Virtualization
/
capstone
mirror of https://gitlab.com/qemu-project/capstone.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

suite: update test_mc.py to better handle output of different formats of MC & CS

This commit is contained in:
Nguyen Anh Quynh 2014-11-10 21:50:54 +08:00
parent 6999d22892
commit df7dde26c9
1 changed files with 100 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

111
suite/test_mc.py
View File

@ -4,17 +4,44 @@ import array, os.path, sys
from subprocess import Popen, PIPE, STDOUT from subprocess import Popen, PIPE, STDOUT
from capstone import * from capstone import *
# convert all hex numbers to decimal numbers in a text
def normalize_hex(a):
while(True):
i = a.find('0x')
if i == -1: # no more hex number
break
hexnum = '0x'
for c in a[i + 2:]:
if c in '0123456789abcdefABCDEF':
hexnum += c
else:
break
num = int(hexnum, 16)
a = a.replace(hexnum, str(num))
return a
def run_mc(arch, hexcode, option, syntax=None): def run_mc(arch, hexcode, option, syntax=None):
def normalize(text): def normalize(text):
# remove tabs # remove tabs
text = text.lower()
items = text.split() items = text.split()
text = ' '.join(items) text = ' '.join(items)
if arch == CS_ARCH_X86: if arch == CS_ARCH_X86:
# remove comment after # # remove comment after #
i = text.find('# ') i = text.find('# ')
if i != -1: if i != -1:
return text[:i].lower() return text[:i].strip()
return text.lower() if arch == CS_ARCH_ARM64:
# remove comment after #
i = text.find('// ')
if i != -1:
return text[:i].strip()
# remove some redundant spaces
text = text.replace('{ ', '{')
text = text.replace(' }', '}')
return text.strip()
#print("Trying to decode: %s" %hexcode) #print("Trying to decode: %s" %hexcode)
if syntax: if syntax:
@ -32,7 +59,7 @@ def run_mc(arch, hexcode, option, syntax=None):
#print lines #print lines
if 'invalid' in lines[0]: if 'invalid' in lines[0]:
#print 'invalid ----' #print 'invalid ----'
return 'FAILED to disassemble' return 'FAILED to disassemble (MC)'
else: else:
#print 'OK:', lines[1] #print 'OK:', lines[1]
return normalize(lines[1].strip()) return normalize(lines[1].strip())
@ -73,6 +100,9 @@ def test_file(fname):
"0": CS_MODE_ARM, "0": CS_MODE_ARM,
"CS_MODE_ARM": CS_MODE_ARM, "CS_MODE_ARM": CS_MODE_ARM,
"CS_MODE_THUMB": CS_MODE_THUMB, "CS_MODE_THUMB": CS_MODE_THUMB,
"CS_MODE_ARM+CS_MODE_V8": CS_MODE_ARM+CS_MODE_V8,
"CS_MODE_THUMB+CS_MODE_V8": CS_MODE_THUMB+CS_MODE_V8,
"CS_MODE_THUMB+CS_MODE_MCLASS": CS_MODE_THUMB+CS_MODE_MCLASS,
"CS_MODE_LITTLE_ENDIAN": CS_MODE_LITTLE_ENDIAN, "CS_MODE_LITTLE_ENDIAN": CS_MODE_LITTLE_ENDIAN,
"CS_MODE_BIG_ENDIAN": CS_MODE_BIG_ENDIAN, "CS_MODE_BIG_ENDIAN": CS_MODE_BIG_ENDIAN,
"CS_MODE_32+CS_MODE_BIG_ENDIAN": CS_MODE_32+CS_MODE_BIG_ENDIAN, "CS_MODE_32+CS_MODE_BIG_ENDIAN": CS_MODE_32+CS_MODE_BIG_ENDIAN,
@ -94,7 +124,10 @@ def test_file(fname):
("CS_ARCH_X86", "CS_MODE_32"): ['-triple=i386'], ("CS_ARCH_X86", "CS_MODE_32"): ['-triple=i386'],
("CS_ARCH_X86", "CS_MODE_64"): ['-triple=x86_64'], ("CS_ARCH_X86", "CS_MODE_64"): ['-triple=x86_64'],
("CS_ARCH_ARM", "CS_MODE_ARM"): ['-triple=armv7'], ("CS_ARCH_ARM", "CS_MODE_ARM"): ['-triple=armv7'],
("CS_ARCH_ARM", "CS_MODE_THUMB"): ['-triple=armv7'], ("CS_ARCH_ARM", "CS_MODE_THUMB"): ['-triple=thumbv7'],
("CS_ARCH_ARM", "CS_MODE_ARM+CS_MODE_V8"): ['-triple=armv8'],
("CS_ARCH_ARM", "CS_MODE_THUMB+CS_MODE_V8"): ['-triple=thumbv8'],
("CS_ARCH_ARM", "CS_MODE_THUMB+CS_MODE_MCLASS"): ['-triple=thumbv7m'],
("CS_ARCH_ARM64", "0"): ['-triple=aarch64'], ("CS_ARCH_ARM64", "0"): ['-triple=aarch64'],
("CS_ARCH_MIPS", "CS_MODE_32+CS_MODE_BIG_ENDIAN"): ['-triple=mips'], ("CS_ARCH_MIPS", "CS_MODE_32+CS_MODE_BIG_ENDIAN"): ['-triple=mips'],
("CS_ARCH_MIPS", "CS_MODE_32+CS_MODE_MICRO"): ['-triple=mipsel', '-mattr=+micromips'], ("CS_ARCH_MIPS", "CS_MODE_32+CS_MODE_MICRO"): ['-triple=mipsel', '-mattr=+micromips'],
@ -106,7 +139,7 @@ def test_file(fname):
