audk/ArmPkg/Library/ArmDisassemblerLib/ThumbDisassembler.c

825 lines
33 KiB
C
Raw Normal View History

/** @file
Thumb Dissassembler. Still a work in progress.
Wrong output is a bug, so please fix it.
Hex output means there is not yet an entry or a decode bug.
gOpThumb[] are Thumb 16-bit, and gOpThumb2[] work on the 32-bit
16-bit stream of Thumb2 instruction. Then there are big case
statements to print everything out. If you are adding instructions
try to reuse existing case entries if possible.
Copyright (c) 2008-2010, Apple Inc. All rights reserved.
All rights reserved. This program and the accompanying materials
are licensed and made available under the terms and conditions of the BSD License
which accompanies this distribution. The full text of the license may be found at
http://opensource.org/licenses/bsd-license.php
THE PROGRAM IS DISTRIBUTED UNDER THE BSD LICENSE ON AN "AS IS" BASIS,
WITHOUT WARRANTIES OR REPRESENTATIONS OF ANY KIND, EITHER EXPRESS OR IMPLIED.
**/
#include <Base.h>
#include <Library/BaseLib.h>
#include <Library/DebugLib.h>
#include <Library/PrintLib.h>
extern CHAR8 *gCondition[];
extern CHAR8 *gReg[];
// Thumb address modes
#define LOAD_STORE_FORMAT1 1
#define LOAD_STORE_FORMAT1_H 101
#define LOAD_STORE_FORMAT1_B 111
#define LOAD_STORE_FORMAT2 2
#define LOAD_STORE_FORMAT3 3
#define LOAD_STORE_FORMAT4 4
#define LOAD_STORE_MULTIPLE_FORMAT1 5
#define PUSH_FORMAT 6
#define POP_FORMAT 106
#define IMMED_8 7
#define CONDITIONAL_BRANCH 8
#define UNCONDITIONAL_BRANCH 9
#define UNCONDITIONAL_BRANCH_SHORT 109
#define BRANCH_EXCHANGE 10
#define DATA_FORMAT1 11
#define DATA_FORMAT2 12
#define DATA_FORMAT3 13
#define DATA_FORMAT4 14
#define DATA_FORMAT5 15
#define DATA_FORMAT6_SP 16
#define DATA_FORMAT6_PC 116
#define DATA_FORMAT7 17
#define DATA_FORMAT8 19
#define CPS_FORMAT 20
#define ENDIAN_FORMAT 21
#define DATA_CBZ 22
#define ADR_FORMAT 23
// Thumb2 address modes
#define B_T3 200
#define B_T4 201
#define BL_T2 202
#define POP_T2 203
#define POP_T3 204
#define STM_FORMAT 205
#define LDM_REG_IMM12_SIGNED 206
#define LDM_REG_IMM12_LSL 207
#define LDM_REG_IMM8 208
#define LDM_REG_IMM12 209
#define LDM_REG_INDIRECT_LSL 210
#define LDM_REG_IMM8_SIGNED 211
#define LDRD_REG_IMM8 212
#define LDREXB 213
#define LDREXD 214
#define SRS_FORMAT 215
#define RFE_FORMAT 216
#define LDRD_REG_IMM8_SIGNED 217
#define ADD_IMM12 218
#define ADD_IMM5 219
#define ADR_THUMB2 220
#define CMN_THUMB2 221
typedef struct {
CHAR8 *Start;
UINT32 OpCode;
UINT32 Mask;
UINT32 AddressMode;
} THUMB_INSTRUCTIONS;
THUMB_INSTRUCTIONS gOpThumb[] = {
// Thumb 16-bit instrucitons
// Op Mask Format
{ "ADC" , 0x4140, 0xffc0, DATA_FORMAT5 },
{ "ADR", 0xa000, 0xf800, ADR_FORMAT }, // ADR <Rd>, <label>
{ "ADD" , 0x1c00, 0xfe00, DATA_FORMAT2 },
{ "ADD" , 0x3000, 0xf800, DATA_FORMAT3 },
{ "ADD" , 0x1800, 0xfe00, DATA_FORMAT1 },
{ "ADD" , 0x4400, 0xff00, DATA_FORMAT8 }, // A8.6.9
{ "ADD" , 0xa000, 0xf100, DATA_FORMAT6_PC },
{ "ADD" , 0xa800, 0xf800, DATA_FORMAT6_SP },
{ "ADD" , 0xb000, 0xff80, DATA_FORMAT7 },
{ "AND" , 0x4000, 0xffc0, DATA_FORMAT5 },
{ "ASR" , 0x1000, 0xf800, DATA_FORMAT4 },
{ "ASR" , 0x4100, 0xffc0, DATA_FORMAT5 },
{ "B" , 0xd000, 0xf000, CONDITIONAL_BRANCH },
{ "B" , 0xe000, 0xf800, UNCONDITIONAL_BRANCH_SHORT },
{ "BLX" , 0x4780, 0xff80, BRANCH_EXCHANGE },
{ "BX" , 0x4700, 0xff87, BRANCH_EXCHANGE },
{ "BIC" , 0x4380, 0xffc0, DATA_FORMAT5 },
{ "BKPT", 0xdf00, 0xff00, IMMED_8 },
{ "CBZ", 0xb100, 0xfd00, DATA_CBZ },
{ "CBNZ", 0xb900, 0xfd00, DATA_CBZ },
{ "CMN" , 0x42c0, 0xffc0, DATA_FORMAT5 },
