptor. The remaining fields are specific to the descriptor. The values of tags are loosely bound to the tag values of Dwarf information entries. However, that does not restrict the use of the information supplied to Dwarf targets. To facilitate versioning of debug information, the tag is augmented with the current debug version (LLVMDebugVersion = 4 << 16 or 0x40000 or 262144.)

The details of the various descriptors follow.

Anchor descriptors
  %llvm.dbg.anchor.type = type {
    uint,   ;; Tag = 0 + LLVMDebugVersion
    uint    ;; Tag of descriptors grouped by the anchor
  }

One important aspect of the LLVM debug representation is that it allows the LLVM debugger to efficiently index all of the global objects without having the scan the program. To do this, all of the global objects use "anchor" descriptors with designated names. All of the global objects of a particular type (e.g., compile units) contain a pointer to the anchor. This pointer allows a debugger to use def-use chains to find all global objects of that type.

The following names are recognized as anchors by LLVM:

  %llvm.dbg.compile_units       = linkonce constant %llvm.dbg.anchor.type  { uint 0, uint 17 } ;; DW_TAG_compile_unit
  %llvm.dbg.global_variables    = linkonce constant %llvm.dbg.anchor.type  { uint 0, uint 52 } ;; DW_TAG_variable
  %llvm.dbg.subprograms         = linkonce constant %llvm.dbg.anchor.type  { uint 0, uint 46 } ;; DW_TAG_subprogram

Using anchors in this way (where the compile unit descriptor points to the anchors, as opposed to having a list of compile unit descriptors) allows for the standard dead global elimination and merging passes to automatically remove unused debugging information. If the globals were kept track of through lists, there would always be an object pointing to the descriptors, thus would never be deleted.

Compile unit descriptors
  %llvm.dbg.compile_unit.type = type {
    uint,   ;; Tag = 17 + LLVMDebugVersion (DW_TAG_compile_unit)
    {  }*,  ;; Compile unit anchor = cast = (%llvm.dbg.anchor.type* %llvm.dbg.compile_units to {  }*)
    uint,   ;; Dwarf language identifier (ex. DW_LANG_C89) 
    sbyte*, ;; Source file name
    sbyte*, ;; Source file directory (includes trailing slash)
    sbyte*  ;; Producer (ex. "4.0.1 LLVM (LLVM research group)")
  }

These descriptors contain a source language ID for the file (we use the Dwarf 3.0 ID numbers, such as DW_LANG_C89, DW_LANG_C_plus_plus, DW_LANG_Cobol74, etc), three strings describing the filename, working directory of the compiler, and an identifier string for the compiler that produced it.

Compile unit descriptors provide the root context for objects declared in a specific source file. Global variables and top level functions would be defined using this context. Compile unit descriptors also provide context for source line correspondence.

Global variable descriptors
  %llvm.dbg.global_variable.type = type {
    uint,   ;; Tag = 52 + LLVMDebugVersion (DW_TAG_variable)
    {  }*,  ;; Global variable anchor = cast (%llvm.dbg.anchor.type* %llvm.dbg.global_variables to {  }*),  
    {  }*,  ;; Reference to context descriptor
    sbyte*, ;; Name
    sbyte*, ;; Display name (fully qualified C++ name)
    sbyte*, ;; MIPS linkage name (for C++)
    {  }*,  ;; Reference to compile unit where defined
    uint,   ;; Line number where defined
    {  }*,  ;; Reference to type descriptor
    bool,   ;; True if the global is local to compile unit (static)
    bool,   ;; True if the global is defined in the compile unit (not extern)
    {  }*   ;; Reference to the global variable
  }

These descriptors provide debug information about globals variables. The provide details such as name, type and where the variable is defined.

Subprogram descriptors
  %llvm.dbg.subprogram.type = type {
    uint,   ;; Tag = 46 + LLVMDebugVersion (DW_TAG_subprogram)
    {  }*,  ;; Subprogram anchor = cast (%llvm.dbg.anchor.type* %llvm.dbg.subprograms to {  }*),  
    {  }*,  ;; Reference to context descriptor
    sbyte*, ;; Name
    sbyte*, ;; Display name (fully qualified C++ name)
    sbyte*, ;; MIPS linkage name (for C++)
    {  }*,  ;; Reference to compile unit where defined
    uint,   ;; Line number where defined
    {  }*,  ;; Reference to type descriptor
    bool,   ;; True if the global is local to compile unit (static)
    bool    ;; True if the global is defined in the compile unit (not extern)
  }

These descriptors provide debug information about functions, methods and subprograms. They provide details such as name, return types and the source location where the subprogram is defined.

