include/llvm-c/Disassembler.h

   1 /*===-- llvm-c/Disassembler.h - Disassembler Public C Interface ---*- C -*-===*\
   2 |*                                                                            *|
   3 |*                     The LLVM Compiler Infrastructure                       *|
   4 |*                                                                            *|
   5 |* This file is distributed under the University of Illinois Open Source      *|
   6 |* License. See LICENSE.TXT for details.                                      *|
   7 |*                                                                            *|
   8 |*===----------------------------------------------------------------------===*|
   9 |*                                                                            *|
  10 |* This header provides public interface to a disassembler library.           *|
  11 |* LLVM provides an implementation of this interface.                         *|
  12 |*                                                                            *|
  13 \*===----------------------------------------------------------------------===*/
  14
  15 #ifndef LLVM_C_DISASSEMBLER_H
  16 #define LLVM_C_DISASSEMBLER_H  1
  17
  18 #include <stddef.h>
  19 #include "llvm/Support/DataTypes.h"
  20
  21 /**
  22  * An opaque reference to a disassembler context.
  23  */
  24 typedef void *LLVMDisasmContextRef;
  25
  26 /**
  27  * The type for the operand information call back function.  This is called to
  28  * get the symbolic information for an operand of an instruction.  Typically
  29  * this is from the relocation information, symbol table, etc.  That block of
  30  * information is saved when the disassembler context is created and passed to
  31  * the call back in the DisInfo parameter.  The instruction containing operand
  32  * is at the PC parameter.  For some instruction sets, there can be more than
  33  * one operand with symbolic information.  To determine the symbolic operand
  34  * information for each operand, the bytes for the specific operand in the
  35  * instruction are specified by the Offset parameter and its byte widith is the
  36  * size parameter.  For instructions sets with fixed widths and one symbolic
  37  * operand per instruction, the Offset parameter will be zero and Size parameter
  38  * will be the instruction width.  The information is returned in TagBuf and is
  39  * Triple specific with its specific information defined by the value of
  40  * TagType for that Triple.  If symbolic information is returned the function
  41  * returns 1 else it returns 0.
  42  */
  43 typedef int (*LLVMOpInfoCallback)(void *DisInfo,
  44                                   uint64_t PC,
  45                                   uint64_t Offset,
  46                                   uint64_t Size,
  47                                   int TagType,
  48                                   void *TagBuf);
  49
  50 /**
  51  * The initial support in LLVM MC for the most general form of a relocatable
  52  * expression is "AddSymbol - SubtractSymbol + Offset".  For some Darwin targets
  53  * this full form is encoded in the relocation information so that AddSymbol and
  54  * SubtractSymbol can be link edited independent of each other.  Many other
  55  * platforms only allow a relocatable expression of the form AddSymbol + Offset
  56  * to be encoded.
  57  *
  58  * The LLVMOpInfoCallback() for the TagType value of 1 uses the struct
  59  * LLVMOpInfo1.  The value of the relocatable expression for the operand,
  60  * including any PC adjustment, is passed in to the call back in the Value
  61  * field.  The symbolic information about the operand is returned using all
  62  * the fields of the structure with the Offset of the relocatable expression
  63  * returned in the Value field.  It is possible that some symbols in the
  64  * relocatable expression were assembly temporary symbols, for example
  65  * "Ldata - LpicBase + constant", and only the Values of the symbols without
  66  * symbol names are present in the relocation information.  The VariantKind
  67  * type is one of the Target specific #defines below and is used to print
  68  * operands like "_foo@GOT", ":lower16:_foo", etc.
  69  */
  70 struct LLVMOpInfoSymbol1 {
  71   uint64_t Present; /* 1 if this symbol is present */
  72   char *Name;     /* symbol name if not NULL */
  73   uint64_t Value; /* symbol value if name is NULL */
  74 };
  75 struct LLVMOpInfo1 {
  76   struct LLVMOpInfoSymbol1 AddSymbol;
  77   struct LLVMOpInfoSymbol1 SubtractSymbol;
  78   uint64_t Value;
  79   uint64_t VariantKind;
  80 };
  81
  82 /**
  83  * The operand VariantKinds for symbolic disassembly.
  84  */
  85 #define LLVMDisassembler_VariantKind_None 0 /* all targets */
  86
  87 /**
  88  * The ARM target VariantKinds.
  89  */
  90 #define LLVMDisassembler_VariantKind_ARM_HI16 1 /* :upper16: */
  91 #define LLVMDisassembler_VariantKind_ARM_LO16 2 /* :lower16: */
  92
  93 /**
  94  * The type for the symbol lookup function.  This may be called by the
  95  * disassembler for such things like adding a comment for a PC plus a constant
  96  * offset load instruction to use a symbol name instead of a load address value.
  97  * It is passed the block information is saved when the disassembler context is
  98  * created and a value of a symbol to look up.  If no symbol is found NULL is
  99  * to be returned.
 100  */
 101 typedef const char *(*LLVMSymbolLookupCallback)(void *DisInfo,
 102                                                 uint64_t SymbolValue);
 103
 104 #ifdef __cplusplus
 105 extern "C" {
 106 #endif /* !defined(__cplusplus) */
 107
 108 /**
 109  * Create a disassembler for the TripleName.  Symbolic disassembly is supported
 110  * by passing a block of information in the DisInfo parameter and specifing the
 111  * TagType and call back functions as described above.  These can all be passed
 112  * as NULL.  If successful this returns a disassembler context if not it
 113  * returns NULL.
 114  */
 115 extern LLVMDisasmContextRef
 116 LLVMCreateDisasm(const char *TripleName,
 117                  void *DisInfo,
 118                  int TagType,
 119                  LLVMOpInfoCallback GetOpInfo,
 120                  LLVMSymbolLookupCallback SymbolLookUp);
 121
 122 /**
 123  * Dispose of a disassembler context.
 124  */
 125 extern void
 126 LLVMDisasmDispose(LLVMDisasmContextRef DC);
 127
 128 /**
 129  * Disassmble a single instruction using the disassembler context specified in
 130  * the parameter DC.  The bytes of the instruction are specified in the parameter
 131  * Bytes, and contains at least BytesSize number of bytes.  The instruction is
 132  * at the address specified by the PC parameter.  If a valid instruction can be
 133  * disassembled its string is returned indirectly in OutString which whos size
 134  * is specified in the parameter OutStringSize.  This function returns the
 135  * number of bytes in the instruction or zero if there was no valid instruction.
 136  */
 137 extern size_t
 138 LLVMDisasmInstruction(LLVMDisasmContextRef DC,
 139                       uint8_t *Bytes,
 140                       uint64_t BytesSize,
 141                       uint64_t PC,
 142                       char *OutString,
 143                       size_t OutStringSize);
 144
 145 #ifdef __cplusplus
 146 }
 147 #endif /* !defined(__cplusplus) */
 148
 149 #endif /* !defined(LLVM_C_DISASSEMBLER_H) */