{ "CMP" , 0x2800, 0xf800, DATA_FORMAT3 },
{ "CMP" , 0x4280, 0xffc0, DATA_FORMAT5 },
{ "CMP" , 0x4500, 0xff00, DATA_FORMAT8 },
{ "CPS" , 0xb660, 0xffe8, CPS_FORMAT },
{ "MOV" , 0x4600, 0xff00, DATA_FORMAT8 },
{ "EOR" , 0x4040, 0xffc0, DATA_FORMAT5 },
{ "LDMIA" , 0xc800, 0xf800, LOAD_STORE_MULTIPLE_FORMAT1 },
{ "LDR" , 0x6800, 0xf800, LOAD_STORE_FORMAT1 },
{ "LDR" , 0x5800, 0xfe00, LOAD_STORE_FORMAT2 },
{ "LDR" , 0x4800, 0xf800, LOAD_STORE_FORMAT3 },
{ "LDR" , 0x9800, 0xf800, LOAD_STORE_FORMAT4 },
{ "LDRB" , 0x7800, 0xf800, LOAD_STORE_FORMAT1_B },
{ "LDRB" , 0x5c00, 0xfe00, LOAD_STORE_FORMAT2 },
{ "LDRH" , 0x8800, 0xf800, LOAD_STORE_FORMAT1_H },
{ "LDRH" , 0x7a00, 0xfe00, LOAD_STORE_FORMAT2 },
{ "LDRSB" , 0x5600, 0xfe00, LOAD_STORE_FORMAT2 },
{ "LDRSH" , 0x5e00, 0xfe00, LOAD_STORE_FORMAT2 },
{ "MOVS", 0x0000, 0xffc0, DATA_FORMAT5 }, // LSL with imm5 == 0 is a MOVS, so this must go before LSL
{ "LSL" , 0x0000, 0xf800, DATA_FORMAT4 },
{ "LSL" , 0x4080, 0xffc0, DATA_FORMAT5 },
{ "LSR" , 0x0001, 0xf800, DATA_FORMAT4 },
{ "LSR" , 0x40c0, 0xffc0, DATA_FORMAT5 },
{ "MOVS", 0x2000, 0xf800, DATA_FORMAT3 },
{ "MOV" , 0x1c00, 0xffc0, DATA_FORMAT3 },
{ "MOV" , 0x4600, 0xff00, DATA_FORMAT8 },
{ "MUL" , 0x4340, 0xffc0, DATA_FORMAT5 },
{ "MVN" , 0x41c0, 0xffc0, DATA_FORMAT5 },
{ "NEG" , 0x4240, 0xffc0, DATA_FORMAT5 },
{ "ORR" , 0x4180, 0xffc0, DATA_FORMAT5 },
{ "POP" , 0xbc00, 0xfe00, POP_FORMAT },
{ "PUSH", 0xb400, 0xfe00, PUSH_FORMAT },
{ "REV" , 0xba00, 0xffc0, DATA_FORMAT5 },
{ "REV16" , 0xba40, 0xffc0, DATA_FORMAT5 },
{ "REVSH" , 0xbac0, 0xffc0, DATA_FORMAT5 },
{ "ROR" , 0x41c0, 0xffc0, DATA_FORMAT5 },
{ "SBC" , 0x4180, 0xffc0, DATA_FORMAT5 },
{ "SETEND" , 0xb650, 0xfff0, ENDIAN_FORMAT },
{ "STMIA" , 0xc000, 0xf800, LOAD_STORE_MULTIPLE_FORMAT1 },
{ "STR" , 0x6000, 0xf800, LOAD_STORE_FORMAT1 },
{ "STR" , 0x5000, 0xfe00, LOAD_STORE_FORMAT2 },
{ "STR" , 0x4000, 0xf800, LOAD_STORE_FORMAT3 },
{ "STR" , 0x9000, 0xf800, LOAD_STORE_FORMAT4 },
{ "STRB" , 0x7000, 0xf800, LOAD_STORE_FORMAT1_B },
{ "STRB" , 0x5800, 0xfe00, LOAD_STORE_FORMAT2 },
{ "STRH" , 0x8000, 0xf800, LOAD_STORE_FORMAT1_H },
{ "STRH" , 0x5200, 0xfe00, LOAD_STORE_FORMAT2 },
{ "SUB" , 0x1e00, 0xfe00, DATA_FORMAT2 },
{ "SUB" , 0x3800, 0xf800, DATA_FORMAT3 },
{ "SUB" , 0x1a00, 0xfe00, DATA_FORMAT1 },
{ "SUB" , 0xb080, 0xff80, DATA_FORMAT7 },
{ "SWI" , 0xdf00, 0xff00, IMMED_8 },
{ "SXTB", 0xb240, 0xffc0, DATA_FORMAT5 },
{ "SXTH", 0xb200, 0xffc0, DATA_FORMAT5 },
{ "TST" , 0x4200, 0xffc0, DATA_FORMAT5 },
{ "UXTB", 0xb2c0, 0xffc0, DATA_FORMAT5 },
{ "UXTH", 0xb280, 0xffc0, DATA_FORMAT5 }
};
THUMB_INSTRUCTIONS gOpThumb2[] = {
//Instruct OpCode OpCode Mask Addressig Mode
{ "ADR", 0xf2af0000, 0xfbff8000, ADR_THUMB2 }, // ADDR <Rd>, <label> ;Needs to go before ADDW
{ "CMN", 0xf1100f00, 0xfff08f00, CMN_THUMB2 }, // CMN <Rn>, <Rm>, {,<shift> #<const>} ;Needs to go before ADD
{ "ADC", 0xf1400000, 0xfbe08000, ADD_IMM12 }, // ADC{S} <Rd>, <Rn>, #<const>
{ "ADC", 0xeb400000, 0xffe08000, ADD_IMM5 }, // ADC{S} <Rd>, <Rn>, <Rm> {,<shift> #<const>}
{ "ADD", 0xf1000000, 0xfbe08000, ADD_IMM12 }, // ADD{S} <Rd>, <Rn>, #<const>
{ "ADD", 0xeb000000, 0xffe08000, ADD_IMM5 }, // ADD{S} <Rd>, <Rn>, <Rm> {,<shift> #<const>}
{ "ADDW", 0xf2000000, 0xfbe08000, ADD_IMM12 }, // ADDW{S} <Rd>, <Rn>, #<const>
{ "AND", 0xf0000000, 0xfbe08000, ADD_IMM12 }, // AND{S} <Rd>, <Rn>, #<const>
{ "AND", 0xea000000, 0xffe08000, ADD_IMM5 }, // AND{S} <Rd>, <Rn>, <Rm> {,<shift> #<const>}
{ "BIC", 0xf0200000, 0xfbe08000, ADD_IMM12 }, // BIC{S} <Rd>, <Rn>, #<const>