Block descriptors
  %llvm.dbg.block = type {
    uint,   ;; Tag = 13 + LLVMDebugVersion (DW_TAG_lexical_block)
    {  }*   ;; Reference to context descriptor
  }

These descriptors provide debug information about nested blocks within a subprogram. The array of member descriptors is used to define local variables and deeper nested blocks.

Basic type descriptors
  %llvm.dbg.basictype.type = type {
    uint,   ;; Tag = 36 + LLVMDebugVersion (DW_TAG_base_type)
    {  }*,  ;; Reference to context (typically a compile unit)
    sbyte*, ;; Name (may be "" for anonymous types)
    {  }*,  ;; Reference to compile unit where defined (may be NULL)
    uint,   ;; Line number where defined (may be 0)
    uint,   ;; Size in bits
    uint,   ;; Alignment in bits
    uint,   ;; Offset in bits
    uint    ;; Dwarf type encoding
  }

These descriptors define primitive types used in the code. Example int, bool and float. The context provides the scope of the type, which is usually the top level. Since basic types are not usually user defined the compile unit and line number can be left as NULL and 0. The size, alignment and offset are expressed in bits and can be 64 bit values. The alignment is used to round the offset when embedded in a composite type (example to keep float doubles on 64 bit boundaries.) The offset is the bit offset if embedded in a composite type.

The type encoding provides the details of the type. The values are typically one of the following;

  DW_ATE_address = 1
  DW_ATE_boolean = 2
  DW_ATE_float = 4
  DW_ATE_signed = 5
  DW_ATE_signed_char = 6
  DW_ATE_unsigned = 7
  DW_ATE_unsigned_char = 8
Derived type descriptors
  %llvm.dbg.derivedtype.type = type {
    uint,   ;; Tag (see below)
    {  }*,  ;; Reference to context
    sbyte*, ;; Name (may be "" for anonymous types)
    {  }*,  ;; Reference to compile unit where defined (may be NULL)
    uint,   ;; Line number where defined (may be 0)
    uint,   ;; Size in bits
    uint,   ;; Alignment in bits
    uint,   ;; Offset in bits
    {  }*   ;; Reference to type derived from
  }

These descriptors are used to define types derived from other types. The value of the tag varies depending on the meaning. The following are possible tag values;

  DW_TAG_formal_parameter = 5
  DW_TAG_member = 13
  DW_TAG_pointer_type = 15
  DW_TAG_reference_type = 16
  DW_TAG_typedef = 22
  DW_TAG_const_type = 38
  DW_TAG_volatile_type = 53
  DW_TAG_restrict_type = 55

DW_TAG_member is used to define a member of a composite type or subprogram. The type of the member is the derived type. DW_TAG_formal_parameter is used to define a member which is a formal argument of a subprogram.

DW_TAG_typedef is used to provide a name for the derived type.

DW_TAG_pointer_type, DW_TAG_reference_type, DW_TAG_const_type, DW_TAG_volatile_type and DW_TAG_restrict_type are used to qualify the derived type.

Derived type location can be determined from the compile unit and line number. The size, alignment and offset are expressed in bits and can be 64 bit values. The alignment is used to round the offset when embedded in a composite type (example to keep float doubles on 64 bit boundaries.) The offset is the bit offset if embedded in a composite type.

Note that the void * type is expressed as a llvm.dbg.derivedtype.type with tag of DW_TAG_pointer_type and NULL derived type.

Composite type descriptors
  %llvm.dbg.compositetype.type = type {
    uint,   ;; Tag (see below)
    {  }*,  ;; Reference to context
    sbyte*, ;; Name (may be "" for anonymous types)
    {  }*,  ;; Reference to compile unit where defined (may be NULL)
    uint,   ;; Line number where defined (may be 0)
    uint,   ;; Size in bits
    uint,   ;; Alignment in bits
    uint,   ;; Offset in bits
    {  }*   ;; Reference to array of member descriptors
  }

These descriptors are used to define types that are composed of 0 or more elements. The value of the tag varies depending on the meaning. The following are possible tag values;

  DW_TAG_array_type = 1
  DW_TAG_enumeration_type = 4
  DW_TAG_structure_type = 19
  DW_TAG_union_type = 23
  DW_TAG_vector_type = 259
  DW_TAG_subroutine_type = 46
  DW_TAG_inheritance = 26

The vector flag indicates that an array type is a native packed vector.

The members of array types (tag = DW_TAG_array_type) or vector types (tag = DW_TAG_vector_type) are subrange descriptors, each representing the range of subscripts at that level of indexing.

The members of enumeration types (tag = DW_TAG_enumeration_type) are enumerator descriptors, each representing the definition of enumeration value for the set.

The members of structure (tag = DW_TAG_structure_type) or union (tag = DW_TAG_union_type) types are any one of the basic, derived or composite type descriptors, each representing a field member of the structure or union.

