Tag Archives: secure development

Two-Factor Authentication Considerations

There was a recent article talking about how a very small percentage of google users actually use 2-factor authentication. You can read the full article at http://www.theregister.co.uk/2018/01/17/no_one_uses_two_factor_authentication/

Why 2-Factor

Two-factor authentication, or multi-factor authentication, is a valuable step in the process to protect accounts from unauthorized users. Traditionally, we have relied just on a username/password combination. That process had its own weaknesses that many applications have moved to improve. For example, many sites now require “complex” passwords. Of course, complex is up for debate. But we have seen the minimum password length go up and limitations on using known weak passwords go up. Each year we see lists of the most common passwords to not use, some being 123456 or Password. I hope no one is using these types of passwords. To be honest, I don’t know of any sites I use that would allow this type of password. So many these days require a mix of characters or special characters.

https://www.youtube.com/watch?v=YxXebkpSLr8

The above controls are meant to help reduce the risk of someone just guessing your password, there are other controls to help try to limit brute forcing techniques. Many accounts offer account lockout after X number of invalid attempts. There are other controls that we also see implemented around protecting the username/password logic. None of these controls help protect against a user reusing passwords on another site that may be compromised. They also do not protect against a user falling for a social engineering attack to trick them into sharing their passwords. To help combat this, many sites will implement a second factor beyond username/password.

The idea of the second factor is that even if you have the username and password, you will not have this other piece of information. In most cases, it is a value that changes every 60 seconds or so, and is delivered over a protected channel. For example, the token used may be sent via SMS, a voice call, or created through a phone application like the Google Authenticator application. So even if the attacker is able to get your password, via a breach, brute force, or just lucky guessing, in theory they would not have access to that second factor.

Why Are People Not Using It?

So why do people not enable the second factor on their Google accounts? Unfortunately, the presentation didn’t appear to explain that, which makes sense since it is difficult to know why people do or do not do certain things. I think there may be a few reasons for it that we will briefly touch on.

First, I think many people just are not aware of enabling the second factor. To be fair, it is sort of buried down in settings that may be difficult to find if you are not really looking for it. If it is not front and center, then there is a much smaller chance people will go seeking it out. To add to the issue, many people really don’t understand what 2-factor authentication means or how it really helps them. Sure, in security we get it, but that doesn’t mean everyone else does. How do we make it more prominent that this is a positive security feature? Many users will already be aware of 2-factor if they use online banking as most of those have started enforcing it.

Many people think that two factor authentication is a burden or it will slow their access down. This is usually not the case unless the application has implemented it poorly. Many sites will allow you to save your computer so you don’t need to enter the 2nd factor every time you access the site. However, it will require it if you access from a different computer.

To complicate things, other applications may not support signing in with 2 factors, like your email client. In these cases, you have to generate an app password which can be very confusing to many users, especially those that are not technically savvy.

There may be a chance that users don’t think they need to protect their email accounts, that it is not sensitive. If you just use email to communicate with friends and receive junk mail, what could be so bad, right? Most people forget that things like password resets are performed using an email account. Having control of an email account provides a lot of control over a lot of things. While it may seem small, email is an important function to protect.

If you are using Gmail, I recommend configuring 2-factor authentication. The following video walks through setting it up using SMS (Although there are other options as well):

Demo- Google 2 factor

If you are developing applications, I recommend looking into providing the option of 2-factor authentication. When you do this, make sure that you are promoting its use in a positive way. If you already have 2-factor with your application, can you run a report to determine what percentage of users are actually using it? If that number is low, what steps can you take to increase them?

Don’t assume that any application is not worthy of the extra security. Many applications are already providing 2-factor and that number will just increase. While we still have the password, we will always be looking for ways to add more protection. When implemented properly, it is simple for the end user, but effective in increasing security. If your user base is not taking advantage of the option, take the time to assess why that is and how it can be improved.

As I was writing this up, I ran into an interesting situation with 2-factor that sparked some more thoughts. When looking to support 2-factor authentication and not using SMS, take careful consideration to the applications you may choose to support. On the Apple App Store alone there are over 200 different authenticator apps available. Some are interchangeable while others are not. This can be another barrier in users choosing to enable 2-factor authentication.

