스택 프레임 그리기 예제 사다리 문서의 문제 4 풀이 코드다. if __name__ == "__main__": 아래가 자체 테스트다.
"""스택 프레임 그리기 예제 사다리의 답 확인. 실행: python3 44_stack-frame-ladder_p4.py
MASM이 없어 x86 32비트 명령 일부를 파이썬으로 흉내 낸다. 문서의 코드를 글자 그대로 넣어 돌린다.
문제 1~4의 명령별 ESP·EBP와 프레임의 주소별 값을 단언한다. 명령 주소는 가짜 값(00401000h + 4 × 순번)이다."""
import re
M32 = 0xFFFFFFFF
REG32 = ["eax", "ecx", "edx", "ebx", "esp", "ebp", "esi", "edi"]
REG8 = {"al": ("eax", 0), "ah": ("eax", 8), "bl": ("ebx", 0), "bh": ("ebx", 8),
"cl": ("ecx", 0), "ch": ("ecx", 8), "dl": ("edx", 0), "dh": ("edx", 8)}
REG16 = {"ax": "eax", "bx": "ebx", "cx": "ecx", "dx": "edx", "si": "esi", "di": "edi"}
class CPU:
"""x86 32비트 명령 일부만 흉내 낸다. 문서에 실린 코드를 글자 그대로 받아 돌린다."""
CODE_BASE = 0x00401000 # 명령 i번의 주소를 CODE_BASE + 4*i 로 친다(가짜 주소)
def __init__(self, source, esp=0x00001000, symbols=None):
self.r = {k: 0 for k in REG32}
self.r["esp"] = esp
self.mem = {} # 주소 -> 바이트
self.zf = self.cf = self.sf = 0
self.sym = dict(symbols or {}) # 변수 이름 -> 주소
self.prog, self.labels = [], {}
self.trace = []
self.halted = False
for raw in source.strip().splitlines():
line = raw.split(";")[0].strip()
if not line:
continue
m = re.match(r"^(\w+)\s+(PROC|ENDP)\b.*$", line)
if m:
if m.group(2) == "PROC":
self.labels[m.group(1)] = len(self.prog)
continue
m = re.match(r"^(\w+):\s*(.*)$", line)
if m:
self.labels[m.group(1)] = len(self.prog)
line = m.group(2).strip()
if not line:
continue
parts = line.split(None, 1)
op = parts[0].lower()
args = [a.strip() for a in parts[1].split(",")] if len(parts) > 1 else []
self.prog.append((op, args, raw.strip()))
self.eip = 0
# ---- 메모리 (리틀 엔디언) ----
def rd(self, addr, size=4):
return sum(self.mem.get((addr + i) & M32, 0) << (8 * i) for i in range(size))
def wr(self, addr, val, size=4):
for i in range(size):
self.mem[(addr + i) & M32] = (val >> (8 * i)) & 0xFF
def addr_of(self, i):
return self.CODE_BASE + 4 * i
# ---- 피연산자 ----
def _ea(self, expr):
"""[ebp+8], [ebp-4], [esi] 꼴의 주소를 계산한다."""
inner = expr.strip()[1:-1].replace(" ", "")
m = re.match(r"^(\w+)(?:([+-])(\w+))?$", inner)
base = self.r[m.group(1).lower()]
if m.group(2):
d = self._imm(m.group(3))
base = base + d if m.group(2) == "+" else base - d
return base & M32
@staticmethod
def _imm(tok):
t = tok.strip().lower()
if t.endswith("h"):
return int(t[:-1], 16)
if t.endswith("b") and set(t[:-1]) <= {"0", "1"}:
return int(t[:-1], 2)
return int(t)
def _parse(self, a):
"""(종류, 값, 크기)를 돌려준다. 종류: reg, reg8, mem, imm."""
