湖湘杯2019两个密码题wp
还是自己太菜的原因,这次湖湘杯只做出来4道题,然后5点的时候就放弃了去跟同学出去玩了,当时感觉进前50无望(这次湖湘杯py情况也很严重啊,可惜烽火台只报不封,挺恶心的)。不过无论如何,这次比赛还是有收获的,总结沉淀一下这两道密码学题目吧:
<!-- more -->
Oracle padding attack
忘了这个题目是啥了,但是攻击原理就是针对CBC模式的特殊攻击方式,Oracle padding attack。详细原理可以从下面获取(介绍的很详细,强烈推荐):
https://www.freebuf.com/articles/database/151167.html
下面简单介绍一下利用原理:
Oracle padding attack的攻击条件还是比较苛刻的:
- 在给服务器返回的数据中,iv可控。(iv是对称加密中的偏移向量,不清楚的同学可以详细了解一下CBC等加密模式的原理。)
- 在响应数据中,攻击者可以通过报错等方式得知padding是否出错。
- 服务器采用对称加密算法,使用CBC模式和PCK#5填充法。(这个填充法就是在最后一组长度不足分组长度的时候,填充剩余多少字节个数,例如16字节分组中最终还剩6个,那就填充6个ascii的6)
- 初始的iv已知。
在服务器解密后,被解密的明文会和iv进行异或(这个是对称加密算法中CBC加密模式的基本原理,这里不懂建议回去复习一下,方便理解),之后会进行去填充处理,一般算法库对这种去填充的实现是:
- 检查最后一个字节的ascii值,这里为了方便解释我们假设是6。
- 检查后6个字节是不是6。
- 如果是6,那么去掉这6个6,否则就报错或提示信息。
这里注意:上述过程中检查的是被我们解密后,再和初始向量异或后的明文串。如果向服务器返回的iv可控,那么我们就可以通过不断改变iv的方法来通过报错达到被解密的数据串可控的目的,同时也可以任意恢复原文!我们假设分组长度为16,简述这个过程如下:
- 控制初始向量,前15个字符串任意,我们控制第16个字符串,从1爆破到255,不报错的那个结果肯定是使密钥解密出的明文串异或后为1。这是我们如果异或真正的初始向量,再异或1就恢复出了明文。这个明文再异或初始向量和我们期望解密的结果,那么这个串解密的结果就可控了。举个例子:假设我们爆破出的最后一个字符是0x20,原本初始向量的值是0x41,那么原本的明文串的最后一个字符就是:0x20 ^ 0x41 ^ 0x01=0x60。假设我们希望服务器解密这个串时最后一个字符是0x02,那么我们就可以调整初始向量为:0x60 ^ 0x41 ^ 0x02=0x23。
- 根据第一步,我们已经爆破出了分组中的最后一个明文,先控制最后一个初始向量,使其经解密后结果为0x02,然后我们再控制倒数第二个初始向量,从1爆破到255,那么不报错的那个结果肯定是使密码解密处的明文串异或后为2。这时我们以此类推,爆破倒数第三个。
原理就是这些,这道题使用脚本攻击如下:
#enconding:utf-8
'''
@Author: b0ring
@MySite: https://unnamebao.github.io/
@Date: 2019-11-10 12:02:53
@Version: 1.0.0
'''
import socket
import base64
import codecs
def num2str(i):
return ("%02x"%i).encode()
def postfix_str(postfix,iv,index):
res_ = b""
for i in range(int(len(postfix)/2)):
res_ = num2str(int(postfix[len(postfix) - (i+1)*2:len(postfix)-i*2],16)^int(iv[32-(i+1)*2:32-i*2],16)^index) + res_
return res_
def brute_one(IV,encryp_text,client,index,postfix):
for i in range(256):
if index == 0:
if i == IV[-1]:
continue
if i == 0:
print(" append:",postfix_str(postfix,IV,index))
iv = b"41"*(16-index) + num2str(i) + postfix_str(postfix,IV,index)
data_to_send = iv + encryp_text + b'\n'
client.send(data_to_send)
data_respond = client.recv(1024)
# print(i,int(iv[32-index*2:32-index*2+2],16)^int(IV[32-index*2:32-index*2+2],16),data_respond)
if b"padding error" in data_respond:
continue
else:
return num2str(i^int(IV[32-index*2:32-index*2+2],16)^index)
if __name__ == "__main__":
client = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
client.connect(('183.129.189.62', 13706))
data = client.recv(1024)
print(data)
c = data.decode().split("\n")[0].replace("Hey, new! Your passport is ","")[:64]
IV= c[:32].encode()
encryp_text = c[32:].encode()
print(" IV:",IV)
print(type(IV))
postfix = b""
for index in range(0,16):
one = brute_one(IV,encryp_text,client,index+1,postfix)
postfix = one + postfix
print(" postfix:",postfix)
print('-----------------------------')
payload = b""
admin = b"Admin"
for i in range(5):