Tinder Mobile Take-Aways

While browsing through the news I noticed an article talking about the Tinder mobile app and a privacy concern. You can read the article at https://www.consumerreports.org/privacy/tinder-app-security-flaws-put-users-privacy-at-risk/. To summarize what is considered the issue is that the mobile application does not transmit the photos that you see using HTTPS. This means that anyone on the same connection can see the traffic and, ultimately, see the photos you are presented. From my understanding, it doesn’t appear the potential attacker can tell who the user is that is viewing these photos as the rest of the traffic is properly using HTTPS.

We have discussed the move to all HTTPS multiple times on this blog and we are seeing a lot of sites making the switch. With web applications it is easy to see if the site is using HTTPS or not with the indicators near the address bar. Of course, these indicators are often confusing to most, but at least we have the ability to see the status. With a mobile application it is much more difficult to tell if data is transmitted using HTTPS or not because there is no visible indicator. Instead, one needs to view the raw traffic or use a web proxy to see how the data is transmitted. This can be misleading to many people because the assumption is that the data is protected because it is hidden under more layers.

In this instance, the ability to see these photos may not be considered that sensitive by many. Assuming that anyone can create an account and see the photos doesn’t make them a secret. People have opted to post their images for others to find them on the network. Of course, level of sensitivity is in the eye of the beholder these days. Another issue that is potentially possible in this situation is that the attacker could manipulate that image traffic to show a different image. This could lead to the end user seeing a different image than the one expected. The usefulness of this could be called into question at any type of large scale.

The take-away here is that when we are building applications we must take care in understanding how we are transmitting all of our data to determine what needs to be protected. As I mentioned, there is already a push to make everything HTTPS all the time. If you have decided not to use HTTPS for your connections, have you documented the reasons? What does your threat model tell you about the risks with that data and its communication. How does that risk line up with your acceptance procedures.

Another interesting tidbit came out of the article mentioned above. In addition to seeing the actual photos, they found it was possible to identify whether or not the end user liked or disliked the photo by comparing the network traffic. The interesting part about this part is that those decisions were encrypted when transmitted. The key point here is that the traffic for each decision was a set size and the sizes were different for like and dislike. By viewing the traffic after seeing a photo, it is possible to determine which ones were liked based on the size of the requests. In this case, it still doesn’t identify the end user that is using the application.

We don’t typically spend a lot of time analyzing the size of the requests we send in the event someone may try to determine what actions we are taking over an encrypted channel. Most of the time these actions are not possible to determine, or the level of effort is way above what is realistic. The easy solution would be to make sure all traffic was encrypted and we wouldn’t be able to know what images were liked or disliked. Maybe it would be possible to still see the difference, but with no way to tie it to specific images. The other option is to attempt to pad the requests so that they are all the same size. This would be for highly sensitive systems as the complexity may not be worth the benefit.

Of course, all of this is based on the attacker being on the same network as the end user so they can intercept or view the traffic in the first place. In the case of a public place, it might just be easier to hover over your shoulder and watch you use the app then intercept the traffic and guess at who is using it.

Both of these topics are good conversation starters within your organization. They help us realize that even just one request that doesn’t use HTTPS may be seen and could raise an issue. It also helps us to see that sometimes even encrypted data can be determined, but that doesn’t mean it is a high risk. Each situation is different and should be properly analyzed to determine the risk it creates for the company and the organization.

New Year’s Resolutions

Here we are, the start of another year. As we reflect on 2017, this is where we really start to focus on what lies ahead in 2018. The new year is always interesting because it usually doesn’t affect our build cycles or releases. With the exception of accounting for vacations. Yet, this is the time of year where many people get re-focused and motivated to change old habits or try something new.

Listen to the Podcast:

As I look back on 2017, there were a lot of news headlines that focused around security. So many of them highlighting breaches, many termed “mega” breaches. The trend of hyped up headlines glorifying monster breaches will likely continue through 2018 and beyond. We know that breaches can, or will, happen. We have seen examples of different techniques used to gain unauthorized access to data. This won’t change, and will most likely become more prevalent going forward. The amount of information available to potential attackers is enormous, making our job of application security that much more important.

