Javascript application testing - Firebug

Recently I was testing a Javascript application. By Javascript application what I mean, is that a lot of the business logic was written on client side. Now it's a reasonably well known fact that client side logic is easily by-passable.

There's plenty of ways to do it ..intercept in Burp, delete the Javascript from the cache folder on your disk, create your own forms or use Firebug. Maybe more. Largely though I find that a combination of manual probing to detect where client side validation is being used, intercepting every single request for every file type and then dropping the Javascript request responsible...works.

There are times though, like in my last application where the entire UI of the app is constructed client side, where dropping Javascript will not load the app at all or mess it up bigtime. So Burp, in such cases, while you can still certainly use it...is not the best tool for the job. The next thing you will hear, if you google about 'Client side testing' is that 4 out of 5 people will say - Use Firebug.

Now once you understand it, it's an absolutely fabulous little tool to use but until then, it can get a little frustrating to use. So what I'm going to do here is to just point out the features I use the most and why I do so.

Inspect Element

The first feature that I use a lot is the 'Inspect Element'. You can open up Firebug and click the button that's highlighted in Red below and then move your mouse over the upper half of the page. As you hover over each part of the page, the corresponding part of the HTML in Firebug will start flashing. It's all in real time.

When you're clear which bit of the page you want to look at in Firebug, click on that element. Firebug too will stop flashing around. I tend to use this a lot when I want to look at the attributes set for a particular form. For example: If I want to check if the autocomplete attribute has been set on the password element, I can focus just on that and click. Then I can read the HTML on that page in the lower pane.

If you want to change the maximum length of an input field in a form, you can double click on the bit you want to change, and enter a new value. The key here is to ensure that whatever you're typing in is syntactically correct otherwise Firebug won't make the change and will leave you frustrated. For example: If you're entering a string, make sure you put it between double quotes.

Javascript

If you want to change just a part of the Javascript on a page, click on the script tab and then reload the page. All the Javascripts on that page will get loaded. Now you can click on the little drop down there (marked in Red) and see all the Javascript loaded. Click on any one to load it up in Firebug. Don't keep clicking on the dropdown marked in Blue. That's just the main tab that you can use to get there.

If it's a small application you might want to manually review the Javascript on each page. Click menu - Let Javascript load - Review. Repeat the cycle. If it's a huge application like the one I was testing though, there's a high chance you'll go mad very quickly :). Trust me. So you need an upgrade to your testing technique. That's where the DOM tab comes in.

DOM

Anything that's loaded in the browser will read/write something from the DOM. Now lets say you loaded a page up which dealt with logging in to an application and you want to check if the authentication logic is client side. That'd be a rare thing and largely quite a stupid thing to do, but it'll do for now :).

So if it is, there's a chance that some JS functions are being called. But you don't know which one and in which Javascript. So as soon as the login page loads, click on the DOM tab and scroll down a little. In whatever Javascript applications I've seen, there's usually a tree full of elements at the top of the DOM. All the reading and writing from/to the DOM happens here. Just below that though there is a large list of functions (all in green font) as shown in the figure below.

Now read through all of those names and make a little list of which functions 'sound' like performing a login sequence. If you click on each of those functions in the DOM, you'll immediately jump to the Javascript which has the code for that function. NOW you can study just that code and see if its insecure. Yes, you're still studying code..sure..but at least the scope is reduced a little.

Setting Breakpoints

Now we've made progress and come up to a point where you know what bit of code is "possibly" doing what. I say possibly, because you haven't actually run it or in this case, clicked on the login button to actually make it happen and see if THAT function is triggered. Let's now confirm that.

If you look at the screen-shot below it'll show you the exact line number on which those 2 functions are called. You want to make sure that those ARE called. So you set a breakpoint on those lines. A breakpoint, for anyone reading this and not too familiar with the term does what it says. It breaks. Breaks what? Execution. Of what? Code. What code? Javascript. When? When I click login.

So I'm just saying..I'll click Login and if the code here is triggered and my guess was right..don't go on. Stop right there. Set breakpoints as follows..set it on the first line of the function and not the definition itself. I'm not sure the line that has "function" is triggered; it may be..I'm just not sure as I haven't had success. Here's a screen-shot with the breakpoints.

Now try and login..

