#!/usr/bin/env python
from os import popen
from sys import argv, stderr, exit

dll = argv[1]
#objdump = "~/xda/wince-cross/usr/local/wince/cross-tools/bin/arm-wince-pe-objdump"
objdump = "~/src/oe/build/openmoko/cross/arm-linux/bin/objdump"

section_name   = []
section_size   = []
section_vma    = []
section_offset = []

imagebase = 0

#
# is this a branch with link instruction?
#

def isbl(inst):
	return (inst[0:2] == 'bl') and ((len(inst) == 2) or (inst[2] != 's'))

#
# is this a load or store instruction?
#

def isldrstr(inst):
	return inst[0:3] in ["ldr", "str"]

#
# extract the address from a string that optionally starts with '; '.
# the address has optional '0x' prefix and optional appended '<symbol_name>'
#

def extract_address(s):
	if s[:2] == '; ':
		start = 2
	else:
		start = 0
	if s[start:start+2] == '0x':
		start += 2
	end = s.find("<")
	if end >= 0:
		return int(s[start:end],16)
	else:
		return int(s[start:],16)

#
# character representation of a word
#

def asciichar(c):
	if c >= 32 and c < 128:
		return chr(c)
	else:
		return '.'

def fourcc(data):
	return "%c%c%c%c" % (asciichar(data & 0xff), asciichar((data>>8) & 0xff),
	                     asciichar((data>>16) & 0xff), asciichar((data>>24) & 0xff))

#
# is this an executable with headers?
#

ftype = popen("file %s" % dll).readlines()[0].split(' ')[1].replace("\n","")
if ftype != "data":

	#
	# parse section headers
	#

	for l in popen("%s -h %s | grep ' \.'" % (objdump, dll)).readlines():
		t = l.split()
		section_name   = section_name   + [t[1]]
		section_size   = section_size   + [int(t[2],16)]
		section_vma    = section_vma    + [int(t[3],16)]
		section_offset = section_offset + [int(t[5],16)]

else:

	section_name   = ["none"]
	section_size   = [256*1024]
	section_vma    = [0]
	section_offset = [0]

#
# virtual memory address <--> file offset translation
#

def vma_to_offset(vma):
	i=0
	while i < len(section_size):
		if (vma >= section_vma[i]) and (vma < section_vma[i]+section_size[i]):
			return vma-section_vma[i]+section_offset[i]
		i=i+1

def offset_to_vma(offset):
	i=0
	while i < len(section_size):
		if (offset >= section_offset[i]) and (offset < section_offset[i]+section_size[i]):
			return offset-section_offset[i]+section_vma[i]
		i=i+1

def vma_in_section(vma):
	i=0
	while i < len(section_size):
		if (vma >= section_vma[i]) and (vma < section_vma[i]+section_size[i]):
			return True
		i=i+1
	return False

#
# parse strings into a dictionary (0xADDR : "string")
#

strings = {}
strlen = {}

for l in popen("strings -tx %s" % dll).readlines():
	l = l.split(None,1)
	if len(l) != 2:
		continue
	addr = offset_to_vma(int(l[0],16))
	text = l[1].replace("\n","")
	strings[addr] = text
	strlen[addr] = (len(text)-1)/4+1

# unicode strings
for l in popen("strings -tx -el %s" % dll).readlines():
	l = l.split(None,1)
	if len(l) != 2:
		continue
	addr = offset_to_vma(int(l[0],16))
	text = l[1].replace("\n","")
	if not strings.has_key(addr):
		strings[addr] = text
		strlen[addr] = len(text)/2+1

#print "STRING TABLE"
#print "vma      offset   string"
#for k in strings.keys():
#	print '%8x %8x %s' % (k, vma_to_offset(k), strings[k])

#
# parse function table into a dictionary (0xADDR : "name")
#

pipe = popen("%s -T %s | grep '^[0-9a-f]\{8\}'" % (objdump, dll))

ll=pipe.readlines()
pipe.close()

ftable = {}

for l in ll:
	addr = int(l[:8],16)
	ftable[addr] = l[44:-1].strip()

#
# custom function information
#

custom_ftable = []
custom_ftable = file(dll+".ftable").readlines()
for l in custom_ftable:
	if l[0] == '#':
		continue
	t = l.split('\t')
	if len(t) < 2:
		continue
	ftable[int(t[0],16)] = t[1]

ftable[0x31ef0] = "DIS31ef0"

#print "FUNCTION TABLE"
#for k in ftable.keys():
#	print "%8x %s" % (k, ftable[k])