For C++ classes (tag = DW_TAG_structure_type), member descriptors provide information about base classes, static members and member functions. If a member is a derived type descriptor and has a tag of DW_TAG_inheritance, then the type represents a base class. If the member of is a global variable descriptor then it represents a static member. And, if the member is a subprogram descriptor then it represents a member function. For static members and member functions, getName() returns the members link or the C++ mangled name. getDisplayName() the simplied version of the name.

The first member of subroutine (tag = DW_TAG_subroutine_type) type elements is the return type for the subroutine. The remaining elements are the formal arguments to the subroutine.

Composite type location can be determined from the compile unit and line number. The size, alignment and offset are expressed in bits and can be 64 bit values. The alignment is used to round the offset when embedded in a composite type (as an example, to keep float doubles on 64 bit boundaries.) The offset is the bit offset if embedded in a composite type.

Subrange descriptors
  %llvm.dbg.subrange.type = type {
    uint,   ;; Tag = 33 + LLVMDebugVersion (DW_TAG_subrange_type)
    uint,   ;; Low value
    uint    ;; High value
  }

These descriptors are used to define ranges of array subscripts for an array composite type. The low value defines the lower bounds typically zero for C/C++. The high value is the upper bounds. Values are 64 bit. High - low + 1 is the size of the array. If low == high the array will be unbounded.

Enumerator descriptors
  %llvm.dbg.enumerator.type = type {
    uint,   ;; Tag = 40 + LLVMDebugVersion (DW_TAG_enumerator)
    sbyte*, ;; Name
    uint    ;; Value
  }

These descriptors are used to define members of an enumeration composite type, it associates the name to the value.

Local variables
  %llvm.dbg.variable.type = type {
    uint,    ;; Tag (see below)
    {  }*,   ;; Context
    sbyte*,  ;; Name
    {  }*,   ;; Reference to compile unit where defined
    uint,    ;; Line number where defined
    {  }*    ;; Type descriptor
  }

These descriptors are used to define variables local to a sub program. The value of the tag depends on the usage of the variable;

  DW_TAG_auto_variable = 256
  DW_TAG_arg_variable = 257
  DW_TAG_return_variable = 258

An auto variable is any variable declared in the body of the function. An argument variable is any variable that appears as a formal argument to the function. A return variable is used to track the result of a function and has no source correspondent.

The context is either the subprogram or block where the variable is defined. Name the source variable name. Compile unit and line indicate where the variable was defined. Type descriptor defines the declared type of the variable.

Debugger intrinsic functions

LLVM uses several intrinsic functions (name prefixed with "llvm.dbg") to provide debug information at various points in generated code.

llvm.dbg.stoppoint
  void %llvm.dbg.stoppoint( uint, uint, { }* )

This intrinsic is used to provide correspondence between the source file and the generated code. The first argument is the line number (base 1), second argument is the column number (0 if unknown) and the third argument the source %llvm.dbg.compile_unit* cast to a { }*. Code following a call to this intrinsic will have been defined in close proximity of the line, column and file. This information holds until the next call to %lvm.dbg.stoppoint.

llvm.dbg.func.start
  void %llvm.dbg.func.start( { }* )

This intrinsic is used to link the debug information in %llvm.dbg.subprogram to the function. It defines the beginning of the function's declarative region (scope). It also implies a call to %llvm.dbg.stoppoint which defines a source line "stop point". The intrinsic should be called early in the function after the all the alloca instructions. It should be paired off with a closing %llvm.dbg.region.end. The function's single argument is the %llvm.dbg.subprogram.type.

llvm.dbg.region.start
  void %llvm.dbg.region.start( { }* )

This intrinsic is used to define the beginning of a declarative scope (ex. block) for local language elements. It should be paired off with a closing %llvm.dbg.region.end. The function's single argument is the %llvm.dbg.block which is starting.

llvm.dbg.region.end
  void %llvm.dbg.region.end( { }* )

This intrinsic is used to define the end of a declarative scope (ex. block) for local language elements. It should be paired off with an opening %llvm.dbg.region.start or %llvm.dbg.func.start. The function's single argument is either the %llvm.dbg.block or the %llvm.dbg.subprogram.type which is ending.

llvm.dbg.declare
  void %llvm.dbg.declare( { } *, { }* )

This intrinsic provides information about a local element (ex. variable.) The first argument is the alloca for the variable, cast to a { }*. The second argument is the %llvm.dbg.variable containing the description of the variable, also cast to a { }*.

Representing stopping points in the source program

LLVM debugger "stop points" are a key part of the debugging representation that allows the LLVM to maintain simple semantics for debugging optimized code. The basic idea is that the front-end inserts calls to the %llvm.dbg.stoppoint intrinsic function at every point in the program where a debugger should be able to inspect the program (these correspond to places a debugger stops when you "step" through it). The front-end can choose to place these as fine-grained as it would like (for example, before every subexpression evaluated), but it is recommended to only put them after every source statement that includes executable code.