Yeah..actually put in a user-name and password..and login. If your function guess was right, you should stop right where you put a breakpoint. If it doesn't stop, it's the wrong path flow for THAT function. So maybe there's some other place login is called. Search for all such instances..and whenever you see a function with login in it...set a breakpoint.  Here's a screenshot where I've searched for login, a breakpoint has triggered and the credentials I entered can be seen in the Right side pane.

Now it's like Burp inside a browser window :). Read. Edit. Forward. Only the forward button here is the Continue button .. the right pointing arrow. If you want to go Step By Step and see what happens...and many times you will want to do this, you can use the Step Into or F11 key to now proceed line by line and watch the path the code takes. Again..when you're editing stuff..make sure you are syntactically correct. Otherwise the change will not persist.

Adding to the DOM

So the last cool feature in Firebug that I'll go over today is how you go about adding an element to the DOM. So suppose you want to add a parameter called 'admin' and set its value to 'YES' so you can try logging in as admin, you have to use the Firebug console to do so. So first decide where you want to add the property.

Once you're clear right click on a different property; in this case any root element like PR_TAB_WIDTH and say "Copy Path". Then click on the leftmost tab called Console and paste the path into the bar at the bottom and change its name and value. Like this..

And hit Enter. Now go to the DOM and you should see a new property called Admin with the value YES. It didn't work here of course..to give me admin privileges, but you get what I am trying to say :)

So that's it...start playing around with Firebug, it's a supremely powerful tool, specially when you're testing JS applications. Until next time..Cya :)

Salt your passwords = Existing accounts password reset

This one is just a simple conclusion that I drew when a customer had MD5 hashed passwords in his database and we recommended them to salt them instead and use PBKDF as the algorithm of choice.

Now the customer added code in, but left the old code in as well which did the simple hash comparisons. So I asked them why and they said..for old users. I started thinking about it and it made sense. Here's why:

Think of it. I have a MD5 password in the database. Say I add a new column for 'salt' , generate 1000 random hashes..one for each user..and add a new column. I then change the code to query the DB for the 'salt' as well, when a person logs in and run the user's password through the PBKDF function before comparing it with the hash. Looks good?

Sure. Except that no one will be able to login. Why? It's super secure now :). No one can find any more vulnerabilities inside. Jokes apart, what's happening here? Lets say the user's original password was abc123 and it had a 32 bit MD5 hash XYZ. Now when the user logs in something like MD5(user entered password) runs..hashes what he entered and compares it to the stored hash. Match? Login successful.

Now there's the salt. So when the user logs in, the new 'salt+pbkdf' code is triggered. So PBKDF(userpwd + salt) is calculated and compared to ..what? The stored MD5 hash. Is there ever going to be a match? Largely no. Because..there is no way you can predict the salt for a particular hash..BEFORE you store it in the DB. Remember those 1000 salts you generated? All random? Did you check if they'd give you the user's password after "decryption"? Nope.

So since there's no way to get the "right salt" for each user ID, the easiest solution is to force a password reset across users. The next time they login everything will automatically fall into place.

Incremental code review - MD5 hashes

This one is a reasonably obvious tip to be frank. I was recently doing a code review retest for a Java app. Now I don't know too much Java so I was finding it hard stepping through code.

One thing I knew though was that if you have the previous code base and the current one and are "re-testing" it's helpful to see which files have changed. Now anyone with a little Linux experience or programming experience will immediately say 'diff'. And yeah..diff is fine..but that will give you a very detailed listing in a slightly cryptic form. Its readable enough if you go slow, but it's not the first step IMO. I'd like a slightly more high level view first. That's how I stumpled upon md5deep.

So I go into the first root directory of the old code and run md5deep -rl > old_list there. Then I change directories to the new code and run md5deep -rl -X old_list > changes over here. Basically all it means is that I compared the new list against the old. Once I have my changes, I know that any fixes they made MUST be in some of those files.

Now I can run diff on each of those file combinations to search a little bit slowly. It's the same really..but it gives me a little more control over the process.

A tip here to further reduce changes is to use md5deep with grep to filter your results out. So I knew I was looking only for java file changes...so I could either do something like:

md5deep -rl -X old_list| egrep -i '.java$'

OR

md5deep -rl * -X ~/list_md5_old |egrep -v ".doc$|.DS_Store|.js|xml|changed_files|hibernate" > changed_files;wc -l changed_files

I used the 2nd one :) coz I knew what all I didn't want. If you don't know that..start with what you know and filter on from there.