#
# disassemble
#

stderr.write("Disassembling...\n")

if ftype == "data":
	pipe = popen("%s -b binary -m arm -D %s" % (objdump, dll))
else:
	pipe = popen("%s -D %s" % (objdump, dll))
	#pipe = popen("%s -d %s" % (objdump, dll))
ll=pipe.readlines()
pipe.close()

datadict   = {}
referenced = {}
jumped_to  = {}

stderr.write("Finding references...\n")

i = 0
while i+1 < len(ll):
	i=i+1
	l=ll[i]
	if (l == '\t...') or (l == '\n'):
		continue
	t = l.split('\t')
	if (len(t[0]) == 0) or (t[0][-1] != ':') or (len(t) < 3):
		continue

	addr = int(t[0][0:-1].strip(),16)
	data = int(t[1],16)
	datadict[addr] = data

	# only evaluate references
	inst = t[2][0:3]

	if isldrstr(inst):
		if t[len(t)-1][:2] == '; ':
			ref = extract_address(t[len(t)-1])
			if vma_in_section(ref):
				referenced[ref] = 0

	if isbl(inst):
		target = extract_address(t[3])
		if vma_in_section(target):
			if target in jumped_to.keys():
				jumped_to[target].append("0x%08x" % addr)
			else:
				jumped_to[target] = ["0x%08x" % addr]

#
# resolve references
#

stderr.write("Resolving references...\n");

for ref in referenced.keys():
	if datadict.has_key(ref):
		referenced[ref] = datadict[ref]
		if datadict.has_key(referenced[ref]):
			referenced[referenced[ref]] = datadict[referenced[ref]]
	else:
		referenced[ref] = 0

#
# go! go! go!
#

stderr.write("Generating disassembled output...\n")

armreg = { 'r0': 0, 'r1': 1, 'r2': 2, 'r3': 3, 'r4': 4, 'r5': 5, 'r6': 6, 'r7': 7,
           'r8': 8, 'r9': 9, 'sl': 10, 'fp': 11, 'ip': 12, 'sp': 13, 'lr': 14, 'pc': 15 }
r = [0, 0, 0, 0, 0, 0, 0, 0,
     0, 0, 0, 0, 0, 0, 0, 0]

section = None

i = 0
while i+1 < len(ll):
	i=i+1
	l=ll[i].strip('\n')	
	if (l == '\t...') or (l == '\n'):
		#print l
		continue
	t = l.split('\t')
	if (len(t[0]) == 0) or (t[0][-1] != ':') or (len(t) < 3):
		if t[0][:23] == "Disassembly of section ":
			section = t[0][23:-1]
			print "Section %s" % section
		continue
	addr = int(t[0][0:-1].strip(),16)
	data = int(t[1],16)
	inst = t[2]

	section_list = [".text", ".rodata", ".data.rel.ro", ".data", ".bss"]
	
	if not (section in section_list):
		continue

	#
	# print function header comment
	#

	if ftable.has_key(addr):
		print "\n/*\n * %s\n */" % ftable[addr]
		if jumped_to.has_key(addr):
			print "/*\n * %s\n */" % jumped_to[addr]

	#
	# interpret some instructions
	#

	r[15] = addr+8
	intstr = ""

	if inst == "ldr":
		a = t[3][:2]
		b = t[3][5:7]
		if armreg.has_key(a) and armreg.has_key(b):
			if t[3][4:10] == "[pc, #":
				target_addr = extract_address(t[4])
				if datadict.has_key(target_addr):
					r[armreg[a]] = datadict[target_addr]
					intstr = "%s = %x" % (a, r[armreg[a]])
				else:
					r[armreg[a]] = 0xbaadbaad
			else:
				c = t[3][9:]
				if t[3][7] == ']':
					target_addr = r[armreg[b]]
				elif c[0] == '#':
					if c[-1] == '!':
						target_addr = r[armreg[b]] + int(c[1:-2])
						r[armreg[b]] += int(c[1:-2])
					else:
						target_addr = r[armreg[b]] + int(c[1:-1])
				elif armreg.has_key(c[:-1]):
					target_addr = r[armreg[b]] + r[armreg[c[:-1]]]
				else:
					target_addr = 0xbaadbaad
				if datadict.has_key(target_addr):
					r[armreg[a]] = datadict[target_addr]
					intstr = "%s = %x (<-- %x)" % (a, r[armreg[a]], target_addr)
				else:
					intstr = "ERR %x" % target_addr
					r[armreg[a]] = 0xbaadbaad
		else:
			intstr = "FIXME"
			if armreg.has_key(a):
				r[armreg[a]] = 0xbaadbaad
	if inst == "mov":
		a = t[3][:2]
		if armreg.has_key(a):
			b = t[3][4:]
			if b[0] == '#':
				r[armreg[a]] = int(b[1:])
				intstr = "%s = %x" % (a, r[armreg[a]])
			elif armreg.has_key(b):
				r[armreg[a]] = r[armreg[b]]
				intstr = "%s = %x" % (a, r[armreg[a]])
			else:
				intstr = "FIXME"
				r[armreg[a]] = 0xbaadbaad
		else:
			intstr = "ERR"
	if inst == "add":
		a = t[3][:2]
		b = t[3][4:6]
		if armreg.has_key(a) and armreg.has_key(b):
			c = t[3][8:]
			if c[0] == '#':
				r[armreg[a]] = (r[armreg[b]] + int(c[1:])) & 0xffffffff
				intstr = "%s = %x" % (a, r[armreg[a]])
			elif armreg.has_key(c):
				r[armreg[a]] = (r[armreg[b]] + r[armreg[c]]) & 0xffffffff
				intstr = "%s = %x" % (a, r[armreg[a]])
			else:
				intstr = "FIXME"
				r[armreg[a]] = 0xbaadbaad
		else:
			intstr = "FIXME"
			if armreg.has_key(a):
				r[armreg[a]] = 0xbaadbaad