{ "BIC", 0xea200000, 0xffe08000, ADD_IMM5 }, // BIC{S} <Rd>, <Rn>, <Rm> {,<shift> #<const>}
{ "EOR", 0xf0800000, 0xfbe08000, ADD_IMM12 }, // EOR{S} <Rd>, <Rn>, #<const>
{ "EOR", 0xea800000, 0xffe08000, ADD_IMM5 }, // EOR{S} <Rd>, <Rn>, <Rm> {,<shift> #<const>}
{ "ORN", 0xf0600000, 0xfbe08000, ADD_IMM12 }, // ORN{S} <Rd>, <Rn>, #<const>
{ "ORN", 0xea600000, 0xffe08000, ADD_IMM5 }, // ORN{S} <Rd>, <Rn>, <Rm> {,<shift> #<const>}
{ "ORR", 0xf0400000, 0xfbe08000, ADD_IMM12 }, // ORR{S} <Rd>, <Rn>, #<const>
{ "ORR", 0xea400000, 0xffe08000, ADD_IMM5 }, // ORR{S} <Rd>, <Rn>, <Rm> {,<shift> #<const>}
{ "RSB", 0xf1c00000, 0xfbe08000, ADD_IMM12 }, // RSB{S} <Rd>, <Rn>, #<const>
{ "RSB", 0xebc00000, 0xffe08000, ADD_IMM5 }, // RSB{S} <Rd>, <Rn>, <Rm> {,<shift> #<const>}
{ "SBC", 0xf1600000, 0xfbe08000, ADD_IMM12 }, // SBC{S} <Rd>, <Rn>, #<const>
{ "SBC", 0xeb600000, 0xffe08000, ADD_IMM5 }, // SBC{S} <Rd>, <Rn>, <Rm> {,<shift> #<const>}
{ "SUB", 0xf1a00000, 0xfbe08000, ADD_IMM12 }, // SUB{S} <Rd>, <Rn>, #<const>
{ "SUB", 0xeba00000, 0xffe08000, ADD_IMM5 }, // SUB{S} <Rd>, <Rn>, <Rm> {,<shift> #<const>}
{ "B", 0xf0008000, 0xf800d000, B_T3 }, // B<c> <label>
{ "B", 0xf0009000, 0xf800d000, B_T4 }, // B<c> <label>
{ "BL", 0xf000d000, 0xf800d000, B_T4 }, // BL<c> <label>
{ "BLX", 0xf000c000, 0xf800d000, BL_T2 }, // BLX<c> <label>
{ "POP", 0xe8bd0000, 0xffff2000, POP_T2 }, // POP <registers>
{ "POP", 0xf85d0b04, 0xffff0fff, POP_T3 }, // POP <register>
{ "PUSH", 0xe8ad0000, 0xffffa000, POP_T2 }, // PUSH <registers>
{ "PUSH", 0xf84d0d04, 0xffff0fff, POP_T3 }, // PUSH <register>
{ "STM" , 0xe8800000, 0xffd0a000, STM_FORMAT }, // STM <Rn>{!},<registers>
{ "STMDB", 0xe9800000, 0xffd0a000, STM_FORMAT }, // STMDB <Rn>{!},<registers>
{ "LDM" , 0xe8900000, 0xffd02000, STM_FORMAT }, // LDM <Rn>{!},<registers>
{ "LDMDB", 0xe9100000, 0xffd02000, STM_FORMAT }, // LDMDB <Rn>{!},<registers>
{ "LDR", 0xf8d00000, 0xfff00000, LDM_REG_IMM12 }, // LDR <rt>, [<rn>, {, #<imm12>]}
{ "LDRB", 0xf8900000, 0xfff00000, LDM_REG_IMM12 }, // LDRB <rt>, [<rn>, {, #<imm12>]}
{ "LDRH", 0xf8b00000, 0xfff00000, LDM_REG_IMM12 }, // LDRH <rt>, [<rn>, {, #<imm12>]}
{ "LDRSB", 0xf9900000, 0xfff00000, LDM_REG_IMM12 }, // LDRSB <rt>, [<rn>, {, #<imm12>]}
{ "LDRSH", 0xf9b00000, 0xfff00000, LDM_REG_IMM12 }, // LDRSH <rt>, [<rn>, {, #<imm12>]}
{ "LDR", 0xf85f0000, 0xff7f0000, LDM_REG_IMM12_SIGNED }, // LDR <Rt>, <label>
{ "LDRB", 0xf81f0000, 0xff7f0000, LDM_REG_IMM12_SIGNED }, // LDRB <Rt>, <label>
{ "LDRH", 0xf83f0000, 0xff7f0000, LDM_REG_IMM12_SIGNED }, // LDRH <Rt>, <label>
{ "LDRSB", 0xf91f0000, 0xff7f0000, LDM_REG_IMM12_SIGNED }, // LDRSB <Rt>, <label>
{ "LDRSH", 0xf93f0000, 0xff7f0000, LDM_REG_IMM12_SIGNED }, // LDRSB <Rt>, <label>
{ "LDR", 0xf8500000, 0xfff00fc0, LDM_REG_INDIRECT_LSL }, // LDR <rt>, [<rn>, <rm> {, LSL #<imm2>]}
{ "LDRB", 0xf8100000, 0xfff00fc0, LDM_REG_INDIRECT_LSL }, // LDRB <rt>, [<rn>, <rm> {, LSL #<imm2>]}
{ "LDRH", 0xf8300000, 0xfff00fc0, LDM_REG_INDIRECT_LSL }, // LDRH <rt>, [<rn>, <rm> {, LSL #<imm2>]}
{ "LDRSB", 0xf9100000, 0xfff00fc0, LDM_REG_INDIRECT_LSL }, // LDRSB <rt>, [<rn>, <rm> {, LSL #<imm2>]}
{ "LDRSH", 0xf9300000, 0xfff00fc0, LDM_REG_INDIRECT_LSL }, // LDRSH <rt>, [<rn>, <rm> {, LSL #<imm2>]}
{ "LDR", 0xf8500800, 0xfff00800, LDM_REG_IMM8 }, // LDR <rt>, [<rn>, {, #<imm8>]}
{ "LDRBT", 0xf8100e00, 0xfff00f00, LDM_REG_IMM8 }, // LDRBT <rt>, [<rn>, {, #<imm8>]}
{ "LDRHT", 0xf8300e00, 0xfff00f00, LDM_REG_IMM8 }, // LDRHT <rt>, [<rn>, {, #<imm8>]}
{ "LDRSB", 0xf9900800, 0xfff00800, LDM_REG_IMM8 }, // LDRHT <rt>, [<rn>, {, #<imm8>]} {!} form?