size = 4
m = re.match(r"^(DWORD|WORD|BYTE)\s+PTR\s+(.*)$", a, re.I)
if m:
size = {"DWORD": 4, "WORD": 2, "BYTE": 1}[m.group(1).upper()]
a = m.group(2).strip()
low = a.lower()
if low in self.r:
return ("reg", low, 4)
if low in REG8:
return ("reg8", low, 1)
if low in REG16:
return ("reg16", low, 2)
if a.startswith("["):
return ("mem", self._ea(a), size)
if low.startswith("offset "):
return ("imm", self.sym[a.split()[1]], 4)
if a in self.sym:
return ("mem", self.sym[a], size)
return ("imm", self._imm(a) & M32, 4)
def get(self, a, size_hint=None):
kind, v, size = self._parse(a)
if kind == "mem" and size_hint:
size = size_hint
if kind == "reg":
return self.r[v]
if kind == "reg16":
return self.r[REG16[v]] & 0xFFFF
if kind == "reg8":
base, sh = REG8[v]
return (self.r[base] >> sh) & 0xFF
if kind == "mem":
return self.rd(v, size)
return v
def put(self, a, val, size_hint=None):
kind, v, size = self._parse(a)
if kind == "mem" and size_hint:
size = size_hint
if kind == "reg":
self.r[v] = val & M32
elif kind == "reg16":
self.r[REG16[v]] = (self.r[REG16[v]] & 0xFFFF0000) | (val & 0xFFFF)
elif kind == "reg8":
base, sh = REG8[v]
self.r[base] = (self.r[base] & ~(0xFF << sh) & M32) | ((val & 0xFF) << sh)
elif kind == "mem":
self.wr(v, val, size)
else:
raise ValueError("즉시값에는 쓸 수 없다: " + a)
# ---- 스택 ----
def push(self, val):
self.r["esp"] = (self.r["esp"] - 4) & M32
self.wr(self.r["esp"], val)
def pop(self):
v = self.rd(self.r["esp"])
self.r["esp"] = (self.r["esp"] + 4) & M32
return v
def _flags(self, res, carry):
self.zf = int(res & M32 == 0)
self.sf = (res >> 31) & 1
self.cf = int(carry)
# ---- 한 명령 실행 ----
def step(self):
op, a, raw = self.prog[self.eip]
nxt = self.eip + 1
if op == "mov":
# 한쪽이 8·16비트 레지스터면 메모리 피연산자도 그 크기로 읽고 쓴다
hint = None
for x in a:
k = self._parse(x)
if k[0] in ("reg8", "reg16"):
hint = k[2]
self.put(a[0], self.get(a[1], hint), hint)
elif op == "lea":
self.put(a[0], self._ea(a[1]))
elif op == "push":
self.push(self.get(a[0]))
elif op == "pop":
self.put(a[0], self.pop())
elif op in ("add", "sub", "cmp"):
x, y = self.get(a[0]), self.get(a[1])
res = x + y if op == "add" else x - y
self._flags(res, res > M32 if op == "add" else x < y)
if op != "cmp":
self.put(a[0], res)
elif op in ("inc", "dec"):
res = self.get(a[0]) + (1 if op == "inc" else -1)
cf = self.cf
self._flags(res, cf) # INC/DEC는 CF를 건드리지 않는다
self.put(a[0], res)
elif op == "mul":
prod = self.r["eax"] * self.get(a[0])
self.r["eax"], self.r["edx"] = prod & M32, (prod >> 32) & M32
self.cf = int(self.r["edx"] != 0)
elif op == "jmp":
nxt = self.labels[a[0]]
elif op in ("jz", "je"):
if self.zf:
nxt = self.labels[a[0]]
elif op in ("jnz", "jne"):
if not self.zf:
nxt = self.labels[a[0]]
elif op == "ja":
if not self.cf and not self.zf:
nxt = self.labels[a[0]]
elif op == "loop":
self.r["ecx"] = (self.r["ecx"] - 1) & M32
if self.r["ecx"]:
nxt = self.labels[a[0]]
elif op == "call":
self.push(self.addr_of(nxt))
nxt = self.labels[a[0]]
elif op == "ret":
target = self.pop()
if a:
self.r["esp"] = (self.r["esp"] + self._imm(a[0])) & M32
nxt = (target - self.CODE_BASE) // 4
elif op == "enter":
self.push(self.r["ebp"])
self.r["ebp"] = self.r["esp"]
self.r["esp"] = (self.r["esp"] - self._imm(a[0])) & M32
elif op == "leave":