One of the biggest lessons to take away from 2017 is that privacy is important. In addition, private data is not limited to PCI or HIPAA. All sorts of data can be considered private and require the custodian to take proper steps to protect it. It doesn’t matter if the data is held by a Fortune 500 company or a one-person shop. To someone, that data is worth something. As we look into 2018, this reminds us that we must understand what data we have. We must know what type of regulations it may fall under, what applications contain it, and how we are protecting it. Just because data may not fall under a regulation doesn’t mean it should be overlooked. In the end, it is the expectation of our customers and clients that we will handle their data responsibly.

Protecting this data is not about how much money you spend or what tools you buy. Every organization is different. Every application development team is different. I encourage everyone to take the time to research and understand what your team needs to be successful. As in the past, throughout the year I will be posting thoughts on different application security topics. If you have any questions or topics, feel free to share them with me. Looking for someone to talk to about application security? Reach out. I have services available to help organizations and individuals reach new heights and solve problems.

What are your New Year’s Resolutions when it comes to application security?

XSS in a Script Tag

Cross-site scripting is a pretty common vulnerability, even with many of the new advances in UI frameworks. One of the first things we mention when discussing the vulnerability is to understand the context. Is it HTML, Attribute, JavaScript, etc.? This understanding helps us better understand the types of characters that can be used to expose the vulnerability.

In this post, I want to take a quick look at placing data within a <script> tag. In particular, I want to look at how embedded <script> tags are processed. Let’s use a simple web page as our example.

<html>
	<head>
	</head>
	<body>
	<script>
		var x = "<a href=test.html>test</a>";
	</script>
	</body>
</html>

The above example works as we expect. When you load the page, nothing is displayed. The link tag embedded in the variable is rated as a string, not parsed as a link tag. What happens, though, when we embed a <script> tag?

<html>
	<head>
	</head>
	<body>
	<script>
		var x = "<script>alert(9)</script>";
	</script>
	</body>
</html>

In the above snippet, actually nothing happens on the screen. Meaning that the alert box does not actually trigger. This often misleads people into thinking the code is not vulnerable to cross-site scripting. if the link tag is not processes, why would the script tag be. In many situations, the understanding is that we need to break out of the (“) delimiter to start writing our own JavaScript commands. For example, if I submitted a payload of (test”;alert(9);t = “). This type of payload would break out of the x variable and add new JavaScript commands. Of course, this doesn’t work if the (“) character is properly encoded to not allow breaking out.

Going back to our previous example, we may have overlooked something very simple. It wasn’t that the script wasn’t executing because it wasn’t being parsed. Instead, it wasn’t executing because our JavaScript was bad. Our issue was that we were attempting to open a <script> within a <script>. What if we modify our value to the following:

<html>
	<head>
	</head>
	<body>
	<script>
		var x = "</script><script>alert(9)</script>";
	</script>
	</body>
</html>

In the above code, we are first closing out the original <script> tag and then we are starting a new one. This removes the embedded nuance and when the page is loaded, the alert box will appear.

This technique works in many places where a user can control the text returned within the <script> element. Of course, the important remediation step is to make sure that data is properly encoded when returned to the browser. By default, Content Security Policy may not be an immediate solution since this situation would indicate that inline scripts are allowed. However, if you are limiting the use of inline scripts to ones with a registered nonce would help prevent this technique. This reference shows setting the nonce (https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/Content-Security-Policy/script-src).

When testing our applications, it is important to focus on the lack of output encoding and less on the ability to fully exploit a situation. Our secure coding standards should identify the types of encoding that should be applied to outputs. If the encodings are not properly implemented then we are citing a violation of our standards.

JavaScript in an HREF or SRC Attribute

The anchor (<a>) HTML tag is commonly used to provide a clickable link for a user to navigate to another page. Did you know it is also possible to set the HREF attribute to execute JavaScript. A common technique is to use the onclick event of the anchor tab to execute a JavaScript method when the user clicks the link. However, to stop the browser from actually redirecting the HREF can be set to javascript:void(0);. This cancels the HREF functionality and allows the JavaScript from the onclick to execute as expected.

In the above example, notice that the HREF is set with a value starting with “javascript:”. This identifier tells the browser to execute the code following that prefix. For those that are security savvy, you might be thinking about cross-site scripting when you hear about executing JavaScript within the browser. For those of you that are new to security, cross-site scripting refers to the ability for an attacker to execute unintended JavaScript in the context of your application (https://www.owasp.org/index.php/Cross-site_Scripting_(XSS)).

