Nguyen Anh Quynh | f49df92 | 2017-05-16 18:15:02 +0700 | [diff] [blame] | 1 | #!/usr/bin/python |
| 2 | # Test tool to disassemble MC files. By Nguyen Anh Quynh, 2017 |
| 3 | import array, os.path, sys |
| 4 | from capstone import * |
| 5 | |
| 6 | |
| 7 | # convert all hex numbers to decimal numbers in a text |
| 8 | def normalize_hex(a): |
| 9 | while(True): |
| 10 | i = a.find('0x') |
| 11 | if i == -1: # no more hex number |
| 12 | break |
| 13 | hexnum = '0x' |
| 14 | for c in a[i + 2:]: |
| 15 | if c in '0123456789abcdefABCDEF': |
| 16 | hexnum += c |
| 17 | else: |
| 18 | break |
| 19 | num = int(hexnum, 16) |
| 20 | a = a.replace(hexnum, str(num)) |
| 21 | return a |
| 22 | |
| 23 | |
| 24 | def test_file(fname): |
| 25 | print("Test %s" %fname); |
| 26 | f = open(fname) |
| 27 | lines = f.readlines() |
| 28 | f.close() |
| 29 | |
| 30 | if not lines[0].startswith('# '): |
| 31 | print("ERROR: decoding information is missing") |
| 32 | return |
| 33 | |
| 34 | # skip '# ' at the front, then split line to get out hexcode |
| 35 | # Note: option can be '', or 'None' |
| 36 | #print lines[0] |
| 37 | #print lines[0][2:].split(', ') |
| 38 | (arch, mode, option) = lines[0][2:].split(', ') |
| 39 | mode = mode.replace(' ', '') |
| 40 | option = option.strip() |
| 41 | |
| 42 | archs = { |
| 43 | "CS_ARCH_ARM": CS_ARCH_ARM, |
| 44 | "CS_ARCH_ARM64": CS_ARCH_ARM64, |
| 45 | "CS_ARCH_MIPS": CS_ARCH_MIPS, |
| 46 | "CS_ARCH_PPC": CS_ARCH_PPC, |
| 47 | "CS_ARCH_SPARC": CS_ARCH_SPARC, |
| 48 | "CS_ARCH_SYSZ": CS_ARCH_SYSZ, |
| 49 | "CS_ARCH_X86": CS_ARCH_X86, |
| 50 | "CS_ARCH_XCORE": CS_ARCH_XCORE, |
| 51 | "CS_ARCH_M68K": CS_ARCH_M68K, |
| 52 | } |
| 53 | |
| 54 | modes = { |
| 55 | "CS_MODE_16": CS_MODE_16, |
| 56 | "CS_MODE_32": CS_MODE_32, |
| 57 | "CS_MODE_64": CS_MODE_64, |
| 58 | "CS_MODE_MIPS32": CS_MODE_MIPS32, |
| 59 | "CS_MODE_MIPS64": CS_MODE_MIPS64, |
| 60 | "0": CS_MODE_ARM, |
| 61 | "CS_MODE_ARM": CS_MODE_ARM, |
| 62 | "CS_MODE_THUMB": CS_MODE_THUMB, |
| 63 | "CS_MODE_ARM+CS_MODE_V8": CS_MODE_ARM+CS_MODE_V8, |
| 64 | "CS_MODE_THUMB+CS_MODE_V8": CS_MODE_THUMB+CS_MODE_V8, |
| 65 | "CS_MODE_THUMB+CS_MODE_MCLASS": CS_MODE_THUMB+CS_MODE_MCLASS, |
| 66 | "CS_MODE_LITTLE_ENDIAN": CS_MODE_LITTLE_ENDIAN, |
| 67 | "CS_MODE_BIG_ENDIAN": CS_MODE_BIG_ENDIAN, |
| 68 | "CS_MODE_64+CS_MODE_LITTLE_ENDIAN": CS_MODE_64+CS_MODE_LITTLE_ENDIAN, |
| 69 | "CS_MODE_64+CS_MODE_BIG_ENDIAN": CS_MODE_64+CS_MODE_BIG_ENDIAN, |
| 70 | "CS_MODE_MIPS32+CS_MODE_MICRO": CS_MODE_MIPS32+CS_MODE_MICRO, |
| 71 | "CS_MODE_MIPS32+CS_MODE_MICRO+CS_MODE_BIG_ENDIAN": CS_MODE_MIPS32+CS_MODE_MICRO+CS_MODE_BIG_ENDIAN, |
| 72 | "CS_MODE_MIPS32+CS_MODE_BIG_ENDIAN+CS_MODE_MICRO": CS_MODE_MIPS32+CS_MODE_MICRO+CS_MODE_BIG_ENDIAN, |
