1
0
mirror of https://github.com/mbirth/wiki.git synced 2024-12-27 23:24:06 +00:00
wiki.mbirth.de/know-how/hacking/_posts/2016-07-07-crypto-virus-decrypted.md

171 lines
7.0 KiB
Markdown
Raw Normal View History

2016-07-08 00:24:31 +01:00
---
title: Crypto-Virus decrypted
language: en
layout: default
2016-07-08 14:42:47 +01:00
created: 2016-07-08 15:41:41 +0200
updated: 2016-07-08 15:41:41 +0200
2016-07-08 00:24:31 +01:00
toc: false
tags:
- know-how
- hacking
- virus
- cryptovirus
- cryptor
---
A friend received the following mail from (supposedly) **FedEx International**:
![]({{ site.url }}/assets/cryptovirusmail.png)
The attached zip file contained a file `000794681.doc.js`. Since `.js` is a
known file type in Windows, it would show up as `000794681.doc` and you'd think
it's a Word file. But clicking on it will run the JavaScript using the *Windows
Scripting Host*.
The entry point
===============
The JavaScript looks like this ([GIST](https://gist.github.com/mars3142/3fc6a5522fcb752cdcbde3a5c1bca434)):
![]({{ site.url }}/assets/codedocjs.png)
The various `g55(number, string)` calls do nothing but sorting the string into
the number's position in an array. In the last but one line, the snippets in
the array are joined in the numeric order which turns out to be valid
JavaScript (or JScript) code again. In the last line, that code is executed.
If you omit the last line and, instead, output the contents of `h72`, you'll
get the unobfuscated code of the first stage: [GIST](https://gist.github.com/mbirth/d21bf52a024d0634f731e90dca94d254).
What this does is basically:
* Download 5 files (incl. backup servers if one is taken down) into `%TEMP%`
* `a1.exe` (some Visual Basic program)
* `a2.exe` (some NSIS installer, unpacks files to `AppData` then waits)
* `a.exe` (php.exe, PHP runtime)
* `php4ts.dll` (PHP library)
* `a.php` (the actual encrypting PHP code)
* If all 3 PHP files are downloaded:
* Write message into `a.txt`
* Register autostart to open `a.txt` after login
* Register extension `.crypted` to open/display `a.txt` whenever you click on
a crypted file
* Copy `a.txt` to your Desktop as `DECRYPT.txt`
* Run the actual encryption of your files (see next chapter)
* Display `a.txt`
2016-07-08 00:27:50 +01:00
* Overwrite the `a.php` (containing the encryption key) with the BitCoin
address (so it can't be undeleted to get the actual key)
2016-07-08 00:24:31 +01:00
* Delete `a.exe`, `php4ts.dll` and, of course, the `a.php`
The files to download are selected by 3 parameters:
2016-07-08 14:37:47 +01:00
* `ad`, the BitCoin address you should send money to
2016-07-08 00:24:31 +01:00
* `id`, some identifier?, can be omitted
* `rnd`, file selector
The `ad` parameter is used to generate your encryption key (see next chapter).
The `rnd` parameter selects the file you download. It's made up of the current
mirror server's number (0..4) and the file to download (1..5). So it starts
with `01`, `02`, etc. and if that server doesn't answer, it'll continue with
`11`, `12` .. `15`. The files you get with `01`, `11`, `21`, etc. are the same.
However, there seem to be some variation with the files returned for `*1`. I'm
not sure if it's time based or randomly selected each time you start a
download. When downloading from different servers in a short time, you'll
mostly get the same file. In rare occasions, one (the last file downloaded)
was different. I've found 5 different variations which differ in length and
bytes starting from position 0x43cb6. Could also be random data to confuse
antivirus products.
I think the files `a1.exe` and `a2.exe` are there to either confuse some
antivirus products or just filler material for future use.
What's also interesting is, that the files are returned with a MIME type of
`image/png`. Also, the download only works when the user agent string
contains `Windows NT`, otherwise you'll get an empty (0 bytes) response.