Using calls to this intrinsic function to demark legal points for the debugger to inspect the program automatically disables any optimizations that could potentially confuse debugging information. To non-debug-information-aware transformations, these calls simply look like calls to an external function, which they must assume to do anything (including reading or writing to any part of reachable memory). On the other hand, it does not impact many optimizations, such as code motion of non-trapping instructions, nor does it impact optimization of subexpressions, code duplication transformations, or basic-block reordering transformations.

Object lifetimes and scoping

In many languages, the local variables in functions can have their lifetime or scope limited to a subset of a function. In the C family of languages, for example, variables are only live (readable and writable) within the source block that they are defined in. In functional languages, values are only readable after they have been defined. Though this is a very obvious concept, it is also non-trivial to model in LLVM, because it has no notion of scoping in this sense, and does not want to be tied to a language's scoping rules.

In order to handle this, the LLVM debug format uses the notion of "regions" of a function, delineated by calls to intrinsic functions. These intrinsic functions define new regions of the program and indicate when the region lifetime expires. Consider the following C fragment, for example:

1.  void foo() {
2.    int X = ...;
3.    int Y = ...;
4.    {
5.      int Z = ...;
6.      ...
7.    }
8.    ...
9.  }

Compiled to LLVM, this function would be represented like this:

void %foo() {
entry:
    %X = alloca int
    %Y = alloca int
    %Z = alloca int
    
    ...
    
    call void %llvm.dbg.func.start( %llvm.dbg.subprogram.type* %llvm.dbg.subprogram )
    
    call void %llvm.dbg.stoppoint( uint 2, uint 2, %llvm.dbg.compile_unit* %llvm.dbg.compile_unit )
    
    call void %llvm.dbg.declare({}* %X, ...)
    call void %llvm.dbg.declare({}* %Y, ...)
    
    ;; Evaluate expression on line 2, assigning to X.
    
    call void %llvm.dbg.stoppoint( uint 3, uint 2, %llvm.dbg.compile_unit* %llvm.dbg.compile_unit )
    
    ;; Evaluate expression on line 3, assigning to Y.
    
    call void %llvm.region.start()
    call void %llvm.dbg.stoppoint( uint 5, uint 4, %llvm.dbg.compile_unit* %llvm.dbg.compile_unit )
    call void %llvm.dbg.declare({}* %X, ...)
    
    ;; Evaluate expression on line 5, assigning to Z.
    
    call void %llvm.dbg.stoppoint( uint 7, uint 2, %llvm.dbg.compile_unit* %llvm.dbg.compile_unit )
    call void %llvm.region.end()
    
    call void %llvm.dbg.stoppoint( uint 9, uint 2, %llvm.dbg.compile_unit* %llvm.dbg.compile_unit )
    
    call void %llvm.region.end()
    
    ret void
}

This example illustrates a few important details about the LLVM debugging information. In particular, it shows how the various intrinsics are applied together to allow a debugger to analyze the relationship between statements, variable definitions, and the code used to implement the function.

The first intrinsic %llvm.dbg.func.start provides a link with the subprogram descriptor containing the details of this function. This call also defines the beginning of the function region, bounded by the %llvm.region.end at the end of the function. This region is used to bracket the lifetime of variables declared within. For a function, this outer region defines a new stack frame whose lifetime ends when the region is ended.

It is possible to define inner regions for short term variables by using the %llvm.region.start and %llvm.region.end to bound a region. The inner region in this example would be for the block containing the declaration of Z.

Using regions to represent the boundaries of source-level functions allow LLVM interprocedural optimizations to arbitrarily modify LLVM functions without having to worry about breaking mapping information between the LLVM code and the and source-level program. In particular, the inliner requires no modification to support inlining with debugging information: there is no explicit correlation drawn between LLVM functions and their source-level counterparts (note however, that if the inliner inlines all instances of a non-strong-linkage function into its caller that it will not be possible for the user to manually invoke the inlined function from a debugger).

Once the function has been defined, the stopping point corresponding to line #2 (column #2) of the function is encountered. At this point in the function, no local variables are live. As lines 2 and 3 of the example are executed, their variable definitions are introduced into the program using %llvm.dbg.declare, without the need to specify a new region. These variables do not require new regions to be introduced because they go out of scope at the same point in the program: line 9.

In contrast, the Z variable goes out of scope at a different time, on line 7. For this reason, it is defined within the inner region, which kills the availability of Z before the code for line 8 is executed. In this way, regions can support arbitrary source-language scoping rules, as long as they can only be nested (ie, one scope cannot partially overlap with a part of another scope).