There's a million ways of doing this of course; I just dropped in what I did :). Here's the blog post which saved me some time..

http://linhost.info/2010/02/checksum-a-directory-with-md5deep/

Cross Domain XHR

Recently I was doing a project (REST web services) where there was an interesting implementation of a defense against CSRF attacks. You would login to the application and immediately be assigned a session ID in a cookie; along with this you would also be assigned an AntiCSRF token, in another cookie. Now this immediately means that both cookies are going to be sent across to the server each time and going to be verified each time. Hence, if this had been the only defense it would be pretty useless; as the additional cookie does..nothing.

In the application though, what was happening was..the cookie 'token' was read, its value grabbed and a new custom header called X-Security-token added in the header with its value. So if on login I get a cookie: token=abcd, the next time I say...do an Edit Profile...I have a new HTTP header called X-Security-token: abcd as part of the request header.

Now my first thought was, Ha, spoofable...the attacker just needs to write a little Javascript, read the cookie (document.cookie) and then add a new custom header and send the request off.

The data was all in JSON format and I wanted a POC to show so I thought I'd just have a simple HTML file with some Jquery code which would make an XHR request by doing all the stuff I talked about above. The code seemed all okay when I tested it on localhost. As in..I set a cookie with the domain as localhost and value as 'arvind' and tried writing JQuery to pick it up and create a new header. That worked and the request DID read the cookie, append it to the header..and send a request to localhost.

The moment I changed the URL to the actual site URL, the request never even fired. I could understand it if the cookie didn't get sent but the request never fired either. More reading revealed that this was due to XHR's limitation of sending cross origin requests. It just refused to send anything at all; I couldn't even see anything in Firebug. I change the URL to localhost..immediately an XHR request gets fired. I tried removing the cookie and resending it...no luck..same behavior. Posts to localhost. Okay. Anywhere else. Browser doesn't allow sending an XHR request anywhere else. Not yet totally clear what's happening.

The other strange thing though is that there is this addon called a REST client which is installed in the same browser and seems to be able to make calls happily to multiple domains other than localhost.

So the question is..What is the difference between an HTML file hosted on a local web server..making XHR calls to abcbank.com AND a firefox addon making calls (not sure if its XHR as I did not see an X-Requested-With header) to abcbank.com? Why does the latter work?

So as a CSRF protection...it doesn't seem safe but I'm not able to get it working with XHR at least. There's some suggestions elsewhere that some versions of Flash allow this; but right now I'm not sure. I'll update this later if I stumble upon something.

SoapUI - XML Entity Injection

So another thing that I tried out with some help from the guys I work with was this attack called XML Entity Injection. It has been around for quite a while and there are plenty of resources on it so I won't talk about it too much. Here is an OWASP reference and another link which was quite cool and explained things well.

In my previous post I described how to use SoapUI through Stunnel to test HTTPS web services. So following up on that, I was using SoapUI to test XXE attacks as well. So I followed stuff given above on the link to try and read the passwd file from the remote machine (which I'd confirmed was a Ubuntu box). To my pleasant surprise it immediately worked and the entire /etc/passwd file got displayed in the Soap response. Yay. Victory jig. And all that ;)

I was brought down to earth though when I saw an account called 'arvind' in the response. For some stupid reason, SoapUI was retrieving my own passwd file and displaying it in the response. Huh?

So I set Burp up again and set Intercept to on to see what was happening. To my surprise, the request itself already had my passwd file..and the response was just reflecting it back. So even before hitting my own client proxy SoapUI had done 'something' with my request, expanded my own entities and THEN sent it to Burp. It wasn't ever reaching the server directly. Meaning..if I used SoapUI there wasn't any way to directly test XXE...unless there was some tweak somewhere where I could turn off request parsing.

Anyway since I had Burp as well in my proxy chain before it hit the server, it didn't really matter and I could test it in Burp. Turned out that the server was not vulnerable at all. Oh well. Maybe next time it will be ;)

SoapUI to Burp - Fuzz away

I was recently testing another web service (lot of them recently) and was using SoapUI as a client to invoked the WSDL. You don't HAVE to use SoapUI; its just that it was the most popular tool out there and seemed to do its job.