{ "LDRSBT",0xf9100e00, 0xfff00f00, LDM_REG_IMM8 }, // LDRHBT <rt>, [<rn>, {, #<imm8>]} {!} form?
{ "LDRSH" ,0xf9300800, 0xfff00800, LDM_REG_IMM8 }, // LDRSH <rt>, [<rn>, {, #<imm8>]}
{ "LDRSHT",0xf9300e00, 0xfff00f00, LDM_REG_IMM8 }, // LDRSHT <rt>, [<rn>, {, #<imm8>]}
{ "LDRT", 0xf8500e00, 0xfff00f00, LDM_REG_IMM8 }, // LDRT <rt>, [<rn>, {, #<imm8>]}
{ "LDRD", 0xe8500000, 0xfe500000, LDRD_REG_IMM8_SIGNED }, // LDRD <rt>, <rt2>, [<rn>, {, #<imm8>]}{!}
{ "LDRD", 0xe8500000, 0xfe500000, LDRD_REG_IMM8 }, // LDRD <rt>, <rt2>, <label>
{ "LDREX", 0xe8500f00, 0xfff00f00, LDM_REG_IMM8 }, // LDREX <Rt>, [Rn, {#imm8}]]
{ "LDREXB", 0xe8d00f4f, 0xfff00fff, LDREXB }, // LDREXB <Rt>, [<Rn>]
{ "LDREXH", 0xe8d00f5f, 0xfff00fff, LDREXB }, // LDREXH <Rt>, [<Rn>]
{ "LDREXD", 0xe8d00f4f, 0xfff00fff, LDREXD }, // LDREXD <Rt>, <Rt2>, [<Rn>]
{ "STR", 0xf8c00000, 0xfff00000, LDM_REG_IMM12 }, // STR <rt>, [<rn>, {, #<imm12>]}
{ "STRB", 0xf8800000, 0xfff00000, LDM_REG_IMM12 }, // STRB <rt>, [<rn>, {, #<imm12>]}
{ "STRH", 0xf8a00000, 0xfff00000, LDM_REG_IMM12 }, // STRH <rt>, [<rn>, {, #<imm12>]}
{ "STR", 0xf8400000, 0xfff00fc0, LDM_REG_INDIRECT_LSL }, // STR <rt>, [<rn>, <rm> {, LSL #<imm2>]}
{ "STRB", 0xf8000000, 0xfff00fc0, LDM_REG_INDIRECT_LSL }, // STRB <rt>, [<rn>, <rm> {, LSL #<imm2>]}
{ "STRH", 0xf8200000, 0xfff00fc0, LDM_REG_INDIRECT_LSL }, // STRH <rt>, [<rn>, <rm> {, LSL #<imm2>]}
{ "STR", 0xf8400800, 0xfff00800, LDM_REG_IMM8 }, // STR <rt>, [<rn>, {, #<imm8>]}
{ "STRH", 0xf8200800, 0xfff00800, LDM_REG_IMM8 }, // STRH <rt>, [<rn>, {, #<imm8>]}
{ "STRBT", 0xf8000e00, 0xfff00f00, LDM_REG_IMM8 }, // STRBT <rt>, [<rn>, {, #<imm8>]}
{ "STRHT", 0xf8200e00, 0xfff00f00, LDM_REG_IMM8 }, // STRHT <rt>, [<rn>, {, #<imm8>]}
{ "STRT", 0xf8400e00, 0xfff00f00, LDM_REG_IMM8 }, // STRT <rt>, [<rn>, {, #<imm8>]}
{ "STRD", 0xe8400000, 0xfe500000, LDRD_REG_IMM8_SIGNED }, // STRD <rt>, <rt2>, [<rn>, {, #<imm8>]}{!}
{ "STREX", 0xe8400f00, 0xfff00f00, LDM_REG_IMM8 }, // STREX <Rt>, [Rn, {#imm8}]]
{ "STREXB", 0xe8c00f4f, 0xfff00fff, LDREXB }, // STREXB <Rd>, <Rt>, [<Rn>]
{ "STREXH", 0xe8c00f5f, 0xfff00fff, LDREXB }, // STREXH <Rd>, <Rt>, [<Rn>]
{ "STREXD", 0xe8d00f4f, 0xfff00fff, LDREXD }, // STREXD <Rd>, <Rt>, <Rt2>, [<Rn>]
{ "SRSDB", 0xe80dc000, 0xffdffff0, SRS_FORMAT }, // SRSDB<c> SP{!},#<mode>
{ "SRS" , 0xe98dc000, 0xffdffff0, SRS_FORMAT }, // SRS{IA}<c> SP{!},#<mode>
{ "RFEDB", 0xe810c000, 0xffd0ffff, RFE_FORMAT }, // RFEDB<c> <Rn>{!}
{ "RFE" , 0xe990c000, 0xffd0ffff, RFE_FORMAT } // RFE{IA}<c> <Rn>{!}
};
CHAR8 *gShiftType[] = {
"LSL",
"LSR",
"ASR",
"ROR"
};
CHAR8 mThumbMregListStr[4*15 + 1];
CHAR8 *
ThumbMRegList (
UINT32 RegBitMask
)
{
UINTN Index, Start, End;
CHAR8 *Str;
BOOLEAN First;
Str = mThumbMregListStr;
*Str = '\0';
AsciiStrCat (Str, "{");
for (Index = 0, First = TRUE; Index <= 15; Index++) {
if ((RegBitMask & (1 << Index)) != 0) {
Start = End = Index;
for (Index++; ((RegBitMask & (1 << Index)) != 0) && (Index <= 9); Index++) {
End = Index;
}
if (!First) {
AsciiStrCat (Str, ",");
} else {
First = FALSE;
}
if (Start == End) {
AsciiStrCat (Str, gReg[Start]);
} else {
AsciiStrCat (Str, gReg[Start]);
AsciiStrCat (Str, "-");
AsciiStrCat (Str, gReg[End]);
}
}
}
if (First) {
AsciiStrCat (Str, "ERROR");
}
AsciiStrCat (Str, "}");
// BugBug: Make caller pass in buffer it is cleaner
return mThumbMregListStr;
}
UINT32
SignExtend32 (
IN UINT32 Data,
IN UINT32 TopBit
)
{
if (((Data & TopBit) == 0) || (TopBit == BIT31)) {
return Data;
}
do {
TopBit <<= 1;
Data |= TopBit;
} while ((TopBit & BIT31) != BIT31);