	if len(intstr) < 8:
		intstr = intstr + "\t\t"
	elif len(intstr) < 16:
		intstr = intstr + "\t"

	print intstr,

	#
	# print line
	#

	if strings.has_key(addr):
		print "%8x:\t\"%s\"" % (addr, strings[addr])
		if(strlen[addr]>1):
			i += strlen[addr]-1
			l = ll[i+1]
			t = l.split('\t')
			if len(t)>2:
				next = int(ll[i+1].strip('\n').split('\t')[1],16)
				if next in [0x0, 0xa]:
					i += 1
	elif referenced.has_key(addr):
		if strings.has_key(data):
			print "%8x:\t\"%s\"" % (addr, strings[data])
			if(strlen[data]>1):
				i += strlen[data]-1
				l = ll[i+1]
				t = l.split('\t')
				if len(t)>2:
					next = int(ll[i+1].strip('\n').split('\t')[1],16)
					if next in [0x0, 0xa]:
						i += 1
		elif referenced.has_key(data):
			if datadict.has_key(referenced[data]):
				if strings.has_key(referenced[data]):
					print "%8x:\t%08x --> \"%s\"" % (addr, data, strings[referenced[data]])
				else:
					print "%8x:\t%08x --> %08x" % (addr, data, datadict[referenced[data]])
			else:
				print "%8x:\t%08x --> ? (at %08x)" % (addr, data, referenced[data])
		else:
			if len(t) > 3:
				print "%8x:\t%08x %s\t%s\t%s\t; (ref.d somewhere)" % (addr, data, fourcc(data), t[2], t[3])
			else:
				print "%8x:\t%08x %s\t%s\t; (ref.d somewhere)" % (addr, data, fourcc(data), t[2])
	elif t[len(t)-1][0:4] == '; 0x':
		ref = int(t[len(t)-1][4:],16)
		if referenced.has_key(ref) and ref != 0:
			if datadict.has_key(referenced[ref]):
				if strings.has_key(referenced[ref]):
					print "%s = 0x%08x --> \"%s\"" % (l, referenced[ref], strings[referenced[ref]])
				else:
					print "%s = 0x%08x --> %08x" % (l, referenced[ref], datadict[referenced[ref]])
			else:
				print "%s --> 0x%08x" % (l, referenced[ref])
		else:
			print l
	elif addr in jumped_to.keys():
		print "\n/*\n *",
		print jumped_to[addr],
		print "\n */\n%s" % l
	elif isldrstr(inst):
		if t[len(t)-1][:2] == '; ':
			ref = extract_address(t[len(t)-1])
			if referenced.has_key(ref):
				if datadict.has_key(referenced[ref]):
					if strings.has_key(referenced[ref]):
						print "%s = 0x%08x --> \"%s\"" % (l, referenced[ref], strings[referenced[ref]])
					else:
						print "%s = 0x%08x --> %08x" % (l, referenced[ref], datadict[referenced[ref]])
				else:
					print "%s --> 0x%08x" % (l, referenced[ref])
			else:
				print "INTERNAL ERROR, REFERENCE SHOULD EXIST"
				exit(1)
		else:
			print l
	elif isbl(inst):
		target = extract_address(t[3])
		if ftable.has_key(target):
			pos = l.rfind("<")
			if pos >= 0:
				print l[:pos],
			else:
				print l,
			print "<%s>" % ftable[target],
			if ftable[target] in ["printf", "puts", "fwrite"]:
				print "r0 = 0x%x" % r[0],
				if strings.has_key(r[0]):
					print "--> \"%s\"" % strings[r[0]],
			elif ftable[target] in ["sprintf, vsprintf", "fprintf"]:
				print "r1 = 0x%x" % r[1],
				if strings.has_key(r[1]):
					print "--> \"%s\"" % strings[r[1]],
			elif ftable[target] in ["putchar", "fputc"]:
				print "r0 = 0x%x" % r[0]
			elif ftable[target] == "fread":
				print "r0 = 0x%x" % r[0]
			elif ftable[target] == "DIS31ef0":
				if strings.has_key(r[1]):
					print "r0 = 0x%x, r1 = \"%s\"" % (r[0], strings[r[1]])
				else:
					print "r0 = 0x%x, r1 = 0x%x" % (r[0], r[1])
			print
		else:
			print l
	else:
		if strings.has_key(data):
			print "%s\t; --> \"%s\"" % (l, strings[data])
		elif ftable.has_key(data):
			print "%s\t; --> <%s>" % (l, ftable[data])
		else:
			print "%8x:\t%08x %s\t%s" % (addr, data, fourcc(data), '\t'.join(t[2:]))
		continue