| 73 | "CS_MODE_BIG_ENDIAN+CS_MODE_V9": CS_MODE_BIG_ENDIAN + CS_MODE_V9, |
| 74 | "CS_MODE_MIPS32+CS_MODE_BIG_ENDIAN": CS_MODE_MIPS32+CS_MODE_BIG_ENDIAN, |
| 75 | "CS_MODE_MIPS32+CS_MODE_LITTLE_ENDIAN": CS_MODE_MIPS32+CS_MODE_LITTLE_ENDIAN, |
| 76 | "CS_MODE_MIPS64+CS_MODE_LITTLE_ENDIAN": CS_MODE_MIPS64+CS_MODE_LITTLE_ENDIAN, |
| 77 | "CS_MODE_MIPS64+CS_MODE_BIG_ENDIAN": CS_MODE_MIPS64+CS_MODE_BIG_ENDIAN, |
| 78 | } |
| 79 | |
| 80 | options = { |
| 81 | "CS_OPT_SYNTAX_ATT": CS_OPT_SYNTAX_ATT, |
| 82 | "CS_OPT_SYNTAX_NOREGNAME": CS_OPT_SYNTAX_NOREGNAME, |
| 83 | } |
| 84 | |
| 85 | mc_modes = { |
| 86 | ("CS_ARCH_X86", "CS_MODE_32"): ['-triple=i386'], |
| 87 | ("CS_ARCH_X86", "CS_MODE_64"): ['-triple=x86_64'], |
| 88 | ("CS_ARCH_ARM", "CS_MODE_ARM"): ['-triple=armv7'], |
| 89 | ("CS_ARCH_ARM", "CS_MODE_THUMB"): ['-triple=thumbv7'], |
| 90 | ("CS_ARCH_ARM", "CS_MODE_ARM+CS_MODE_V8"): ['-triple=armv8'], |
| 91 | ("CS_ARCH_ARM", "CS_MODE_THUMB+CS_MODE_V8"): ['-triple=thumbv8'], |
| 92 | ("CS_ARCH_ARM", "CS_MODE_THUMB+CS_MODE_MCLASS"): ['-triple=thumbv7m'], |
| 93 | ("CS_ARCH_ARM64", "0"): ['-triple=aarch64'], |
| 94 | ("CS_ARCH_MIPS", "CS_MODE_MIPS32+CS_MODE_BIG_ENDIAN"): ['-triple=mips'], |
| 95 | ("CS_ARCH_MIPS", "CS_MODE_MIPS32+CS_MODE_MICRO"): ['-triple=mipsel', '-mattr=+micromips'], |
| 96 | ("CS_ARCH_MIPS", "CS_MODE_MIPS64"): ['-triple=mips64el'], |
| 97 | ("CS_ARCH_MIPS", "CS_MODE_MIPS32"): ['-triple=mipsel'], |
| 98 | ("CS_ARCH_MIPS", "CS_MODE_MIPS64+CS_MODE_BIG_ENDIAN"): ['-triple=mips64'], |
| 99 | ("CS_ARCH_MIPS", "CS_MODE_MIPS32+CS_MODE_MICRO+CS_MODE_BIG_ENDIAN"): ['-triple=mips', '-mattr=+micromips'], |
| 100 | ("CS_ARCH_MIPS", "CS_MODE_MIPS32+CS_MODE_BIG_ENDIAN+CS_MODE_MICRO"): ['-triple=mips', '-mattr=+micromips'], |
| 101 | ("CS_ARCH_PPC", "CS_MODE_BIG_ENDIAN"): ['-triple=powerpc64'], |
| 102 | ('CS_ARCH_SPARC', 'CS_MODE_BIG_ENDIAN'): ['-triple=sparc'], |
| 103 | ('CS_ARCH_SPARC', 'CS_MODE_BIG_ENDIAN+CS_MODE_V9'): ['-triple=sparcv9'], |
| 104 | ('CS_ARCH_SYSZ', '0'): ['-triple=s390x', '-mcpu=z196'], |
| 105 | } |
| 106 | |
| 107 | #if not option in ('', 'None'): |
| 108 | # print archs[arch], modes[mode], options[option] |
| 109 | |
| 110 | #print(arch, mode, option) |
| 111 | md = Cs(archs[arch], modes[mode]) |
| 112 | |
| 113 | if arch == 'CS_ARCH_ARM' or arch == 'CS_ARCH_PPC' : |
| 114 | md.syntax = CS_OPT_SYNTAX_NOREGNAME |
| 115 | |
| 116 | if fname.endswith('3DNow.s.cs'): |
| 117 | md.syntax = CS_OPT_SYNTAX_ATT |
| 118 | |
| 119 | for line in lines[1:]: |
| 120 | # ignore all the input lines having # in front. |
| 121 | if line.startswith('#'): |
| 122 | continue |