return Data;
}
//
// Some instructions specify the PC is always considered aligned
// The PC is after the instruction that is excuting. So you pass
// in the instruction address and you get back the aligned answer
//
PCAlign4 (
IN UINT32 Data
)
{
return (Data + 4) & 0xfffffffc;
}
/**
Place a dissasembly of of **OpCodePtr into buffer, and update OpCodePtr to
point to next instructin.
We cheat and only decode instructions that access
memory. If the instruction is not found we dump the instruction in hex.
@param OpCodePtrPtr Pointer to pointer of ARM Thumb instruction to disassemble.
@param Buf Buffer to sprintf disassembly into.
@param Size Size of Buf in bytes.
@param Extended TRUE dump hex for instruction too.
**/
VOID
DisassembleThumbInstruction (
IN UINT16 **OpCodePtrPtr,
OUT CHAR8 *Buf,
OUT UINTN Size,
IN BOOLEAN Extended
)
{
UINT16 *OpCodePtr;
UINT16 OpCode;
UINT32 OpCode32;
UINT32 Index;
UINT32 Offset;
UINT16 Rd, Rn, Rm, Rt, Rt2;
BOOLEAN H1, H2, imod;
UINT32 PC, Target;
CHAR8 *Cond;
BOOLEAN S, J1, J2, P, U, W;
OpCodePtr = *OpCodePtrPtr;
OpCode = **OpCodePtrPtr;
// Thumb2 is a stream of 16-bit instructions not a 32-bit instruction.
OpCode32 = (((UINT32)OpCode) << 16) | *(OpCodePtr + 1);
// These register names match branch form, but not others
Rd = OpCode & 0x7;
Rn = (OpCode >> 3) & 0x7;
Rm = (OpCode >> 6) & 0x7;
H1 = (OpCode & BIT7) != 0;
H2 = (OpCode & BIT6) != 0;
imod = (OpCode & BIT4) != 0;
PC = (UINT32)(UINTN)OpCodePtr;
// Increment by the minimum instruction size, Thumb2 could be bigger
*OpCodePtrPtr += 1;
for (Index = 0; Index < sizeof (gOpThumb)/sizeof (THUMB_INSTRUCTIONS); Index++) {
if ((OpCode & gOpThumb[Index].Mask) == gOpThumb[Index].OpCode) {
if (Extended) {
Offset = AsciiSPrint (Buf, Size, "0x%04x %-6a", OpCode, gOpThumb[Index].Start);
} else {
Offset = AsciiSPrint (Buf, Size, "%-6a", gOpThumb[Index].Start);
}
switch (gOpThumb[Index].AddressMode) {
case LOAD_STORE_FORMAT1:
// A6.5.1 <Rd>, [<Rn>, #<5_bit_offset>]
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, [r%d #0x%x]", Rd, Rn, (OpCode >> 4) & 0x7c);
return;
case LOAD_STORE_FORMAT1_H:
// A6.5.1 <Rd>, [<Rn>, #<5_bit_offset>]
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, [r%d #0x%x]", Rd, Rn, (OpCode >> 5) & 0x3f);
return;
case LOAD_STORE_FORMAT1_B:
// A6.5.1 <Rd>, [<Rn>, #<5_bit_offset>]
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, [r%d #0x%x]", Rd, Rn, (OpCode >> 6) & 0x1f);
return;
case LOAD_STORE_FORMAT2:
// A6.5.1 <Rd>, [<Rn>, <Rm>]
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, [r%d, r%d]", Rd, Rn, Rm);
return;
case LOAD_STORE_FORMAT3:
// A6.5.1 <Rd>, [PC, #<8_bit_offset>]
Target = (OpCode & 0xff) << 2;
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, [pc, #0x%x] ;0x%08x", (OpCode >> 8) & 7, Target, PCAlign4 (PC) + Target);
return;
case LOAD_STORE_FORMAT4:
// Rt, [SP, #imm8]
Target = (OpCode & 0xff) << 2;
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, [sp, #0x%x]", (OpCode >> 8) & 7, Target);
return;
case LOAD_STORE_MULTIPLE_FORMAT1:
// <Rn>!, {r0-r7}
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d!, %a", (OpCode >> 8) & 7, ThumbMRegList (OpCode & 0xff));
return;
case POP_FORMAT:
// POP {r0-r7,pc}
AsciiSPrint (&Buf[Offset], Size - Offset, " %a", ThumbMRegList ((OpCode & 0xff) | ((OpCode & BIT8) == BIT8 ? BIT15 : 0)));
return;
case PUSH_FORMAT:
// PUSH {r0-r7,lr}
AsciiSPrint (&Buf[Offset], Size - Offset, " %a", ThumbMRegList ((OpCode & 0xff) | ((OpCode & BIT8) == BIT8 ? BIT14 : 0)));
return;
case IMMED_8:
// A6.7 <immed_8>