So one of the main tests in a web service is to fuzz all the methods. For the new guys it just means the following:

a) Identify the method and find out what arguments it has
b) Find out what valid arguments need to be passed for the method to work
c) Fill each argument with a huge bunch of malicious data specific to SQL Inj or XSS or any other type of injection and see how the server reacts

Now you could do this manually and put in a ' , < , > , #, OR 1=1 and so on manually and see what responses you get back. That, I guarantee you is not going to be too much of fun after around 10 minutes max :). So you want to automate this process.

SoapUI has its own security scan from version 4.0 onwards and this is what a large number of people working with web services tend to use. I'm not sure what I was doing wrong, but the free version allowed me to scan just 4 specific parameters [username, password, request and one more]. It has a huge number of scans; SoapUI but if it is not going to allow to me to fuzz the parameters I want, it kinda is useless. Maybe its because I was using the free version and the Pro version supports scanning all parameters. Anyone tried that out? If not, no worries, I will ..sometime :)

So now I thought of chaining SoapUI to Burp and using Burp to do all my fuzzing instead. This turned out to be harder than I thought and frustratingly very little good help on how to do it. Note that it was a HTTPS connection; if its HTTP it is easy enough. SoapUI seems to have something called a HTTP Tunnel which I guess you have to configure with keystores and trust stores...but whatever I did I couldn't get it to work. And I couldn't find any other fuzzers which would reliably work. I am sure there might well be something...I was on a deadline though; you know how that goes.

So I pinged a few of my colleagues for help and Max helpfully suggested that I use stunnel to make the SSL connection to the server instead. That actually worked and  I could get everything working. Here is what I did and why it worked.

a) Set SoapUI proxy settings to Burp - File: Preferences: Proxy Settings. Now load up any 1 request in the SoapUI request editor and edit its Endpoint to point to http://abcbank.com:443/blah/blah instead of https://abcbank.com:443/blah/blah. So it means...if you want a response for this method go to http://abcbank.com:443. This is just to ensure that the request gets to Burp...in the first place. If you keep the endpoint as https it gives you a 'Cannot cast to HTTP' error. I think that is because the proxy setting in SoapUI is ONLY for http.

b) Start Burp

c) Open Terminal (Ubuntu) ..probably CMD on Windows although untested and start stunnel as follows: sudo stunnel -cf -d 127.0.0.1:443 -r www.abcbank.com:443

d) Go to Burp - Options - Hostname resolution. This is the one to the left of the Alerts tab and set  the hostname of the remote site to 127.0.0.1. So you should have an entry in there which is Enabled and says:
www.abcbank.com 127.0.0.1

e) Load the WSDL up in SoapUI and invoke any method. Look into Burp; you should see a request in its history. Look into stunnel; you should see some data getting written to its console [Service stunnel3 accepted connection from 127.0.0.1:50318]. Yay. Done :).

f) The request is now in Burp. Uncheck the Enabled box in the Hostname Options in Burp. Now have fun with Burp Intruder/Scanner as usual.

For the curious..what exactly happened and why did this work? Well to start off, we simply set a proxy for SoapUI, so anything from SoapUI will go to Burp. Now Burp, like any other proxy needs to forward a request it got to the destination, else it'd be useless rt? ;)

Now just like your browser resolves a hostname to try and open up a site for you, Burp needs to do that as well...but it first looks into its hostname section to see if you've hard coded something there. You have..in this case.. www.abcbank.com = 127.0.0.1. So the SoapUI request goes to 127.0.0.1.

What's the destination port when you sent it from SoapUI? 443. And we haven't changed that in Burp. Its just a different hostname..that's all. So the traffic gets sent to 127.0.0.1:443.

And what's listening on 127.0.0.1:443? Stunnel of course. And Stunnel has been configured with the -r argument to forward all its traffic to abcbank.com:443.. So stunnel, since it understands SSL..completes the HTTPS connection with abcbank.com and returns the response to Stunnel - Burp and then SoapUI as well. Very very neat :)

Oh and once you have all of SoapUI's requests in Burp, you don't need to use SoapUI any more; you can just use the awesome Burp. You also do not need Stunnel anymore..as Burp can understand HTTPS by itself. The reason we needed Stunnel was because SoapUI did not seem to directly understand SSL. I am NOT stating this as fact..because I'm not sure; but certainly, from what I searched and worked with, it isn't that easy.

I hope that helps anyone who is trying to get this setup. Again big thanks to Max, Google ;) and this post.