payload += num2str(admin[i] ^ int(postfix[i*2:i*2+2],16) ^ int(IV[i*2:i*2+2],16))
for i in range(5,16):
payload += num2str(11 ^ int(postfix[i*2:i*2+2],16) ^ int(IV[i*2:i*2+2],16))
print(type(payload))
print(payload)
print(len(payload))
payload += encryp_text + b'\n'
client.send(payload)
print(client.recv(1024))
Easy RSA
首先看一下题目吧:
from Crypto.Util.number import *
import libnum
import gmpy2
flag = open("flag.txt","rb").read()
m=libnum.s2n(flag)
p=getPrime(1024)
q=getPrime(1024)
n=p*q
e=65537
c=pow(m,e,n)
phi=(p-1)*(q-1)
d=gmpy2.invert(e,phi)
dp=d%(p-1)
print dp,n,e,c
#84373069210173690047629226878686144017052129353931011112880892379361035492516066159394115482289291025932915787077633999791002846189004408043685986856359812230222233165493645074459765748901898518115384084258143483508823079115319711227124403284267559950883054402576935436305927705016459382628196407373896831725 22000596569856085362623019573995240143720890380678581299411213688857584612953014122879995808816872221032805734151343458921719334360194024890377075521680399678533655114261000716106870610083356478621445541840124447459943322577740268407217950081217130055057926816065068275999620502766866379465521042298370686053823448099778572878765782711260673185703889168702746195779250373642505375725925213796848495518878490786035363094086520257020021547827073768598600151928787434153003675096254792245014217044607440890694190989162318846104385311646123343795149489946251221774030484424581846841141819601874562109228016707364220840611 65537 14874271064669918581178066047207495551570421575260298116038863877424499500626920855863261194264169850678206604144314318171829367575688726593323863145664241189167820996601561389159819873734368810449011761054668595565217970516125181240869998009561140277444653698278073509852288720276008438965069627886972839146199102497874818473454932012374251932864118784065064885987416408142362577322906063320726241313252172382519793691513360909796645028353257317044086708114163313328952830378067342164675055195428728335222242094290731292113709866489975077052604333805889421889967835433026770417624703011718120347415460385182429795735
我们从中提取一些信息(由于52上使用markdown生成数学公式有问题,建议出现阅读问题的朋友来我的博客阅读本文章:https://blog.b0ring.cf/#/posts/%E6%B9%96%E6%B9%98%E6%9D%AF2019%E4%B8%A4%E4%B8%AA%E5%AF%86%E7%A0%81%E9%A2%98wp):
m = flag # m中的值是flag的明文串
n = p q # 这个很清晰 学过RSA应该理解
e = 65537 # 提醒下,65537是个质数,虽然好像是不是质数无所谓
c = pow(m,e,n) # 这里就是典型的RSA加密了,如果已知d,可以通过 m = pow(c,d,n)解出明文,其中,(e,n)是公钥,d是私钥。
phi = (p-1)(q-1) # 这里计算了n的欧拉函数,解释一下,欧拉函数是小于一个数有多少与其互质的数,质数的欧拉函数就是自身减一,而两个质数相乘的数,它的欧拉函数就是两个质数相加减一相乘
d = gmpy2.invert(e,phi) # 这里就是求同余phi下与e互质的数,就是私钥d
dp = d % (p-1) # 就是d对(p-1)求余
梳理一下:
e,c,n已知,dp = d % (p-1)已知,求m
这个地方涉及到一个原理:
对于任意因子a、b(不需要是质数),$x=a*b$,如果$c \equiv m \mod x$,那么 $c \equiv m \mod a$和$c \equiv m \mod b$ 都成立。