| 123 | #print("Check %s" %line) |
| 124 | code = line.split(' = ')[0] |
| 125 | asm = ''.join(line.split(' = ')[1:]) |
| 126 | hex_code = code.replace('0x', '') |
| 127 | hex_code = hex_code.replace(',', '') |
| 128 | hex_data = hex_code.decode('hex') |
| 129 | #hex_bytes = array.array('B', hex_data) |
| 130 | |
| 131 | x = list(md.disasm(hex_data, 0)) |
| 132 | if len(x) > 0: |
| 133 | if x[0].op_str != '': |
| 134 | cs_output = "%s %s" %(x[0].mnemonic, x[0].op_str) |
| 135 | else: |
| 136 | cs_output = x[0].mnemonic |
| 137 | else: |
| 138 | cs_output = 'FAILED to disassemble' |
| 139 | |
| 140 | cs_output2 = normalize_hex(cs_output) |
| 141 | cs_output2 = cs_output2.replace(' ', '') |
| 142 | |
| 143 | if arch == 'CS_ARCH_MIPS': |
| 144 | # normalize register alias names |
| 145 | cs_output2 = cs_output2.replace('$at', '$1') |
| 146 | cs_output2 = cs_output2.replace('$v0', '$2') |
| 147 | cs_output2 = cs_output2.replace('$v1', '$3') |
| 148 | |
| 149 | cs_output2 = cs_output2.replace('$a0', '$4') |
| 150 | cs_output2 = cs_output2.replace('$a1', '$5') |
| 151 | cs_output2 = cs_output2.replace('$a2', '$6') |
| 152 | cs_output2 = cs_output2.replace('$a3', '$7') |
| 153 | |
| 154 | cs_output2 = cs_output2.replace('$t0', '$8') |
| 155 | cs_output2 = cs_output2.replace('$t1', '$9') |
| 156 | cs_output2 = cs_output2.replace('$t2', '$10') |
| 157 | cs_output2 = cs_output2.replace('$t3', '$11') |
| 158 | cs_output2 = cs_output2.replace('$t4', '$12') |
| 159 | cs_output2 = cs_output2.replace('$t5', '$13') |
| 160 | cs_output2 = cs_output2.replace('$t6', '$14') |
| 161 | cs_output2 = cs_output2.replace('$t7', '$15') |
| 162 | cs_output2 = cs_output2.replace('$t8', '$24') |
| 163 | cs_output2 = cs_output2.replace('$t9', '$25') |
| 164 | |
| 165 | cs_output2 = cs_output2.replace('$s0', '$16') |
| 166 | cs_output2 = cs_output2.replace('$s1', '$17') |
| 167 | cs_output2 = cs_output2.replace('$s2', '$18') |
| 168 | cs_output2 = cs_output2.replace('$s3', '$19') |
| 169 | cs_output2 = cs_output2.replace('$s4', '$20') |
| 170 | cs_output2 = cs_output2.replace('$s5', '$21') |
| 171 | cs_output2 = cs_output2.replace('$s6', '$22') |
| 172 | cs_output2 = cs_output2.replace('$s7', '$23') |
| 173 | |
| 174 | cs_output2 = cs_output2.replace('$k0', '$26') |
| 175 | cs_output2 = cs_output2.replace('$k1', '$27') |
| 176 | |
| 177 | print("\t%s = %s" %(hex_code, cs_output)) |
| 178 | |
| 179 | |
| 180 | if __name__ == '__main__': |
| 181 | if len(sys.argv) == 1: |
| 182 | fnames = sys.stdin.readlines() |
| 183 | for fname in fnames: |
| 184 | test_file(fname.strip()) |
| 185 | else: |
| 186 | #print("Usage: ./test_mc.py <input-file.s.cs>") |
| 187 | test_file(sys.argv[1]) |
| 188 | |