It is worth noting that this scoping mechanism is used to control scoping of all declarations, not just variable declarations. For example, the scope of a C++ using declaration is controlled with this and could change how name lookup is performed.

C/C++ front-end specific debug information

The C and C++ front-ends represent information about the program in a format that is effectively identical to Dwarf 3.0 in terms of information content. This allows code generators to trivially support native debuggers by generating standard dwarf information, and contains enough information for non-dwarf targets to translate it as needed.

This section describes the forms used to represent C and C++ programs. Other languages could pattern themselves after this (which itself is tuned to representing programs in the same way that Dwarf 3 does), or they could choose to provide completely different forms if they don't fit into the Dwarf model. As support for debugging information gets added to the various LLVM source-language front-ends, the information used should be documented here.

The following sections provide examples of various C/C++ constructs and the debug information that would best describe those constructs.

C/C++ source file information

Given the source files "MySource.cpp" and "MyHeader.h" located in the directory "/Users/mine/sources", the following code;

#include "MyHeader.h"

int main(int argc, char *argv[]) {
  return 0;
}

a C/C++ front-end would generate the following descriptors;

...
;;
;; Define types used.  In this case we need one for compile unit anchors and one
;; for compile units.
;;
%llvm.dbg.anchor.type = type { uint, uint }
%llvm.dbg.compile_unit.type = type { uint, {  }*, uint, uint, sbyte*, sbyte*, sbyte* }
...
;;
;; Define the anchor for compile units.  Note that the second field of the
;; anchor is 17, which is the same as the tag for compile units
;; (17 = DW_TAG_compile_unit.)
;;
%llvm.dbg.compile_units = linkonce constant %llvm.dbg.anchor.type { uint 0, uint 17 }, section "llvm.metadata"

;;
;; Define the compile unit for the source file "/Users/mine/sources/MySource.cpp".
;;
%llvm.dbg.compile_unit1 = internal constant %llvm.dbg.compile_unit.type {
    uint add(uint 17, uint 262144), 
    {  }* cast (%llvm.dbg.anchor.type* %llvm.dbg.compile_units to {  }*), 
    uint 1, 
    uint 1, 
    sbyte* getelementptr ([13 x sbyte]* %str1, int 0, int 0), 
    sbyte* getelementptr ([21 x sbyte]* %str2, int 0, int 0), 
    sbyte* getelementptr ([33 x sbyte]* %str3, int 0, int 0) }, section "llvm.metadata"
    
;;
;; Define the compile unit for the header file "/Users/mine/sources/MyHeader.h".
;;
%llvm.dbg.compile_unit2 = internal constant %llvm.dbg.compile_unit.type {
    uint add(uint 17, uint 262144), 
    {  }* cast (%llvm.dbg.anchor.type* %llvm.dbg.compile_units to {  }*), 
    uint 1, 
    uint 1, 
    sbyte* getelementptr ([11 x sbyte]* %str4, int 0, int 0), 
    sbyte* getelementptr ([21 x sbyte]* %str2, int 0, int 0), 
    sbyte* getelementptr ([33 x sbyte]* %str3, int 0, int 0) }, section "llvm.metadata"

;;
;; Define each of the strings used in the compile units.
;;
%str1 = internal constant [13 x sbyte] c"MySource.cpp\00", section "llvm.metadata";
%str2 = internal constant [21 x sbyte] c"/Users/mine/sources/\00", section "llvm.metadata";
%str3 = internal constant [33 x sbyte] c"4.0.1 LLVM (LLVM research group)\00", section "llvm.metadata";
%str4 = internal constant [11 x sbyte] c"MyHeader.h\00", section "llvm.metadata";
...
C/C++ global variable information

Given an integer global variable declared as follows;

int MyGlobal = 100;

a C/C++ front-end would generate the following descriptors;

;;
;; Define types used. One for global variable anchors, one for the global
;; variable descriptor, one for the global's basic type and one for the global's
;; compile unit.
;;
%llvm.dbg.anchor.type = type { uint, uint }
%llvm.dbg.global_variable.type = type { uint, {  }*, {  }*, sbyte*, {  }*, uint, {  }*, bool, bool, {  }*, uint }
%llvm.dbg.basictype.type = type { uint, {  }*, sbyte*, {  }*, int, uint, uint, uint, uint }
%llvm.dbg.compile_unit.type = ...
...
;;
;; Define the global itself.
;;
%MyGlobal = global int 100
...
;;
;; Define the anchor for global variables.  Note that the second field of the
;; anchor is 52, which is the same as the tag for global variables
;; (52 = DW_TAG_variable.)
;;
%llvm.dbg.global_variables = linkonce constant %llvm.dbg.anchor.type { uint 0, uint 52 }, section "llvm.metadata"