self.r["esp"] = self.r["ebp"]
self.r["ebp"] = self.pop()
elif op == "pushad":
old = self.r["esp"]
for k in REG32:
self.push(old if k == "esp" else self.r[k])
elif op == "popad":
for k in reversed(REG32):
v = self.pop()
if k != "esp":
self.r[k] = v
elif op == "hlt":
self.halted = True
else:
raise ValueError("모르는 명령: " + raw)
self.trace.append((raw, dict(self.r), self.zf, self.cf))
self.eip = nxt
def run(self, start, limit=100000, watch=None):
self.eip = self.labels[start]
n = 0
while not self.halted:
self.step()
if watch:
watch(self)
n += 1
assert n < limit, "끝나지 않는다"
return self
P1 = """
main PROC
push 9
push 4
call Sub2
hlt
main ENDP
Sub2 PROC
push ebp
mov ebp,esp
mov eax,[ebp+12]
sub eax,[ebp+8]
pop ebp
ret 8
Sub2 ENDP
"""
P2 = """
main PROC
push 7
call P
hlt
main ENDP
P PROC
push ebp
mov ebp,esp
sub esp,8
mov DWORD PTR [ebp-4],1
mov eax,[ebp+8]
mov [ebp-8],eax
add eax,[ebp-4]
mov esp,ebp
pop ebp
ret 4
P ENDP
"""
P3 = """
main PROC
push 30
push 20
push 10
call R
hlt
main ENDP
R PROC
push ebp
mov ebp,esp
sub esp,4
push esi
push edi
mov esi,[ebp+8]
mov edi,[ebp+16]
mov [ebp-4],esi
add [ebp-4],edi
mov eax,[ebp-4]
pop edi
pop esi
mov esp,ebp
pop ebp
ret 12
R ENDP
"""
P4 = """
main PROC
push 2
push 8
call Q
add esp,8
hlt
main ENDP
Q PROC
enter 4,0
mov eax,[ebp+8]
add eax,[ebp+12]
mov [ebp-4],eax
add eax,[ebp-4]
leave
ret
Q ENDP
"""
def run(src, esp, ebp, **regs):
c = CPU(src, esp=esp)
c.r["ebp"] = ebp
c.r.update(regs)
c.run("main")
return c, [(raw, r["esp"], r["ebp"], r["eax"]) for raw, r, _, _ in c.trace]
if __name__ == "__main__":
# 문제 1
c, t = run(P1, 0x1000, 0x1010)
assert [x[1] for x in t[:9]] == [0xFFC, 0xFF8, 0xFF4, 0xFF0, 0xFF0, 0xFF0, 0xFF0, 0xFF4, 0x1000]
assert t[4][2] == 0xFF0 and c.r["eax"] == 5 and c.r["ebp"] == 0x1010
assert [c.rd(a) for a in (0xFFC, 0xFF8, 0xFF4, 0xFF0)] == [9, 4, c.addr_of(3), 0x1010]
# 문제 2
c, t = run(P2, 0x1000, 0x1010)
d = {x[0]: x for x in t}
assert d["sub esp,8"][1:3] == (0xFEC, 0xFF4)
assert [c.rd(a) for a in (0xFFC, 0xFF8, 0xFF4, 0xFF0, 0xFEC)] == [7, c.addr_of(2), 0x1010, 1, 7]
assert d["mov esp,ebp"][1] == 0xFF4 and d["pop ebp"][1:3] == (0xFF8, 0x1010) and d["ret 4"][1] == 0x1000
assert c.r["eax"] == 8
# 문제 3
c, t = run(P3, 0x2000, 0x2020, esi=0x5555, edi=0x6666)
d = {x[0]: x for x in t}
assert d["call R"][1] == 0x1FF0 and d["mov ebp,esp"][2] == 0x1FEC
assert d["sub esp,4"][1] == 0x1FE8 and d["push edi"][1] == 0x1FE0
words = [c.rd(a) for a in range(0x1FFC, 0x1FDC, -4)]
assert words == [30, 20, 10, c.addr_of(4), 0x2020, 40, 0x5555, 0x6666]
assert d["pop edi"][1] == 0x1FE4 and d["pop esi"][1] == 0x1FE8
assert d["mov esp,ebp"][1] == 0x1FEC and d["pop ebp"][1] == 0x1FF0 and d["ret 12"][1] == 0x2000
assert (c.r["eax"], c.r["esi"], c.r["edi"], c.r["ebp"]) == (40, 0x5555, 0x6666, 0x2020)
# 문제 4
c, t = run(P4, 0x0012FF80, 0x0012FF94)
d = {x[0]: x for x in t}
assert d["call Q"][1] == 0x12FF74
assert d["enter 4,0"][1:3] == (0x12FF6C, 0x12FF70)
assert [c.rd(a) for a in (0x12FF7C, 0x12FF78, 0x12FF74, 0x12FF70, 0x12FF6C)] == \
[2, 8, c.addr_of(3), 0x12FF94, 10]
assert d["leave"][1:3] == (0x12FF74, 0x12FF94)
assert d["ret"][1] == 0x12FF78 and d["add esp,8"][1] == 0x12FF80
assert c.r["eax"] == 20
print("ALL CHECKS PASSED")