AsciiSPrint (&Buf[Offset], Size - Offset, " 0x%x", OpCode & 0xff);
return;
case CONDITIONAL_BRANCH:
// A6.3.1 B<cond> <target_address>
// Patch in the condition code. A little hack but based on "%-6a"
Cond = gCondition[(OpCode >> 8) & 0xf];
Buf[Offset-5] = *Cond++;
Buf[Offset-4] = *Cond;
AsciiSPrint (&Buf[Offset], Size - Offset, " 0x%04x", PC + 4 + SignExtend32 ((OpCode & 0xff) << 1, BIT8));
return;
case UNCONDITIONAL_BRANCH_SHORT:
// A6.3.2 B <target_address>
AsciiSPrint (&Buf[Offset], Size - Offset, " 0x%04x", PC + 4 + SignExtend32 ((OpCode & 0x3ff) << 1, BIT11));
return;
case BRANCH_EXCHANGE:
// A6.3.3 BX|BLX <Rm>
AsciiSPrint (&Buf[Offset], Size - Offset, " %a", gReg[Rn | (H2 ? 8:0)]);
return;
case DATA_FORMAT1:
// A6.4.3 <Rd>, <Rn>, <Rm>
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, r%d, r%d", Rd, Rn, Rm);
return;
case DATA_FORMAT2:
// A6.4.3 <Rd>, <Rn>, #3_bit_immed
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, r%d, 0x%x", Rd, Rn, Rm);
return;
case DATA_FORMAT3:
// A6.4.3 <Rd>|<Rn>, #imm8
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, #0x%x", (OpCode >> 8) & 7, OpCode & 0xff);
return;
case DATA_FORMAT4:
// A6.4.3 <Rd>|<Rm>, #immed_5
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, r%d, 0x%x", Rn, Rd, (OpCode >> 6) & 0x1f);
return;
case DATA_FORMAT5:
// A6.4.3 <Rd>|<Rm>, <Rm>|<Rs>
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, r%d", Rd, Rn);
return;
case DATA_FORMAT6_SP:
// A6.4.3 <Rd>, <reg>, #<8_Bit_immed>
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, sp, 0x%x", (OpCode >> 8) & 7, (OpCode & 0xff) << 2);
return;
case DATA_FORMAT6_PC:
// A6.4.3 <Rd>, <reg>, #<8_Bit_immed>
AsciiSPrint (&Buf[Offset], Size - Offset, " r%d, pc, 0x%x", (OpCode >> 8) & 7, (OpCode & 0xff) << 2);
return;
case DATA_FORMAT7:
// A6.4.3 SP, SP, #<7_Bit_immed>
AsciiSPrint (&Buf[Offset], Size - Offset, " sp, sp, 0x%x", (OpCode & 0x7f)*4);
return;
case DATA_FORMAT8:
// A6.4.3 <Rd>|<Rn>, <Rm>
AsciiSPrint (&Buf[Offset], Size - Offset, " %a, %a", gReg[Rd | (H1 ? 8:0)], gReg[Rn | (H2 ? 8:0)]);
return;
case CPS_FORMAT:
// A7.1.24
AsciiSPrint (&Buf[Offset], Size - Offset, "%a %a%a%a", imod ? "ID":"IE", ((OpCode & BIT2) == 0) ? "":"a", ((OpCode & BIT1) == 0) ? "":"i", ((OpCode & BIT0) == 0) ? "":"f");
return;
case ENDIAN_FORMAT:
// A7.1.24
AsciiSPrint (&Buf[Offset], Size - Offset, " %a", (OpCode & BIT3) == 0 ? "LE":"BE");
return;
case DATA_CBZ:
// CB{N}Z <Rn>, <Lable>
Target = ((OpCode >> 2) & 0x3e) | (((OpCode & BIT9) == BIT9) ? BIT6 : 0);
AsciiSPrint (&Buf[Offset], Size - Offset, " %a, %08x", gReg[Rd], PC + 4 + Target);
return;
case ADR_FORMAT:
// ADR <Rd>, <Label>
Target = (OpCode & 0xff) << 2;
AsciiSPrint (&Buf[Offset], Size - Offset, " %a, %08x", gReg[(OpCode >> 8) & 7], PCAlign4 (PC) + Target);
return;
}
}
}
// Thumb2 are 32-bit instructions
*OpCodePtrPtr += 1;
Rt = (OpCode32 >> 12) & 0xf;
Rt2 = (OpCode32 >> 8) & 0xf;
Rd = (OpCode32 >> 8) & 0xf;
Rm = (OpCode32 & 0xf);
Rn = (OpCode32 >> 16) & 0xf;
for (Index = 0; Index < sizeof (gOpThumb2)/sizeof (THUMB_INSTRUCTIONS); Index++) {
if ((OpCode32 & gOpThumb2[Index].Mask) == gOpThumb2[Index].OpCode) {
if (Extended) {
Offset = AsciiSPrint (Buf, Size, "0x%04x %-6a", OpCode32, gOpThumb2[Index].Start);
} else {
Offset = AsciiSPrint (Buf, Size, " %-6a", gOpThumb2[Index].Start);
}
switch (gOpThumb2[Index].AddressMode) {
case B_T3:
Cond = gCondition[(OpCode32 >> 22) & 0xf];
Buf[Offset-5] = *Cond++;
Buf[Offset-4] = *Cond;
// S:J2:J1:imm6:imm11:0
Target = ((OpCode32 << 1) & 0xffe) + ((OpCode32 >> 4) & 0x3f000);