I want to walk through a simple scenario of where this could be abused. In this scenario, the application will attempt to track the page the user came from to set up where the Cancel button will redirect to. Imagine you have a list page that allows you to view details of a specific item. When you click the item it takes you to that item page and passes a BackUrl in the query string. So the link may look like:

https://developsec.com/item.php?backUrl=/items.php

On the page, there is a hyperlink created that sets the HREF to the backUrl property, like below:

<a href=”<?php echo $_GET[“backUrl”];?>”>Back</a>

When the page executes as expected you should get an output like this:

<a href=”/items.php”>Back</a>

There is a big problem though. The application is not performing any type of output encoding to protect against cross-site scripting. If we instead pass in backUrl=”%20onclick=”alert(10); we will get the following output:

<a href=”” onclick=”alert(10);“>Back</a>

In the instance above, we have successfully inserted the onclick event by breaking out of the HREF attribute. The bold section identifies the malicious string we added. When this link is clicked it will prompt an alert box with the number 10.

To remedy this, we could (or typically) use output encoding to block the escape from the HREF attribute. For example, if we can escape the double quotes (” -> &quot; then we cannot get out of the HREF attribute. We can do this (in PHP as an example) using htmlentities() like this:

<a href=”<?php echo htmlentities($_GET[“backUrl”],ENT_QUOTES);?>”>Back</a>

When the value is rendered the quotes will be escapes like the following:

<a href=”&quot; onclick=&"alert(10);“>Back</a>

Notice in this example, the HREF actually has the entire input (in bold), rather than an onclick event actually being added. When the user clicks the link it will try to go to https://www.developsec.com/” onclick=”alert(10); rather than execute the JavaScript.

But Wait… JavaScript

It looks like we have solved the XSS problem, but there is a piece still missing. Remember at the beginning of the post how we mentioned the HREF supports the javascript: prefix? That will allow us to bypass the current encodings we have performed. This is because with using the javascript: prefix, we are not trying to break out of the HREF attribute. We don’t need to break out of the double quotes to create another attribute. This time we will set backUrl=javascript:alert(11); and we can see how it looks in the response:

<a href=”javascript:alert(11);“>Back</a>

When the user clicks on the link, the alert will trigger and display on the page. We have successfully bypassed the XSS protection initially put in place.

Mitigating the Issue

There are a few steps we can take to mitigate this issue. Each has its pros and many can be used in conjunction with each other. Pick the options that work best for your environment.

  • URL Encoding – Since the HREF is meant to be a URL, you could perform URL encoding. URL encoding will render the javascript benign in the above instances because the colon (:) will get encoded. You should be using URL encoding for URLs anyway, right?
  • Implement Content Security Policy (CSP) – CSP can help limit the ability for inline scripts to be executed. In this case, it is an inline script so something as simple as ‘Content-Security-Policy:default-src ‘self’ could be sufficient. Of course, implementing CSP requires research and great care to get it right for your application.
  • Validate the URL – It is a good idea to validate that the URL used is well formed and pointing to a relative path. If the system is unable to parse the URL then it should not be used and a default back URL can be substituted.
  • URL White Listing – Creating a white list of valid URLs for the back link can be effective at limiting what input is used by the end user. This can cut down on the values that are actually returned blocking any malicious scripts.
  • Remove javascript: – This really isn’t recommended as different encodings can make it difficult to effectively remove the string. The other techniques listed above are much more effective.

The above list is not exhaustive, but does give an idea of ways to help reduce the risk of JavaScript within the HREF attribute of a hyper link.

Iframe SRC

It is important to note that this situation also applies to the IFRAME SRC attribute. it is possible to set the SRC of an IFRAME using the javascript: notation. In doing so, the javascript executes when the page is loaded.

Wrap Up

When developing applications, make sure you take this use case into consideration if you are taking URLs from user supplied input and setting that in an anchor tag or IFrame SRC.

If you are responsible for testing applications, take note when you identify URLs in the parameters. Investigate where that data is used. If you see it is used in an anchor tag, look to see if it is possible to insert JavaScript in this manner.