举个例子来理解: $16 \equiv 1 \mod 15$
那么 $16 \equiv 1 \mod 3$,$16 \equiv 1 \mod 5$
而我们知道,d和e在同余phi下是互逆的,所以:
$d*e \equiv 1 \mod phi$
$d*e \equiv 1 \mod (p-1)*(q-1)$
$d*e \equiv 1 \mod (p-1)$
$dp*e \equiv 1 \mod (p-1)$
因此,$(dp*e - 1)$一定是(p-1)的倍数,且p是素数。而且,我们又知道,$dp < (p-1)$,因此遍历1~65537,一定可以找到一个数,使$(dp*e - 1) \equiv 0 \mod (p-1)$
使用脚本如下:
#enconding:utf-8
'''
@Author: b0ring
@MySite: https://unnamebao.github.io/
@Date: 2019-11-10 12:36:18
@Version: 1.0.0
'''
import Crypto.Util.number as number
def fastpow(Co, CoCo, CoCoCo):
CoCoCoCo = 1
while CoCo != 0:
if (CoCo & 1) == 1:
CoCoCoCo = (CoCoCoCo * Co) % CoCoCo
CoCo >>= 1
Co = (Co * Co) % CoCoCo
return CoCoCoCo
def gcd(num_1,num_2):
p,q=max(num_1,num_2),min(num_1,num_2)
if q == 0:
return p
r = p%q
return gcd(q,r)
def EX_GCD(a,b,arr): #扩展欧几里得
if b == 0:
arr[0] = 1
arr[1] = 0
return a
g = EX_GCD(b, a % b, arr)
t = arr[0]
arr[0] = arr[1]
arr[1] = t - int(a // b) * arr[1]
return g
def ModReverse(a,n):
arr = [0,1,]
gcd = EX_GCD(a,n,arr)
if gcd == 1:
return (arr[0] % n + n) % n
else:
return -1
dp = 84373069210173690047629226878686144017052129353931011112880892379361035492516066159394115482289291025932915787077633999791002846189004408043685986856359812230222233165493645074459765748901898518115384084258143483508823079115319711227124403284267559950883054402576935436305927705016459382628196407373896831725
n = 22000596569856085362623019573995240143720890380678581299411213688857584612953014122879995808816872221032805734151343458921719334360194024890377075521680399678533655114261000716106870610083356478621445541840124447459943322577740268407217950081217130055057926816065068275999620502766866379465521042298370686053823448099778572878765782711260673185703889168702746195779250373642505375725925213796848495518878490786035363094086520257020021547827073768598600151928787434153003675096254792245014217044607440890694190989162318846104385311646123343795149489946251221774030484424581846841141819601874562109228016707364220840611
e = 65537
c = 14874271064669918581178066047207495551570421575260298116038863877424499500626920855863261194264169850678206604144314318171829367575688726593323863145664241189167820996601561389159819873734368810449011761054668595565217970516125181240869998009561140277444653698278073509852288720276008438965069627886972839146199102497874818473454932012374251932864118784065064885987416408142362577322906063320726241313252172382519793691513360909796645028353257317044086708114163313328952830378067342164675055195428728335222242094290731292113709866489975077052604333805889421889967835433026770417624703011718120347415460385182429795735
# tmp1 = (dp * e) -1
for i in range(65537)[::-1]:
if (dp*e-1)%i == 0 and number.isPrime((dp*e-1)//i + 1):
p = (dp*e-1)//i + 1
break
q = n // p
phi = (p-1)*(q-1)
d = ModReverse(e,phi)
m = fastpow(c,d,n)