Target |= ((OpCode32 & BIT11) == BIT11)? BIT19 : 0; // J2
Target |= ((OpCode32 & BIT13) == BIT13)? BIT18 : 0; // J1
Target |= ((OpCode32 & BIT26) == BIT26)? BIT20 : 0; // S
Target = SignExtend32 (Target, BIT20);
AsciiSPrint (&Buf[Offset], Size - Offset, " 0x%08x", PC + 4 + Target);
return;
case B_T4:
// S:I1:I2:imm10:imm11:0
Target = ((OpCode32 << 1) & 0xffe) + ((OpCode32 >> 4) & 0x3ff000);
S = (OpCode32 & BIT26) == BIT26;
J1 = (OpCode32 & BIT13) == BIT13;
J2 = (OpCode32 & BIT11) == BIT11;
Target |= (!(J2 ^ S) ? BIT22 : 0); // I2
Target |= (!(J1 ^ S) ? BIT23 : 0); // I1
Target |= (S ? BIT24 : 0); // S
Target = SignExtend32 (Target, BIT24);
AsciiSPrint (&Buf[Offset], Size - Offset, " 0x%08x", PC + 4 + Target);
return;
case BL_T2:
// BLX S:I1:I2:imm10:imm11:0
Target = ((OpCode32 << 1) & 0xffc) + ((OpCode32 >> 4) & 0x3ff000);
S = (OpCode32 & BIT26) == BIT26;
J1 = (OpCode32 & BIT13) == BIT13;
J2 = (OpCode32 & BIT11) == BIT11;
Target |= (!(J2 ^ S) ? BIT23 : 0); // I2
Target |= (!(J1 ^ S) ? BIT24 : 0); // I1
Target |= (S ? BIT25 : 0); // S
Target = SignExtend32 (Target, BIT25);
AsciiSPrint (&Buf[Offset], Size - Offset, " 0x%08x", PCAlign4 (PC) + Target);
return;
case POP_T2:
// <reglist> some must be zero, handled in table
AsciiSPrint (&Buf[Offset], Size - Offset, " %a", ThumbMRegList (OpCode32 & 0xffff));
return;
case POP_T3:
// <register>
AsciiSPrint (&Buf[Offset], Size - Offset, " %a", gReg[(OpCode32 >> 12) & 0xf]);
return;
case STM_FORMAT:
// <Rn>{!}, <registers>
W = (OpCode32 & BIT21) == BIT21;
AsciiSPrint (&Buf[Offset], Size - Offset, " %a%a, %a", gReg[(OpCode32 >> 16) & 0xf], W ? "!":"", ThumbMRegList (OpCode32 & 0xffff));
return;
case LDM_REG_IMM12_SIGNED:
// <rt>, <label>
Target = OpCode32 & 0xfff;
if ((OpCode32 & BIT23) == 0) {
// U == 0 means subtrack, U == 1 means add
Target = -Target;
}
AsciiSPrint (&Buf[Offset], Size - Offset, " %a, %a", gReg[(OpCode32 >> 12) & 0xf], PCAlign4 (PC) + Target);
return;
case LDM_REG_INDIRECT_LSL:
// <rt>, [<rn>, <rm> {, LSL #<imm2>]}
Offset += AsciiSPrint (&Buf[Offset], Size - Offset, " %a, [%a, %a", gReg[Rt], gReg[Rn], gReg[Rm]);
if (((OpCode32 >> 4) && 3) == 0) {
AsciiSPrint (&Buf[Offset], Size - Offset, "]");
} else {
AsciiSPrint (&Buf[Offset], Size - Offset, ", LSL #%d]", (OpCode32 >> 4) && 3);
}
return;
case LDM_REG_IMM12:
// <rt>, [<rn>, {, #<imm12>]}
Offset += AsciiSPrint (&Buf[Offset], Size - Offset, " %a, [%a", gReg[Rt], gReg[Rn]);
if ((OpCode32 && 0xfff) == 0) {
AsciiSPrint (&Buf[Offset], Size - Offset, "]");
} else {
AsciiSPrint (&Buf[Offset], Size - Offset, ", #0x%x]", OpCode32 & 0xfff);
}
return;
case LDM_REG_IMM8:
ASSERT (FALSE);
// <rt>, [<rn>, {, #<imm8>}]{!}
W = (OpCode32 & BIT8) == BIT8;
U = (OpCode32 & BIT9) == BIT9;
P = (OpCode32 & BIT10) == BIT10;
Offset += AsciiSPrint (&Buf[Offset], Size - Offset, " %a, [%a", gReg[Rt], gReg[Rn]);
if (P) {
if ((OpCode32 && 0xff) == 0) {
AsciiSPrint (&Buf[Offset], Size - Offset, "]%a", W?"!":"");
} else {
AsciiSPrint (&Buf[Offset], Size - Offset, ", #%a0x%x]%a", OpCode32 & 0xff, U?"":"-" , W?"!":"");
}
} else {
AsciiSPrint (&Buf[Offset], Size - Offset, "], #%a0x%x]", OpCode32 & 0xff, U?"":"-");
}
return;
case LDRD_REG_IMM8_SIGNED:
// LDRD <rt>, <rt2>, [<rn>, {, #<imm8>]}{!}
P = (OpCode32 & BIT24) == BIT24; // index = P
U = (OpCode32 & BIT23) == BIT23;
W = (OpCode32 & BIT21) == BIT21;
Offset += AsciiSPrint (&Buf[Offset], Size - Offset, " %a, %a, [%a", gReg[Rt], gReg[Rt2], gReg[Rn]);
if (P) {
if ((OpCode32 && 0xff) == 0) {
AsciiSPrint (&Buf[Offset], Size - Offset, "]");
} else {