For those performing static analysis or code review, look for areas where the HREF or SRC attributes are set with untrusted data and make sure proper encoding has been applied. This is less of a concern if the base path of the URL has been hard-coded and the untrusted input only makes up parameters of the URL. These should still be properly encoded.

Understanding Your Application Platform

Building applications today includes the use of some pretty impressive platforms. These platforms have so much built in capability, many of the most common tasks are easily accomplished through simple method calls. As developers, we rely on these frameworks to provide a certain level of functionality. Much of which we may never even use.

When it comes to security, the platform can be a love/hate relationship. On the one hand, developers may have little control over how the platform handles certain tasks. On the other, the platform may provide excellent security controls. As we mature these platforms, we see a lot of new, cool security features enabled by default. Many view engines have cross-site scripting protections built in by default. Many of the systems use ORM to help reduce SQL Injection vulnerabilities. The problem we often run into is we don’t really know what our platform does and does not provide.

A question was posed about what was the most secure application platform, or which would you recommend. The problem to that question is that the answer really is “It depends.” The frameworks are not all created equally. Some have better XSS preventions. Others may have default CSRF prevention. What a framework does or doesn’t have can also change next month. They are always being updated. Rather than pick the most secure platform, I recommend to people to take the time to understand your platform of choice.

Does it matter if you use PHP, Java, .Net, Python, or Ruby? They all have some built in features, they all have their downfalls. So rather than trying to swap platforms every time a new one gets better features, spend some time to understand the platform you have in front of you. When you understand the risks that you face, you can then determine how those line up with your platform. Do you output user input to a web browser? If so, cross site scripting is a concern. Determine how your platform handles that. It may be that the platform auto encodes that data for you. The encoding may only happen in certain contexts. it may be the platform doesn’t provide any encoding, but rather leaves that up to you.

While the secure by default is more secure, as it reduces the risk of human oversight, applications can still be very secure without it. This is where that understanding comes into play. If I understand my platform and know that it doesn’t encode for me then I must make the effort to protect that. This may even include creating your own function or library that is used enterprise wide to help solve the problem. Unfortunately, when you don’t understand your platform, you never realize that this is a step you must take. Then it is overlooked and you are vulnerable.

I am also seeing more platforms starting to provide security guidelines or checklists to help developers with secure implementation. They may know of areas the platform doesn’t create a protection, so they show how to get around that. Maybe something is not enabled by default, but they recommend and show how to enable that. The more content like this that is produced the more we will understand how to securely create applications.

Whatever platform you use, understanding it will make the most difference. If the platform doesn’t have good documentation, push for it. Ask around or even do the analysis yourself to understand how security works in your situations.

MySpace Account Takeover – Take-aways

Have you ever forgotten your password, or lost access to your accounts? I know I have. The process of getting your access back can range from very easy to quite difficult. In one case, I had an account that required that a pin code be physically mailed to me in 7-10 days. Of course, this was a financial account that required extra protections.

I came across this article (https://www.wired.com/story/myspace-security-account-takeover/) that identified that MySpace’s process for regaining access to an account was easily by-passable with just a few pieces of information that is commonly easily found.

According to the issue reported, the following details were needed to gain control of an account:

  • Account Holder’s Name
  • UserName
  • Date of Birth

In addition, the email address was part of the form, but was not actually validated.

With the amount of information available on social media platforms, and even though past breaches, it is very difficult to come up with good questions to help validate a user is who they say they are. In this case, the first 2 pieces of information are reportedly available on a user’s account page and are not difficult to find. It is critical that we give great consideration to the way we attempt to validate a user’s identity.

Security or secret questions have been known to be weak for a long time. While they can help provide some assistance in identifying the user, they should not be used alone. One recommendation is to include a side channel for some verification. For example, requiring a user to receive an email at the address on record before answering these questions helps reduce some of the risk. This method now requires that an attacker gain access to the user’s email account to receive a temporary account reset link. Although this is possible, it adds another hurdle to help protect the user’s account.

Similarly, you could look to send a code to a phone number that is currently on record. Like email, it would then require the interception of the phone message, adding difficulty to the process. One of the points here is that you only want to send to email or phone that is already on record. Never allow the user to provide a new email address or phone number as they could add an untrusted phone number and easily take over the account.

