15821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Copyright (c) 2012 The Chromium Authors. All rights reserved.
25821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Use of this source code is governed by a BSD-style license that can be
35821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// found in the LICENSE file.
45821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
55821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Definition of MiniDisassembler.
65821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
75821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)#ifndef TRACELINE_SIDESTEP_MINI_DISASSEMBLER_H_
85821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)#define TRACELINE_SIDESTEP_MINI_DISASSEMBLER_H_
95821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
105821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)#include "sidestep/mini_disassembler_types.h"
115821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
125821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)namespace sidestep {
135821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
145821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// This small disassembler is very limited
155821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// in its functionality, and in fact does only the bare minimum required by the
165821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// preamble patching utility.  It may be useful for other purposes, however.
175821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//
185821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// The limitations include at least the following:
195821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//  -# No support for coprocessor opcodes, MMX, etc.
205821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//  -# No machine-readable identification of opcodes or decoding of
215821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//     assembly parameters. The name of the opcode (as a string) is given,
225821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//     however, to aid debugging.
235821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//
245821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// You may ask what this little disassembler actually does, then?  The answer is
255821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// that it does the following, which is exactly what the patching utility needs:
265821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//  -# Indicates if opcode is a jump (any kind) or a return (any kind)
275821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//     because this is important for the patching utility to determine if
285821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//     a function is too short or there are jumps too early in it for it
295821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//     to be preamble patched.
305821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//  -# The opcode length is always calculated, so that the patching utility
315821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//     can figure out where the next instruction starts, and whether it
325821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//     already has enough instructions to replace with the absolute jump
335821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//     to the patching code.
345821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//
355821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// The usage is quite simple; just create a MiniDisassembler and use its
365821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Disassemble() method.
375821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)//
385821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// If you would like to extend this disassembler, please refer to the
395821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// IA-32 Intel Architecture Software Developer's Manual Volume 2:
405821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// Instruction Set Reference for information about operand decoding
415821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)// etc.
425821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)class MiniDisassembler {
435821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) public:
445821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
455821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Creates a new instance and sets defaults.
465821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  //
475821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // operand_default_32_bits: If true, the default operand size is
485821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // set to 32 bits, which is the default under Win32. Otherwise it is 16 bits.
495821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // address_default_32_bits: If true, the default address size is
505821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // set to 32 bits, which is the default under Win32. Otherwise it is 16 bits.
515821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  MiniDisassembler(bool operand_default_32_bits,
525821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                   bool address_default_32_bits);
535821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
545821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Equivalent to MiniDisassembler(true, true);
555821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  MiniDisassembler();
565821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
575821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Attempts to disassemble a single instruction starting from the
585821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // address in memory it is pointed to.
595821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  //
605821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // start: Address where disassembly should start.
615821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // instruction_bytes: Variable that will be incremented by
625821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // the length in bytes of the instruction.
635821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Returns enItJump, enItReturn or enItGeneric on success.  enItUnknown
645821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // if unable to disassemble, enItUnused if this seems to be an unused
655821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // opcode. In the last two (error) cases, cbInstruction will be set
665821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // to 0xffffffff.
675821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  //
685821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Postcondition: This instance of the disassembler is ready to be used again,
695821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // with unchanged defaults from creation time.
705821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  InstructionType Disassemble(unsigned char* start,
715821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                              unsigned int* instruction_bytes);
725821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
735821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles) private:
745821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
755821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Makes the disassembler ready for reuse.
765821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  void Initialize();
775821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
785821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Sets the flags for address and operand sizes.
795821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Returns Number of prefix bytes.
805821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  InstructionType ProcessPrefixes(unsigned char* start, unsigned int* size);
815821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
825821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Sets the flag for whether we have ModR/M, and increments
835821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // operand_bytes_ if any are specifies by the opcode directly.
845821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Returns Number of opcode bytes.
855821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  InstructionType ProcessOpcode(unsigned char* start,
865821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                                unsigned int table,
875821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)                                unsigned int* size);
885821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
895821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Checks the type of the supplied operand.  Increments
905821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // operand_bytes_ if it directly indicates an immediate etc.
915821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // operand.  Asserts have_modrm_ if the operand specifies
925821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // a ModR/M byte.
935821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  bool ProcessOperand(int flag_operand);
945821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
955821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Increments operand_bytes_ by size specified by ModR/M and
965821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // by SIB if present.
975821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Returns 0 in case of error, 1 if there is just a ModR/M byte,
985821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // 2 if there is a ModR/M byte and a SIB byte.
995821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  bool ProcessModrm(unsigned char* start, unsigned int* size);
1005821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1015821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Processes the SIB byte that it is pointed to.
1025821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // start: Pointer to the SIB byte.
1035821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // mod: The mod field from the ModR/M byte.
1045821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Returns 1 to indicate success (indicates 1 SIB byte)
1055821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  bool ProcessSib(unsigned char* start, unsigned char mod, unsigned int* size);
1065821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1075821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // The instruction type we have decoded from the opcode.
1085821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  InstructionType instruction_type_;
1095821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1105821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Counts the number of bytes that is occupied by operands in
1115821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // the current instruction (note: we don't care about how large
1125821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // operands stored in registers etc. are).
1135821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  unsigned int operand_bytes_;
1145821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1155821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // True iff there is a ModR/M byte in this instruction.
1165821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  bool have_modrm_;
1175821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1185821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // True iff we need to decode the ModR/M byte (sometimes it just
1195821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // points to a register, we can tell by the addressing mode).
1205821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  bool should_decode_modrm_;
1215821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1225821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Current operand size is 32 bits if true, 16 bits if false.
1235821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  bool operand_is_32_bits_;
1245821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1255821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Default operand size is 32 bits if true, 16 bits if false.
1265821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  bool operand_default_is_32_bits_;
1275821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1285821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Current address size is 32 bits if true, 16 bits if false.
1295821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  bool address_is_32_bits_;
1305821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1315821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Default address size is 32 bits if true, 16 bits if false.
1325821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  bool address_default_is_32_bits_;
1335821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1345821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Huge big opcode table based on the IA-32 manual, defined
1355821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // in Ia32OpcodeMap.cpp
1365821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  static const OpcodeTable s_ia32_opcode_map_[];
1375821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1385821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Somewhat smaller table to help with decoding ModR/M bytes
1395821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // when 16-bit addressing mode is being used.  Defined in
1405821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Ia32ModrmMap.cpp
1415821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  static const ModrmEntry s_ia16_modrm_map_[];
1425821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1435821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Somewhat smaller table to help with decoding ModR/M bytes
1445821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // when 32-bit addressing mode is being used.  Defined in
1455821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Ia32ModrmMap.cpp
1465821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  static const ModrmEntry s_ia32_modrm_map_[];
1475821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1485821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // Indicators of whether we got certain prefixes that certain
1495821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // silly Intel instructions depend on in nonstandard ways for
1505821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  // their behaviors.
1515821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)  bool got_f2_prefix_, got_f3_prefix_, got_66_prefix_;
1525821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)};
1535821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1545821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)};  // namespace sidestep
1555821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)
1565821806d5e7f356e8fa4b058a389a808ea183019Torne (Richard Coles)#endif  // TRACELINE_SIDESTEP_MINI_DISASSEMBLER_H_
157