Investigation into the state of NIM malware Part 2
By: Jason Reaves
We did an investigation in the usage of NIM malware earlier in 2021[1] after the discovery of NimRod/Nimza[2,5] and RustyBuer[3,4] which are both being leveraged by actors associated with the TrickBot crew. This is just a continuation of that earlier report or a dump of data from continuing to track NIM based malware over time.
NimGrabber
MD5: 8753b73e2486523f0bd1120b0d87df21
SHA-1: 5048bb4ac50049ba1490347920a65d7309e1ebfb
SHA-256: 8ec44187e50c15a7c4c89af4a1e99c63c855539101ec1ef4588d2e12e05f7d2bMD5: 8c7086bf1606da31134b36cef134a91f
SHA-1: ee85527be9f6017e9dee0a9979a9b948a8d244be
SHA-256: bc74f22b5407ac67b8d7dcb05262bee0dc9581620448c2b6514ed519ab7f6bd2
These samples call themselves NimGrabber based on the report it builds to be sent back to the actor.
The data stolen includes Discord tokens are data from popular browsers, including:
- Vivaldi
- Microsoft Edge
- Opera
- Chrome
- Yandex
- Brave
After harvesting the data and building out the report it is exfiltrated by sending the data to a Discord webhook:
hxxps://discord.]com/api/webhooks/891404241124605982/evrTSNuCyasJcFT1KiY35gFCugpWZFoE7VfNXtlhrEfJLWeYK-p3dYvM4Ev84YoPbKAL
CobaltStrike Stagers
MD5: 9d6c1908baa481203faa31bce05ab8b2
SHA-1: 76441909714108823c90f7ed9603d21bab53d801
SHA-256: 41f40f8bbaeaeb811e5a9f8ba7870e6165fc749fe1121d09da30b127291ef351
This stager has a base64 encoded blob onboard but then leverages 3DES for decrypting the decoded data. As it turns out this stager is built based on a project called NimShellCodeLoader[6] which can leverage multiple encryption routines including our previously mentioned Caesar or lookup table based encodings from our previous report[1].
After Base64 decoding we are left with a blob that contains the length of the plaintext followed by the 3DES key and then the encrypted shellcode blob. Unfortunately my version of 3DES in python was incompatible, I’ve talked about how this problem can come up with encryption algorithms in the past. The easiest solution here since we have the source code is to utilize that to decode the data.
proc D3DES_Decrypt(plainBuffer:cstring,keyBuffer:cstring,cipherBuffer:cstring,n:cint):cint {.importc,cdecl.}
import base64,strutils,sequtilsconst source {.strdefine.}: string = ""
var code*:cstring
var codelen*:cint = 0proc de3des(enbase64:string): void =
let shellcode:string = decode(enbase64)
let plain_len_byte = cast[int16]([shellcode[0],shellcode[1]])
let input_encode:cstring = shellcode[26..high(shellcode)]
let key:cstring = shellcode[2..25]
var str = cast[cstring](alloc0(plain_len_byte));
discard D3DES_Decrypt(input_encode,key,str,cast[cint](plain_len_byte))
code = str
codelen = plain_len_bytewhen isMainModule:
let data = readFile("blah.bin")
de3des(data)
let f = open("blah.dec", fmWrite)
let b = writeBuffer(f, code, codelen)
close(f)
If we wanted to make a standalone decoder in python we could either search for alternative python implementations or just compile our NIM code into a shared library and call it using ctypes from python. Regardless having the code allows us to easily decode the shellcode blobs from samples.
1.14.66.81
/3mXe
User-Agent: Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 5.1; Trident/4.0; .NET CLR 2.0.50727)
Sample:
MD5: 5fd028a9bd6087c70c0a09cc2ac8f2fd
SHA-1: 0d6b34d4c9678dd35155093fde7aaca1847d3f09
SHA-256: d34bc5060dd7e433bd11f16fb7f2ef289511476a2ba32721078483fbc0372024
Data:
81.69.224.81
/dJv9
User-Agent: Mozilla/5.0 (compatible; MSIE 10.0; Windows NT 6.2; WOW64; Trident/6.0; MDDCJS)
Sample:
MD5: da61a622aff329ad08141387c9432b1a
SHA-1: c49649cc71e7aaf243265959ec372173947a34a0
SHA-256: ff261192a1defd66fcd5924e04c04cf255859beda3a02bb58dfe6d3e211d9c04
Data:
192.168.111.141
/LwAY
User-Agent: Mozilla/5.0 (compatible; MSIE 10.0; Windows NT 6.1; WOW64; Trident/6.0)
Sample:
MD5: c48763be59b1ba3fe81b06a31a5369bc
SHA-1: 2e4086ca701304d67d14063b2105c20ff85a366c
SHA-256: 40f8ca4c9f19d0330e42c98b9d0396b9f0caf191c6a544df4e4edb6837ed542c
This CobaltStrike stager appears to be more custom than the others, it still leverages base64 but has a replacement set of characters that it will do before decoding and loaded the shellcode onboard.