Encryption code
===============
The downloaded PHP script (`a.php`) seems to be freshly obfuscated each time
you download it. But while the obfuscated version differs every time, the
deobfuscated code is the same.
2016-07-08 14:37:47 +01:00
This is how it looks after download:
![]({{ site.url }}/assets/codeaphp.png)
After removing the first layer of obfuscation, it turns into this:
![]({{ site.url }}/assets/codeaphp2.png)
And now it becomes clearer, what's happening here. In `$h772` we have an
`eval()` command with the code to *evaluate* being encoded in base64.
In `$e51` there's a regular expression with the `/e` modifier, which enabled
callbacks in older PHP versions. This means, whenever the parser finds a
match, a given piece of code is called. And as you might have guessed, that
code is the one in `$h772`. And to have the parser find something, the subject
in the `preg_replace` call is exactly the same as the search string in `$e51`.
So to get to the actual code, we just have to `base64_decode()` the string
without `eval()`ing it. And this brings us to the final code. You can see all
iterations in this [GIST](https://gist.github.com/mbirth/11979a35a152478427928dbdf593797b).
(I've annotated and reformatted the final stage.)
Fun fact: The base64 encoded code contains proper indentation and Windows line
endings (CR+LF instead of LF only). They could've saved a lot of space by
removing all unneccessary spaces and line breaks.
The code does this:
* Try all drives from `C:` to `Z:`
* For each drive, check the contents of the root directory
* If it's a folder and it's not `winnt`, `windows`, `appdata`, etc., check that
folder (recursively)
* If it's a file, check if the extension is `zip`, `rar`, `doc`, etc. and if so,
encrypt it and add `.crypted` to the filename
Now the interesting things are right at the beginning of the `Tree($p)` method.
There are 3 variables defined:
* `$a` which defines the intended action, `e` is for encryption, `d` for
decryption
* `$k` is the key, which is decoded from base64
* `$s` is a backslash character to not have to mess around with proper escaping
I noticed that with different `ad` values in the download URL (see previous
chapter), you'll get different encryption keys in `$k`. `ad` is the BitCoin
address you are asked to send money to. This means the encryption key is based
on the BitCoin address.
2016-07-08 00:24:31 +01:00
E.g., an `ad` value of `17DmGrhMXJcvsmj9tihgTRGAhACynuBmSo` returns a PHP
script with the key:
MWKTbqXczBBUtCGOY6rxrB6Q2ECoaLUCGHDI5C54QaQHiP5010q99mPQNqAKkMkCtCicYss0uCCIDHPa5DiMDF6wYajvGFmaKJD4mtscEVSXPLUuduRStiug/kCCoA16swZZvi2c
If you change `ad` to `27DmGrhMXJcvsmj9tihgTRGAhACynuBmSo`, the key changes to:
MmSWbqXczBBUtCGOY6rxrB6Q2ECoaLUCGHDI5C54QaQHiP5010q99mPQNqAKkMkCtCicYss0uCCIDHPa5DiMDF6wYajvGFmaKJD4mtscEVSXPLUuduRStiug/kCCoA16swZZvi2c
^^^
2016-07-08 14:37:47 +01:00
In other words: This script is generic and can be used for different BitCoin
accounts which will generate different encryption keys. And you need to wire
your money to the correct one to get the correct decryption key.
Decrypting encrypted files
==========================
So let's imagine you accidentally run the script and all your important files
are now encrypted and renamed to `important.doc.crypted`. What can you do?
Well, we've learned that the encryption key is based on the `ad` value. And that
you still have in that mail with the bogus zip file. Using that, we can download
the PHP script again, which contains the actual key.
2016-07-08 00:24:31 +01:00
2016-07-08 14:37:47 +01:00
After deobfuscating it, you just have to change `$a='e'` to `$a='d'` and run it
and it should decrypt all your files again. Problem solved.