ASMX Webservices - XSS

I tested a few public Web services recently. 1 of them had a front end so it was easier to visualize but that apart it was quite tough, specially as numerous methods were so complex that the developers had a tough time trying to answer some of my questions.

Now assuming that the application that consumes the web services is not in testing scope, all you have is the .asmx and the .wsdl. At times, these methods might be invokable from the browser itself, via the ASMX web interface like this. Other times invoking is restricted to the local machine alone, in which case you have to use a third party client like SoapUI or write your own using a library. I wrote mine using a library called Suds (Python). Obviously, writing the client is more work and if the arguments that need to be passed are complex, it can be quite tricky.

There are 2 ways to test for SQL Injection or XSS. In my case I could invoke the methods via the browser, so that made it easier and I could find a SQL injection.

a) You can code your custom client and fuzz through that.
b) You can simply write a very simple client, with Burp set as a proxy. You can then use Burp Intruder or its inbuilt scanner to fuzz the application.

I wasn't lucky with the XSS though. The content type of the response was text/xml. I saw a few attack vectors for XSS in xml but all of them involved closing the existing XML tag but for any of them..I'd have to break out of the existing structure. And < and > were getting encoded..to &lt and &gt. So unless I could do that, I couldn't see anything anywhere that would make this vulnerable to XSS.

And much like my Flash post earlier, the content type is text/xml. So unless I could find a way to get a browser to forcefully treat it as text/html, there wasn't going to be a way to do this.

A colleague said this - "...In terms of triggering XSS it will be harder unless you can convince the client to render it using an XSLT" coz that would convert the XML into HTML. But I wasn't sure how I'd go about this. And then I ran out of time :).

So here I am. You guys know of how this is possible?

Update: I posted to the list and had a few ideas here.

Testing a Flex application

So I have been testing quite a bit recently. A while back I ran into an application which had a Flex frontend. The Flex UI talked to a backend server which ran Flourine FX. This eventually talked to a .NET application - MSSQL Db combination. The Flex client used AMF 3.0 to talk to the server.

This whole thread by the way is just a sort of a rant on how I could NOT do things and all the things I failed at. It offers no solutions. It is however interesting if you want to know what to avoid. If that's okay..read on.

Now I knew that Burp had support for AMF for sure. And the application was browser based, so everything should have been normal. Testing for SQL Inj, XSS, Authorization bypass tend to take the most time in a test. Usually though, if you can reliably intercept and replay traffic, it is quite doable. I had no reason to think that this would be anywhere different; considering that Burp talked AMF.

a) Authorization Test

Now, this particular app had 2 types of users: user & manager. So I tried to replay a manager request as a user, but it kept failing. So I think..okay..protected. Then I took a menu which both users had access to. Replayed request. Fail. Huh? Both have access, why should it fail? Something very funny going on. Or maybe I have no clue (more likely :)). The last test I did was trap a user request and replay it as the same user, inside the same session, inside the same TAB. Surely this should work?? FAIL. Huh?

Now I'm confused. This means that something is happening ON the client itself. Sent 2 user requests to Burp Comparer. Turns out there are 3 values that change with every request. They are: ClientID, MessageID(under body) and DSId (under body and then headers). Now the ClientID and DSId didn't seem like they were too important, as in, tampering with those didn't give me an invalid request. However anytime I touched the MessageID, I was rebuffed.

So drilling down further MessageID seems to be of the structure of a 32 bit Guid. Spoofing an older valid message ID or a newer message ID or a pre-generated but not sent (Dropped request) message ID all fail. There is nothing that comes back in the response either; like an Anti CSRF token in a hidden field. So I am sure it is nothing that is stored on the server. If it isn't though, how does it recognize just THAT ONE message ID? And no... it does not seem to be all client side either. This is because I could see AMF requests being sent; dropping them also caused an error.

b) SQL Injection and XSS

At this point I hit a limitation with Burp. In a normal request (non AMF) you can add, modify and delete parameters. Right? So in AMF, while I can still modify data in individual parameters, I cannot add or delete anything through Burp.

This probably has something to do with the structure of the request. I tried Charles too, but have the same problem. Now I can sit and manually put a single quote and < > in every parameter but I'd probably die before I complete the test. Meaning...I cannot fuzz every parameter reliably using Burp. That's cause although there is repeater and intruder support to parse AMF, it doesn't detect the exact places properly...so I can choose which parameters to fuzz.