print(number.long_to_bytes(m))
当然了,这里可以这样用是因为e比较小,那么如果e很大怎么办呢?其实我们已知e和dp在同余(p-1)下是互逆的,那么:
对于任意一个数r,$r^{e*dp} \equiv r \mod p$
为什么呢?因为根据欧拉定理,我们有:
$r^{ \phi(p) } \equiv 1 \mod p$
又因为p是素数,$\phi(p) = p-1$
$e * dp \equiv 1 \mod (p-1)$
$r^{e*dp} = r^{n*\phi(p)+1} \equiv 1*r \mod p$
因此
$r^{dp*e} - r = n*p$
也就是说,对于任意的r,我们可以利用$r^{n*dp} -r$来找到p的一个倍数,再使用扩展欧几里得算法去求其与n的最大公因数,就可以算出p了,脚本具体步骤如下:
#enconding:utf-8
'''
@Author: b0ring
@MySite: https://unnamebao.github.io/
@Date: 2019-11-10 12:36:18
@Version: 1.0.0
'''
import Crypto.Util.number as number
def fastpow(Co, CoCo, CoCoCo):
CoCoCoCo = 1
while CoCo != 0:
if (CoCo & 1) == 1:
CoCoCoCo = (CoCoCoCo * Co) % CoCoCo
CoCo >>= 1
Co = (Co * Co) % CoCoCo
return CoCoCoCo
def gcd(num_1,num_2):
p,q=max(num_1,num_2),min(num_1,num_2)
if q == 0:
return p
r = p%q
return gcd(q,r)
def EX_GCD(a,b,arr): #扩展欧几里得
if b == 0:
arr[0] = 1
arr[1] = 0
return a
g = EX_GCD(b, a % b, arr)
t = arr[0]
arr[0] = arr[1]
arr[1] = t - int(a // b) * arr[1]
return g
def ModReverse(a,n):
arr = [0,1,]
gcd = EX_GCD(a,n,arr)
if gcd == 1:
return (arr[0] % n + n) % n
else:
return -1
dp = 84373069210173690047629226878686144017052129353931011112880892379361035492516066159394115482289291025932915787077633999791002846189004408043685986856359812230222233165493645074459765748901898518115384084258143483508823079115319711227124403284267559950883054402576935436305927705016459382628196407373896831725
n = 22000596569856085362623019573995240143720890380678581299411213688857584612953014122879995808816872221032805734151343458921719334360194024890377075521680399678533655114261000716106870610083356478621445541840124447459943322577740268407217950081217130055057926816065068275999620502766866379465521042298370686053823448099778572878765782711260673185703889168702746195779250373642505375725925213796848495518878490786035363094086520257020021547827073768598600151928787434153003675096254792245014217044607440890694190989162318846104385311646123343795149489946251221774030484424581846841141819601874562109228016707364220840611
e = 65537
c = 14874271064669918581178066047207495551570421575260298116038863877424499500626920855863261194264169850678206604144314318171829367575688726593323863145664241189167820996601561389159819873734368810449011761054668595565217970516125181240869998009561140277444653698278073509852288720276008438965069627886972839146199102497874818473454932012374251932864118784065064885987416408142362577322906063320726241313252172382519793691513360909796645028353257317044086708114163313328952830378067342164675055195428728335222242094290731292113709866489975077052604333805889421889967835433026770417624703011718120347415460385182429795735
mp = fastpow(e,e*dp,n)-e
p = gcd(mp,n)
q = n // p
phi = (p-1)*(q-1)
d = ModReverse(e,phi)
m = fastpow(c,d,n)
print(number.long_to_bytes(m))