AsciiSPrint (&Buf[Offset], Size - Offset, ", #%a0x%x]%a", U?"":"-", (OpCode32 & 0xff) << 2, W?"!":"");
}
} else {
if ((OpCode32 && 0xff) != 0) {
AsciiSPrint (&Buf[Offset], Size - Offset, ", #%a0x%x", U?"":"-", (OpCode32 & 0xff) << 2);
}
}
return;
case LDRD_REG_IMM8:
// LDRD <rt>, <rt2>, <label>
Target = (OpCode32 & 0xff) << 2;
if ((OpCode32 & BIT23) == 0) {
// U == 0 means subtrack, U == 1 means add
Target = -Target;
}
AsciiSPrint (&Buf[Offset], Size - Offset, " %a, %a, %a", gReg[Rt], gReg[Rt2], PC + 4 + Target);
return;
case LDREXB:
// LDREXB <Rt>, [Rn]
AsciiSPrint (&Buf[Offset], Size - Offset, " %a, [%a]", gReg[Rt], gReg[Rn]);
return;
case LDREXD:
// LDREXD <Rt>, <Rt2>, [<Rn>]
AsciiSPrint (&Buf[Offset], Size - Offset, " %a, ,%a, [%a]", gReg[Rt], gReg[Rt2], gReg[Rn]);
return;
case SRS_FORMAT:
// SP{!}, #<mode>
W = (OpCode32 & BIT21) == BIT21;
AsciiSPrint (&Buf[Offset], Size - Offset, " SP%a, #0x%x", W?"!":"", OpCode32 & 0x1f);
return;
case RFE_FORMAT:
// <Rn>{!}
AsciiSPrint (&Buf[Offset], Size - Offset, " %a%a, #0x%x", gReg[Rn], W?"!":"");
return;
case ADD_IMM12:
// ADD{S} <Rd>, <Rn>, #<const> i:imm3:imm8
if ((OpCode32 & BIT20) == BIT20) {
Buf[Offset - 3] = 'S'; // assume %-6a
}
Target = (OpCode32 & 0xff) | ((OpCode32 >> 4) & 0x700) | ((OpCode & BIT26) == BIT26 ? BIT11 : 0);
AsciiSPrint (&Buf[Offset], Size - Offset, " %a, ,%a, #0x%x", gReg[Rd], gReg[Rn], Target);
return;
case ADD_IMM5:
// ADC <Rd>, <Rn>, <Rm> {,LSL #<const>} imm3:imm2
if ((OpCode32 & BIT20) == BIT20) {
Buf[Offset - 3] = 'S'; // assume %-6a
}
Target = ((OpCode32 >> 6) & 3) | ((OpCode32 >> 10) & 0x1c0);
Offset += AsciiSPrint (&Buf[Offset], Size - Offset, " %a, ,%a, %a", gReg[Rd], gReg[Rn], gReg[Rm]);
if (Target != 0) {
AsciiSPrint (&Buf[Offset], Size - Offset, ", LSL %d", gShiftType[(OpCode >> 5) & 3], Target);
}
return;
case ADR_THUMB2:
// ADDR <Rd>, <label>
Target = (OpCode32 & 0xff) | ((OpCode32 >> 8) & 0x700) | ((OpCode & BIT26) == BIT26 ? BIT11 : 0);
if ((OpCode & (BIT23 | BIT21)) == (BIT23 | BIT21)) {
Target = PCAlign4 (PC) - Target;
} else {
Target = PCAlign4 (PC) + Target;
}
AsciiSPrint (&Buf[Offset], Size - Offset, " %a, 0x%08x", gReg[Rd], Target);
return;
case CMN_THUMB2:
// CMN <Rn>, <Rm>, {,<shift> #<const>}
if ((OpCode32 & BIT20) == BIT20) {
Buf[Offset - 3] = 'S'; // assume %-6a
}
Target = ((OpCode32 >> 6) & 3) | ((OpCode32 >> 10) & 0x1c0);
Offset += AsciiSPrint (&Buf[Offset], Size - Offset, " %a, ,%a", gReg[Rn], gReg[Rm]);
if (Target != 0) {
AsciiSPrint (&Buf[Offset], Size - Offset, ", LSL %d", gShiftType[(OpCode >> 5) & 3], Target);
}
return;
}
}
}
AsciiSPrint (Buf, Size, "0x%08x", OpCode32);
}
VOID
DisassembleArmInstruction (
IN UINT32 **OpCodePtr,
OUT CHAR8 *Buf,
OUT UINTN Size,
IN BOOLEAN Extended
);
/**
Place a dissasembly of of **OpCodePtr into buffer, and update OpCodePtr to
point to next instructin.
We cheat and only decode instructions that access
memory. If the instruction is not found we dump the instruction in hex.
@param OpCodePtrPtr Pointer to pointer of ARM Thumb instruction to disassemble.
@param Thumb TRUE for Thumb(2), FALSE for ARM instruction stream
@param Extended TRUE dump hex for instruction too.
@param Buf Buffer to sprintf disassembly into.
@param Size Size of Buf in bytes.
**/
VOID
DisassembleInstruction (
IN UINT8 **OpCodePtr,
IN BOOLEAN Thumb,
IN BOOLEAN Extended,
OUT CHAR8 *Buf,
OUT UINTN Size
)
{
if (Thumb) {
DisassembleThumbInstruction ((UINT16 **)OpCodePtr, Buf, Size, Extended);
} else {
DisassembleArmInstruction ((UINT32 **)OpCodePtr, Buf, Size, Extended);
}
}