The closest I came was to carefully look at the 'Raw' request which tended to be just 1 value and fuzz just that. I got a few errors here but nothing concrete from a SQL Inj perspective. Ditto XSS. I tried deblaze, Pinta and a ton of tools which are already out there on the OWASP Flash page. None worked. It almost certainly meant I'd have to write my own client specially for FlourineFX. A custom BlazeDS client written by a colleague also failed. On this particular engagement I ran out of time.

The last thing I wanted to talk about was that the Response content type was x-amf. So all the Stack overflow and Google threads I read, talked about how this was real hard to do. I bounced this off a few guys and everyone did say...If there is no HTML content type responses...you can't get JS to execute. Which made sense.

c) CSRF

The fact that the message ID was 'unspoofable' automatically gave this Flex app protection against CSRF. No token. No nothing.

So in a nutshell, its super hard to test for SQL, XSS, CSRF and Replay traffic with existing tools. Almost certainly a custom FlourineFX client has got to be written. How and when and by whom I don't know. Maybe I'll write one some day :)

You guys have any ideas on how anything could have been done better? I saw a talk on BlackHat12 which talked about this. I'll check that out. That apart..anything?

Reverse Engineering - Android APK

There's nothing really complex about this post. In the past I've always maintained that reversing is kind of tough; and that's true if its an EXE,DLL etc. But in the case of an Android application; it really is very easy.

While all this information is already there, here is a very short blog post summarizing how you get from APK to Source. I used a Vuln app available here; but you can use anything really.

If all you want is the source, do the following:

1. Use dex2jar downloaded from here and run ./dex2jar.sh . This results in a JAR file getting created in the directory where the APK is already present. The JAR file contains all the JAVA class files; namely the Java byte code - something that you get once you compile your Java code.

2. Convert your Java byte code into actual Java code. You need something called a Java decompiler for this. You can download one called the Jd-Gui from here - http://jd.benow.ca/#jd-gui-download.Load the JAR into it and use the 'Save all sources' feature to save all the source (Java) files to your disk. Now you can review it like you would review any Java code.

Some other interesting things about APKs though:

1. An APK can be extracted to a folder. Its just a Zip file so any archiving program should do; I use the inbuilt Ubuntu GUI archiving tool. You can also use unzip, 7z, WinRar or anything you want.

2. Look at the file AndroidManifest.xml. It'll open in a Text editor but its largely binary; hence unreadable. Use Apktool to decode this XML file as well as every other XML file in the APK. You can get it from here. Run apktool against the APK. It'll run and give you all the XML files totally decoded; so you can now read them.

3. All the application code is in classes.dex. This is Android Byte code, reversible to Java Byte Code. So we first 'dedex'. Use 'dedexer' that can be found here. Run it as follows: java -jar ddx1.22.jar -d

classes.dex.

Nothing in this post is original. I just wanted a place to refer to instead of Googling a million times for the syntax.

Market app - Android 2.2

Was doing a test on an Android app (first ever ;)) and was trying to install an app which was available on Android market. As it turns out, the emulator doesn't have the app pre-installed. So I followed instructions here. That largely works except that despite following everything the app vanishes the next time around. So after more digging I came up on this thread which provided the missing piece. I'll sum it up quickly.

--- You need to edit files on the emulator which will prevent the emulator from checking if its a valid phone or not
--- You need to upload the packages for the Market app and the GoogleServicesFramework as those are not shipped
--- You can do both of the above only if you have WRITE access to the emulator's system directory.

So far everything is covered by Varun's blog. The 1 point which he hasn't made clear is that once you've pushed all your changes to build.prop, uploaded the 2 apps mentioned above and deleted SdkSetup.apk you have to sync your changes and save the temp image in which all these changes are made.

Think of it in this way..all the changes you make are temporary and made in a file /tmp/android-. The filename starts with emulator. Once you're done making your changes save this file somewhere. Shut the emulator down. Now rename the saved file to system.img and copy it into ~/.android/avd/.avd. Now boot the emulator. It should work.