;;
;; Define the global variable descriptor.  Note the reference to the global
;; variable anchor and the global variable itself.
;;
%llvm.dbg.global_variable = internal constant %llvm.dbg.global_variable.type {
    uint add(uint 52, uint 262144), 
    {  }* cast (%llvm.dbg.anchor.type* %llvm.dbg.global_variables to {  }*), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([9 x sbyte]* %str1, int 0, int 0), 
    sbyte* getelementptr ([1 x sbyte]* %str2, int 0, int 0), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    uint 1,
    {  }* cast (%llvm.dbg.basictype.type* %llvm.dbg.basictype to {  }*), 
    bool false, 
    bool true, 
    {  }* cast (int* %MyGlobal to {  }*) }, section "llvm.metadata"
    
;;
;; Define the basic type of 32 bit signed integer.  Note that since int is an
;; intrinsic type the source file is NULL and line 0.
;;    
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([4 x sbyte]* %str3, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 32, 
    uint 32, 
    uint 0, 
    uint 5 }, section "llvm.metadata"

;;
;; Define the names of the global variable and basic type.
;;
%str1 = internal constant [9 x sbyte] c"MyGlobal\00", section "llvm.metadata"
%str2 = internal constant [1 x sbyte] c"\00", section "llvm.metadata"
%str3 = internal constant [4 x sbyte] c"int\00", section "llvm.metadata"
C/C++ function information

Given a function declared as follows;

int main(int argc, char *argv[]) {
  return 0;
}

a C/C++ front-end would generate the following descriptors;

;;
;; Define types used. One for subprogram anchors, one for the subprogram
;; descriptor, one for the global's basic type and one for the subprogram's
;; compile unit.
;;
%llvm.dbg.subprogram.type = type { uint, {  }*, {  }*, sbyte*, {  }*, bool, bool }
%llvm.dbg.anchor.type = type { uint, uint }
%llvm.dbg.compile_unit.type = ...
	
;;
;; Define the anchor for subprograms.  Note that the second field of the
;; anchor is 46, which is the same as the tag for subprograms
;; (46 = DW_TAG_subprogram.)
;;
%llvm.dbg.subprograms = linkonce constant %llvm.dbg.anchor.type { uint 0, uint 46 }, section "llvm.metadata"

;;
;; Define the descriptor for the subprogram.  TODO - more details.
;;
%llvm.dbg.subprogram = internal constant %llvm.dbg.subprogram.type {
    uint add(uint 46, uint 262144), 
    {  }* cast (%llvm.dbg.anchor.type* %llvm.dbg.subprograms to {  }*), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([5 x sbyte]* %str1, int 0, int 0), 
    sbyte* getelementptr ([1 x sbyte]* %str2, int 0, int 0), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*),
    uint 1,
    {  }* null, 
    bool false, 
    bool true }, section "llvm.metadata"

;;
;; Define the name of the subprogram.
;;
%str1 = internal constant [5 x sbyte] c"main\00", section "llvm.metadata"
%str2 = internal constant [1 x sbyte] c"\00", section "llvm.metadata"

;;
;; Define the subprogram itself.
;;
int %main(int %argc, sbyte** %argv) {
...
}
C/C++ basic types

The following are the basic type descriptors for C/C++ core types;

bool
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([5 x sbyte]* %str1, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 32, 
    uint 32, 
    uint 0, 
    uint 2 }, section "llvm.metadata"
%str1 = internal constant [5 x sbyte] c"bool\00", section "llvm.metadata"
char
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([5 x sbyte]* %str1, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 8, 
    uint 8, 
    uint 0, 
    uint 6 }, section "llvm.metadata"
%str1 = internal constant [5 x sbyte] c"char\00", section "llvm.metadata"
unsigned char
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([14 x sbyte]* %str1, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 8, 
    uint 8, 
    uint 0, 
    uint 8 }, section "llvm.metadata"
%str1 = internal constant [14 x sbyte] c"unsigned char\00", section "llvm.metadata"
short
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([10 x sbyte]* %str1, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 16, 
    uint 16, 
    uint 0, 
    uint 5 }, section "llvm.metadata"
%str1 = internal constant [10 x sbyte] c"short int\00", section "llvm.metadata"
unsigned short
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([19 x sbyte]* %str1, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 16, 
    uint 16, 
    uint 0, 
    uint 7 }, section "llvm.metadata"
%str1 = internal constant [19 x sbyte] c"short unsigned int\00", section "llvm.metadata"
int
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([4 x sbyte]* %str1, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 32, 
    uint 32, 
    uint 0, 
    uint 5 }, section "llvm.metadata"
%str1 = internal constant [4 x sbyte] c"int\00", section "llvm.metadata"
unsigned int
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([13 x sbyte]* %str1, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 32, 
    uint 32, 
    uint 0, 
    uint 7 }, section "llvm.metadata"
%str1 = internal constant [13 x sbyte] c"unsigned int\00", section "llvm.metadata"
long long
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([14 x sbyte]* %str1, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 64, 
    uint 64, 
    uint 0, 
    uint 5 }, section "llvm.metadata"
%str1 = internal constant [14 x sbyte] c"long long int\00", section "llvm.metadata"
unsigned long long
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([23 x sbyte]* %str1, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 64, 
    uint 64, 
    uint 0, 
    uint 7 }, section "llvm.metadata"
%str1 = internal constant [23 x sbyte] c"long long unsigned int\00", section "llvm.metadata"
float
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([6 x sbyte]* %str1, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 32, 
    uint 32, 
    uint 0, 
    uint 4 }, section "llvm.metadata"
%str1 = internal constant [6 x sbyte] c"float\00", section "llvm.metadata"
double
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([7 x sbyte]* %str1, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 64, 
    uint 64, 
    uint 0, 
    uint 4 }, section "llvm.metadata"
%str1 = internal constant [7 x sbyte] c"double\00", section "llvm.metadata"
C/C++ derived types