There was another piece of this that should be considered, the unvalidated email address. The request for the email address in this reset process may have been meant as another verification of account information. However, as commonly seen, the value wasn’t actually validated on the server. This is an easy oversight to make when working with web forms.

Remember that validation should always be performed on the server. This is because it is simple to bypass client-side validation with the use of simple add-ins or a web proxy.

Validation of data fields should be a standard part of QA testing. Ensure that when testing forms, such as the account recovery, that the form reacts accordingly with many types of inputs. This includes what happens if a field is not provided or if it does not line up with the expected answer. This same situation has been seen on other sites where the current password may be required to update the password, but is not actually verified.

These types of stories help remind us to review these types of functions and features within our application. For older applications, it may have been a while since we have touched these features and how they were designed initially is no longer recommended. Take a moment to look back at your account recovery process to make sure it is up to par with your current requirements.

Looking for some help with your application security program? Just want someone to talk to about these types of topics? Reach out to us. We are here to help and offer a wide range of services to help make your day easier. Reach out to james@developsec.com for more information.

Validation: Client vs. Server

Years ago, I remember being on a technical interview phone call for a senior developer position. What stood out was when the interviewer asked me about performing input validation. The question was in regards to if validation should be on the client or the server. My answer: The server.

What took me by surprise was when the response was that my answer was incorrect. In fact, I was told that Microsoft recommends performing validation on the client. This was inaccurate information, but I let it go and continued with the interview.
Recently, I have been having more conversations around input validation. In particular, the question of client or server side. While it is easy to state that validation should always be performed on the server, lets dig into this a little more to better understand your situation.

From a pure security perspective, input validation must be performed on the server. There is one simple reason for this: Any protections built using client-side techniques can be bypassed by using a simple web proxy. Using JavaScript to enforce that a field contains an email address can be easily bypassed by intercepting the request and changing it after the JavaScript has executed.

If you look at the threat model of your application, requests from the client to the server cross a trust boundary. Because of this trust boundary we know that the data needs to be validated again. Why? There is no way to know what happened to the data before it was received. We can assume the request was sent from a browser, used by a typical user. However, we don’t know if the data was manipulated after leaving the browser, or even sent from a browser at all.

That, however, is from a strict security standpoint. We must not forget that client-side validation serves a purpose as well. While client-side validation may not be trusted by the server, it tends to be more focused on immediate feedback to the user. Not only does this save a round trip, or many round trips, to the server, it cuts down on the processing the server needs to handle.
If we take an example of purely validating required fields on a form, we can immediately see the benefit of client-side validation. Even a small form, if not complete can create a lot of inefficiency if the user is constantly posting it without all the required fields. The ability to alert to this on the client makes it much quicker and cuts down on the number of invalid requests to the server.

Of course, this doesn’t mean that the user can’t fill in all the required fields to pass the client-side validation, intercept the request, and then remove some of those fields. In this case, server-side validation would catch this. The goal, however, of client-side validation is to provide a reactive user interface that is fast.

Understanding how each validation location functions and what the real purpose is helps us identify when to use each. While server-side validation is always required, client-side validation can be a great addition to the application.

Properly Placing XSS Output Encoding

Cross-Site Scripting flaws, as well as other injection flaws, are pretty well understood. We know how they work and how to mitigate them. One of the key factors in mitigation of these flaws is output encoding or escaping. For SQL, we escape by using parameters. For cross-site scripting we use context sensitive output encoding.

In this post, I don’t want to focus on the how of output encoding for cross-site scripting. Rather, I want to focus on when in the pipeline it should be done. Over the years I have had a lot of people ask if it is ok to encode the data before storing it in the database. I recently came across this insufficient solution and thought it was a good time to address it again. I will look at a few different cases that indicate why the solution is insufficient and explain a more sufficient approach.

The Database Is Not Trusted

The database should not be considered a trusted resource. This is a common oversight in many organizations, most likely due to the fact that the database is internal. It lives in a protected portion of your production environment. While that is true, it has many sources that have access to it. Even if it is just your application that uses the database today, there are still administrators and update scripts that most likely access it. We can’t rule out the idea of a rogue administrator, even if it is very slim.