("CS_ARCH_PPC", "CS_MODE_BIG_ENDIAN"): ['-triple=powerpc64'], ("CS_ARCH_PPC", "CS_MODE_BIG_ENDIAN"): ['-triple=powerpc64'],
('CS_ARCH_SPARC', 'CS_MODE_BIG_ENDIAN'): ['-triple=sparc'], ('CS_ARCH_SPARC', 'CS_MODE_BIG_ENDIAN'): ['-triple=sparc'],
('CS_ARCH_SPARC', 'CS_MODE_BIG_ENDIAN+CS_MODE_V9'): ['-triple=sparcv9'], ('CS_ARCH_SPARC', 'CS_MODE_BIG_ENDIAN+CS_MODE_V9'): ['-triple=sparcv9'],
('CS_ARCH_SYSZ', '0'): ['-triple=s390x'], ('CS_ARCH_SYSZ', '0'): ['-triple=s390x', '-mcpu=z196'],
} }
#if not option in ('', 'None'): #if not option in ('', 'None'):
@ -120,15 +153,19 @@ def test_file(fname):
# tell llvm-mc to use Intel syntax # tell llvm-mc to use Intel syntax
mc_option = '-output-asm-variant=1' mc_option = '-output-asm-variant=1'
if arch == 'CS_ARCH_ARM': if arch == 'CS_ARCH_ARM' or arch == 'CS_ARCH_PPC' :
md.syntax = CS_OPT_SYNTAX_NOREGNAME md.syntax = CS_OPT_SYNTAX_NOREGNAME
if fname.endswith('3DNow.s.cs'):
md.syntax = CS_OPT_SYNTAX_ATT
for line in lines[1:]: for line in lines[1:]:
# ignore all the input lines having # in front. # ignore all the input lines having # in front.
if line.startswith('#'): if line.startswith('#'):
continue continue
#print("Check %s" %line) #print("Check %s" %line)
code = line.split(' = ')[0] code = line.split(' = ')[0]
asm = ''.join(line.split(' = ')[1:])
hex_code = code.replace('0x', '') hex_code = code.replace('0x', '')
hex_code = hex_code.replace(',', '') hex_code = hex_code.replace(',', '')
hex_data = hex_code.decode('hex') hex_data = hex_code.decode('hex')
@ -143,11 +180,63 @@ def test_file(fname):
else: else:
cs_output = 'FAILED to disassemble' cs_output = 'FAILED to disassemble'
mc_output = run_mc(archs[arch], code, mc_modes[(arch, mode)], mc_option) cs_output2 = normalize_hex(cs_output)
if (cs_output != mc_output): cs_output2 = cs_output2.replace(' ', '')
print("Mismatch: %s" %code)
print("\tMC = %s" %mc_output) if arch == 'CS_ARCH_MIPS':
print("\tCS = %s" %cs_output) # normalize register alias names
cs_output2 = cs_output2.replace('$at', '$1')
cs_output2 = cs_output2.replace('$v0', '$2')
cs_output2 = cs_output2.replace('$v1', '$3')
cs_output2 = cs_output2.replace('$a0', '$4')
cs_output2 = cs_output2.replace('$a1', '$5')
cs_output2 = cs_output2.replace('$a2', '$6')
cs_output2 = cs_output2.replace('$a3', '$7')
cs_output2 = cs_output2.replace('$t0', '$8')
cs_output2 = cs_output2.replace('$t1', '$9')
cs_output2 = cs_output2.replace('$t2', '$10')
cs_output2 = cs_output2.replace('$t3', '$11')
cs_output2 = cs_output2.replace('$t4', '$12')
cs_output2 = cs_output2.replace('$t5', '$13')
cs_output2 = cs_output2.replace('$t6', '$14')
cs_output2 = cs_output2.replace('$t7', '$15')
cs_output2 = cs_output2.replace('$t8', '$24')
cs_output2 = cs_output2.replace('$t9', '$25')
cs_output2 = cs_output2.replace('$s0', '$16')
cs_output2 = cs_output2.replace('$s1', '$17')
cs_output2 = cs_output2.replace('$s2', '$18')
cs_output2 = cs_output2.replace('$s3', '$19')
cs_output2 = cs_output2.replace('$s4', '$20')
cs_output2 = cs_output2.replace('$s5', '$21')
cs_output2 = cs_output2.replace('$s6', '$22')
cs_output2 = cs_output2.replace('$s7', '$23')
cs_output2 = cs_output2.replace('$k0', '$26')
cs_output2 = cs_output2.replace('$k1', '$27')
#print("Running MC ...")
if fname.endswith('thumb-fp-armv8.s.cs'):
mc_output = run_mc(archs[arch], code, ['-triple=thumbv8'], mc_option)
elif fname.endswith('mips64-alu-instructions.s.cs'):
mc_output = run_mc(archs[arch], code, ['-triple=mips64el', '-mcpu=mips64r2'], mc_option)
else:
mc_output = run_mc(archs[arch], code, mc_modes[(arch, mode)], mc_option)
mc_output2 = normalize_hex(mc_output)
if arch == 'CS_ARCH_MIPS':
mc_output2 = mc_output2.replace(' 0(', '(')
mc_output2 = mc_output2.replace(' ', '')
mc_output2 = mc_output2.replace('opaque', '')
if (cs_output2 != mc_output2):
asm = asm.replace(' ', '').strip().lower()
if asm != cs_output2:
print("Mismatch: %s" %line.strip())
print("\tMC = %s" %mc_output)
print("\tCS = %s" %cs_output)
if __name__ == '__main__': if __name__ == '__main__':