I've said nothing new frankly. If you use Varun's post and the post on the XDA forums together, you will be able to do this. However I was stuck for quite a while here, so thought a few guys would benefit by me linking the two together :)

Blogging elsewhere for a bit

I've been blogging at Infosec Institute (a little :)) and on the TeamMentor development blog (a lot) ... so that's why this isn't as updated with my experiences. Do peek at my work at II or on TeamMentor if you're interested in knowing what I am up to :)

Olly - Hit trace..finding code flow

Many a time we come across malware which is extremely confusing and has lots of JMP and CALL statements. It becomes very difficult to find out what exactly the malware is doing; if you use a purely manual technique..and you start seeing stars after a while. At least I did :)

So to help you understand the control flow better you can use something called a Hit Trace in Olly. Here is how you can best use a Hit Trace:-

a) Load the executable into Olly. You'll have to use 2.01; I couldn't find this in Olly 1.10.

b) Till what point in the code, do you want to understand code flow? Identify that point and set a breakpoint there using F2.

c) Once you've set the breakpoint, navigate to Trace - Run hit trace. This will automatically run the malware until the breakpoint that you had set earlier.

d) If there's some anti debug measures or the process simply exits before your breakpoint...you won't be able to see the Hit Trace at all. However, if the program is still active, you should see a little red dot to the left of each assembly instruction. This just says....'This instruction has been run at least once'.

You will notice a lot of gaps in between the 'red dot' lines. That is because those instructions were never triggered at all. You'll see this specially in cases of JMP instructions.

There's another feature called 'Run Trace' in Olly and the way to use that to identify code flow is to Log to File and then study the log.

Or lastly (as far as I know) you can set breakpoints and then use the 'Trace Over/Into and Animate over/Into) to step into and over code interactively using the '+' and '-' keys to go forward or back.

But the Hit Trace is quite neat and it offers a very nice option to identify control flow. Until next time ...bye :)

EXE dependencies - Read from PEB

I was trying to understand where executables load all their dependencies from. So for example: When I load up minesweeper (winmine.exe) there are a host of additional DLLs that are loaded up as well. That's fine as there are functions inside each of those DLLs that winmine uses. What was interesting to find out though, was how the list of those DLLs was built in the first place. Where did the OS loader look...into the process..to find out 'what else to load'?

So I loaded up winmine.exe in Olly, Dep Walker and a few other tools to get an understanding of 'what all' was loaded. Once that was done I read up a little about the PE header to understand where dependencies are accessed from. In PE terminology, these are called imports. It turned out there was a Data Directory for the Import Table. So i thought; lets read the import table; that should give me a list.

So I wrote a little bit of Python code [after learning Python from Google's Python class ;)] to read the IMPORT TABLE from the PE header. This did give me quite a few DLLs but it still did not tally with what Olly was displaying.

The next thing I did was to modify my code to be recursive. So...

--- Load EXE. Get dependent DLLs from import table.
--- Load 1st DLL from previous list. Get its dependent DLLs. Add to list.
--- Load 2nd DLL from previous list. Get its dependent DLLs. Add to list.

And so on... until all dependencies were resolved.This was doubtlessly an improvement as I was closer to what was displayed in Olly [Alt+E]. It still wasn't complete though.

Then I read some more, read how Dependency Walker works, asked on Woodmann and re-read a few sections from Xeno Kovah's Life Of Binaries class. Turns out there are somethings called Delayed Imports. These aren't loaded statically at load time; like the Import table but only when a particular function needs to be called..later on...at runtime.

So I modified the code a little to recursively get all the DELAYED IMPORTS too. This would have been enough but it gave me a huge huge list which contained much much more than what Olly ever showed me :). So obviously this wasn't efficient either and it was something else.

So as usual when I'm stuck I bug the nice guys on Woodmann and I'm rarely disappointed. This time was no exception and I was pointed in the direction of something called a PEB.

The PEB or the Process Environment Block is apparently the only "part" (very loosely used) of a process that is in User Mode. Everything else is in Kernel mode. So if you want to find out every DLL that an EXE depends upon; its best to query the PEB instead of doing all that stuff I did earlier :)

I want to quickly touch upon the structure of the PEB before I finish this post. To look at the entire structure of the PEB you can visit this link. You need to query the PEB_LDR_DATA structure to get the list of Loaded DLLs. Here is an article that I found quite useful while I was learning stuff about all this.

And lastly..here is the complete thread of my discussion on Woodmann, if at all you are interested :).

Links - All recent articles

I've been primarily writing for Infosec Institute these days; so there's very few updates on my blog. I could duplicate stuff here; but that's not of much use to me or anyone :). So if you're interested in what I'm writing about, do take a look here.