Here we can see the base64 like string but with non standard characters mixed in:
Before being Base64 decoded some of the characters are replaced:
After replacing the characters and base64 decoding it we are left with CobaltStrike stager shellcode:
/Mds9
User-Agent: Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; QQDownload 733; .NET CLR 2.0.50727)
www[.monksec[.tk
This server was still at the time of analysis:
'DOMAINS': 'www[.monksec[.tk,/j.ad'
'SUBMITURI': '/submit.php'
'WATERMARK': '305419896'
'USERAGENT': 'Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; WOW64; Trident/5.0; MANM)'
Ransomware
MD5: 8c09df22e86a70bfc3aa541e5ccab80b
SHA-1: bfb1b412fde1ba4bf26d10ac7d76915051fafc3e
SHA-256: a7517bca3c161893f0af6884415538defe3da75534e6b2b75720a61ae1c77abf
Ransomware is the hot item lately in the media and finding tools or samples associated with them written in new languages should not be surprising, in this case however the ransomware turned out to be from a blog[7].
Sample:
MD5: d65a8c0a2e771bb3b9664559552e69a7
SHA-1: 4912dfe9fa35f5e209a5c019f5abce9a4e64132d
SHA-256: 29f56a007e3f9b19a9a6ad9eab2af8edcdb3164a6f3323f0463c2983f83cdcef
This is another ransomware sample that also appears to be in testing but this time they have added a connection piece that receives content from a HTTP request to an embedded domain. Internally the objects are named ‘exfilurl’ for the embedded domain:
And ‘rware’ for the function responsible for encryption files:
File encryption appears to be a hash of a hardcoded string that will be used as an AES key in bcmode:
With the file then being overwritten:
The ransomware sample only targets files a hardcoded location with a ‘.local’ domain along with the source filename artifact in the binary of ‘RwareEmulator.nim’ makes us suspect this is still testing or demonstrations being done.
Backdoor
Another sample found appears to also be related to testing:
MD5: 28d231ca6d2b5c219eb23abbf2e6eec6
SHA-1: 3512085d14fada2a5afc123efc4a087a48b98be1
SHA-256: 181b1d6ba674bd6d4f786c6838b2ba36f79e7210cb1b7cef9de93aa68153b488
This sample operates more as a simple reverse shell backdoor called NimRev, there are a number of examples of NIM based reverse shells on github but the disassembled binary most closely resembles the code from this blog[8].
The sample takes an IP and port as parameters and connects back to them, then signals to the server that it is ready to receive a command by sending a ‘> ‘:
After receiving a command it detonates the command and sends back the output:
While reverse shells are fairly simplistic backdoors they can very effective.
Another NimRev sample:
MD5: 533082e631bb02f640dde461b9b71939
SHA-1: 955578c34fe814d0ffaf87b8e38ca2510dbc7f9f
SHA-256: 8503bd24fd22b322a8ccd2290a655630cef45fe144a386ef57cb9a7cb8ae5bef
This one is much the same except that the prompt it sends back to the server is ‘nimrev> ‘ instead of just ‘> ‘.
YARA rules
rule caesar_cs_stager
{
meta:
author = "Jason Reaves"
strings:
$a1 = "caesar__w9"
condition:
all of them
}rule nim_sc_loader
{
meta:
author = "Jason Reaves"
strings:
$a1 = "de3des__"
$a2 = "D3DES_Decrypt"
$a3 = "Direct_LoadPcy"
condition:
any of them
}rule nimrev
{
meta:
author="Jason Reaves"
strings:
$a1 = "nimrev.nim"
$b1 = "result__"
$b2 = "VirtualAlloc"
$b3 = "recv__"
$b4 = "send__"
$b5 = "execProces"
condition:
$a1 or all of ($b*)
}
References
1: https://medium.com/walmartglobaltech/investigation-into-the-state-of-nim-malware-14cc543af811
2: https://medium.com/walmartglobaltech/nimar-loader-4f61c090c49e
3: https://medium.com/walmartglobaltech/buerloader-updates-3e34c1949b96
4: https://www.proofpoint.com/us/blog/threat-insight/new-variant-buer-loader-written-rust
5: https://www.proofpoint.com/us/blog/threat-insight/nimzaloader-ta800s-new-initial-access-malware
6: https://github.com/aeverj/NimShellCodeLoader
7: https://ilankalendarov.github.io/posts/nim-ransomware/
8: https://trustfoundry.net/writing-basic-offensive-tooling-in-nim/