Given the following as an example of C/C++ derived type;

typedef const int *IntPtr;

a C/C++ front-end would generate the following descriptors;

;;
;; Define the typedef "IntPtr".
;;
%llvm.dbg.derivedtype1 = internal constant %llvm.dbg.derivedtype.type {
    uint add(uint 22, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([7 x sbyte]* %str1, int 0, int 0), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    int 1, 
    uint 0, 
    uint 0, 
    uint 0, 
    {  }* cast (%llvm.dbg.derivedtype.type* %llvm.dbg.derivedtype2 to {  }*) }, section "llvm.metadata"
%str1 = internal constant [7 x sbyte] c"IntPtr\00", section "llvm.metadata"

;;
;; Define the pointer type.
;;
%llvm.dbg.derivedtype2 = internal constant %llvm.dbg.derivedtype.type {
    uint add(uint 15, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* null, 
    {  }* null, 
    int 0, 
    uint 32, 
    uint 32, 
    uint 0, 
    {  }* cast (%llvm.dbg.derivedtype.type* %llvm.dbg.derivedtype3 to {  }*) }, section "llvm.metadata"

;;
;; Define the const type.
;;
%llvm.dbg.derivedtype3 = internal constant %llvm.dbg.derivedtype.type {
    uint add(uint 38, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* null, 
    {  }* null, 
    int 0, 
    uint 0, 
    uint 0, 
    uint 0, 
    {  }* cast (%llvm.dbg.basictype.type* %llvm.dbg.basictype1 to {  }*) }, section "llvm.metadata"	

;;
;; Define the int type.
;;
%llvm.dbg.basictype1 = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([4 x sbyte]* %str2, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 32, 
    uint 32, 
    uint 0, 
    uint 5 }, section "llvm.metadata"
%str2 = internal constant [4 x sbyte] c"int\00", section "llvm.metadata"
C/C++ struct/union types

Given the following as an example of C/C++ struct type;

struct Color {
  unsigned Red;
  unsigned Green;
  unsigned Blue;
};

a C/C++ front-end would generate the following descriptors;

;;
;; Define basic type for unsigned int.
;;
%llvm.dbg.basictype = internal constant %llvm.dbg.basictype.type {
    uint add(uint 36, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([13 x sbyte]* %str1, int 0, int 0), 
    {  }* null, 
    int 0, 
    uint 32, 
    uint 32, 
    uint 0, 
    uint 7 }, section "llvm.metadata"
%str1 = internal constant [13 x sbyte] c"unsigned int\00", section "llvm.metadata"

;;
;; Define composite type for struct Color.
;;
%llvm.dbg.compositetype = internal constant %llvm.dbg.compositetype.type {
    uint add(uint 19, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([6 x sbyte]* %str2, int 0, int 0), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    int 1, 
    uint 96, 
    uint 32, 
    uint 0, 
    {  }* null,
    {  }* cast ([3 x {  }*]* %llvm.dbg.array to {  }*) }, section "llvm.metadata"
%str2 = internal constant [6 x sbyte] c"Color\00", section "llvm.metadata"

;;
;; Define the Red field.
;;
%llvm.dbg.derivedtype1 = internal constant %llvm.dbg.derivedtype.type {
    uint add(uint 13, uint 262144), 
    {  }* null, 
    sbyte* getelementptr ([4 x sbyte]* %str3, int 0, int 0), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    int 2, 
    uint 32, 
    uint 32, 
    uint 0, 
    {  }* cast (%llvm.dbg.basictype.type* %llvm.dbg.basictype to {  }*) }, section "llvm.metadata"
%str3 = internal constant [4 x sbyte] c"Red\00", section "llvm.metadata"

;;
;; Define the Green field.
;;
%llvm.dbg.derivedtype2 = internal constant %llvm.dbg.derivedtype.type {
    uint add(uint 13, uint 262144), 
    {  }* null, 
    sbyte* getelementptr ([6 x sbyte]* %str4, int 0, int 0), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    int 3, 
    uint 32, 
    uint 32, 
    uint 32, 
    {  }* cast (%llvm.dbg.basictype.type* %llvm.dbg.basictype to {  }*) }, section "llvm.metadata"
%str4 = internal constant [6 x sbyte] c"Green\00", section "llvm.metadata"

;;
;; Define the Blue field.
;;
%llvm.dbg.derivedtype3 = internal constant %llvm.dbg.derivedtype.type {
    uint add(uint 13, uint 262144), 
    {  }* null, 
    sbyte* getelementptr ([5 x sbyte]* %str5, int 0, int 0), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    int 4, 
    uint 32, 
    uint 32, 
    uint 64, 
    {  }* cast (%llvm.dbg.basictype.type* %llvm.dbg.basictype to {  }*) }, section "llvm.metadata"
%str5 = internal constant [5 x sbyte] c"Blue\00", section "llvm.metadata"

;;
;; Define the array of fields used by the composite type Color.
;;
%llvm.dbg.array = internal constant [3 x {  }*] [
      {  }* cast (%llvm.dbg.derivedtype.type* %llvm.dbg.derivedtype1 to {  }*),
      {  }* cast (%llvm.dbg.derivedtype.type* %llvm.dbg.derivedtype2 to {  }*),
      {  }* cast (%llvm.dbg.derivedtype.type* %llvm.dbg.derivedtype3 to {  }*) ], section "llvm.metadata"
C/C++ enumeration types

Given the following as an example of C/C++ enumeration type;

enum Trees {
  Spruce = 100,
  Oak = 200,
  Maple = 300
};

a C/C++ front-end would generate the following descriptors;

;;
;; Define composite type for enum Trees
;;
%llvm.dbg.compositetype = internal constant %llvm.dbg.compositetype.type {
    uint add(uint 4, uint 262144), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    sbyte* getelementptr ([6 x sbyte]* %str1, int 0, int 0), 
    {  }* cast (%llvm.dbg.compile_unit.type* %llvm.dbg.compile_unit to {  }*), 
    int 1, 
    uint 32, 
    uint 32, 
    uint 0, 
    {  }* null, 
    {  }* cast ([3 x {  }*]* %llvm.dbg.array to {  }*) }, section "llvm.metadata"
%str1 = internal constant [6 x sbyte] c"Trees\00", section "llvm.metadata"

;;
;; Define Spruce enumerator.
;;
%llvm.dbg.enumerator1 = internal constant %llvm.dbg.enumerator.type {
    uint add(uint 40, uint 262144), 
    sbyte* getelementptr ([7 x sbyte]* %str2, int 0, int 0), 
    int 100 }, section "llvm.metadata"
%str2 = internal constant [7 x sbyte] c"Spruce\00", section "llvm.metadata"

;;
;; Define Oak enumerator.
;;
%llvm.dbg.enumerator2 = internal constant %llvm.dbg.enumerator.type {
    uint add(uint 40, uint 262144), 
    sbyte* getelementptr ([4 x sbyte]* %str3, int 0, int 0), 
    int 200 }, section "llvm.metadata"
%str3 = internal constant [4 x sbyte] c"Oak\00", section "llvm.metadata"

;;
;; Define Maple enumerator.
;;
%llvm.dbg.enumerator3 = internal constant %llvm.dbg.enumerator.type {
    uint add(uint 40, uint 262144), 
    sbyte* getelementptr ([6 x sbyte]* %str4, int 0, int 0), 
    int 300 }, section "llvm.metadata"
%str4 = internal constant [6 x sbyte] c"Maple\00", section "llvm.metadata"

;;
;; Define the array of enumerators used by composite type Trees.
;;
%llvm.dbg.array = internal constant [3 x {  }*] [
  {  }* cast (%llvm.dbg.enumerator.type* %llvm.dbg.enumerator1 to {  }*),
  {  }* cast (%llvm.dbg.enumerator.type* %llvm.dbg.enumerator2 to {  }*),
  {  }* cast (%llvm.dbg.enumerator.type* %llvm.dbg.enumerator3 to {  }*) ], section "llvm.metadata"

Valid CSS Valid HTML 4.01 Chris Lattner
LLVM Compiler Infrastructure
Last modified: $Date$