We also may not know if other applications access our database. What if there is a mobile application that has access? We can’t guarantee that every source of data is going to properly encode the data before it gets sent to the database. This leaves us with a gap in coverage and a potential for cross-site scripting.

Input Validation My Not Be Enough

I always recommend to developers to have good input validation. Of course, the term good is different for everyone. At the basic level, your input validation may limit a few characters. At the advanced level it may try to limit different types of attacks. In some cases, it is difficult to use input validation to eliminate all cross-site scripting payloads. Why? Due to the different contexts and the types of data that some forms accept, a payload may still squeak by. Are you protecting against all payloads across all the different contexts? How do you know what context the data will be used in?

In addition to potentially missing a payload, what about when another developer creates a function and forgets to include the input validation? Even more likely, what happens when a new application starts accessing your database and doesn’t perform the same input validation. it puts us back into the same scenario described in the section above regarding the database isn’t trusted.

You Don’t Know the Context

When receiving and storing data, the chances are good I don’t know where the data will be used. What is the context of the data when output? Is it going into a span tag, an attribute or even straight into JavaScript? Will it be used by a reporting engine? The context matters because it determines how we encode the data. Different contexts are concerned with different characters. Can I just throw data into the database with an html encoding context? At this point, I am transforming the data at a time where there is no transformation required. Cross-site scripting doesn’t execute in my SQL column.

A Better Approach

As you can see, the above techniques are useful, however, they appear insufficient for multiple reasons. I recommend performing the output encoding immediately before the data is actually used. By that, I mean to encode right before it is output to the client. This way it is very clear what the context is and the appropriate encoding can be implemented.
Don’t forget to perform input validation on your data, but remember it is typically not meant to stop all attack scenarios. Instead it is there to help reduce them. We must be aware that other applications may access our data and those applications may no follow the same procedures we do. Due to this, making sure we make encoding decisions at the last moment provides the best coverage. Of course, this relies on the developers remembering to perform the output encoding.

Sub Resource Integrity – SRI

Do you rely on content distribution networks or CDNs to provide some of your resources? You may not consider some of your resources in this category, but really it is any resource that is provided outside of your server. For example, maybe you pull in the jQuery JavaScript file from ajax.googleapis.com rather than hosting the file on your server.
These CDNs provide a great way to give fast access to these resources. But how do you know you are getting the file you expect?

As an attacker, if I can attack multiple people vs just one, it is a better chance of success. A CDN provides a central location to potentially affect many applications, vs. targeting just one. Would you know if the CDN has modified that file you are expecting?

Welcome Sub Resource Integrity, or SRI. SRI provides the ability to validate the signature of the file against a predetermined hash. It is common for websites that provide files for downloads to provide a hash to validate the file is not corrupt. After downloading the file, you would compute the hash using the same algorithm (typically MD5) and then compare it to the hash listed on the server.

SRI works in a similar way. To implement this, as a developer you create a hash of the expected resource using a specified hashing algorithm. Then, you would add an integrity attribute to your resource, whether it is a script element or stylesheet. When the browser requests the resource, it will compute the hash, compare it to the integrity attribute and if successful, will load the resource. if it is unsuccessful, the file will not be loaded.

How it works

Lets look at how we would implement this for jQuery hosted at google. We will be including the reference from https://ajax.googleapis.com/ajax/libs/jquery/3.2.1/jquery.min.js

Initially, we might start by just creating a script tag with that as the source. This will work, but doesn’t provide any integrity check. There are a few different ways we can create the digest. An easy way is to use https://www.srihash.org/. The site provides a way to enter in the url to the resource and it will create the script tag for you.

Another option is to generate the hash yourself. To do this you will start by downloading the resource to your local system.

Once the file is downloaded, you can generate the hash by executing the following command:


openssl dgst -sha384 -binary Downloads/jquery.min.js | openssl base64 -A

Make sure you change Downloads/jquery.min.js to your downloaded file path. You should see a hash similar to:

xBuQ/xzmlsLoJpyjoggmTEz8OWUFM0/RC5BsqQBDX2v5cMvDHcMakNTNrHIW2I5f

Now, we can build our script tag as follows (Don’t forget to add the hashing algorithm to the integrity attribute:

<script src=”https://ajax.googleapis.com/ajax/libs/jquery/3.2.1/jquery.min.js” integrity=”sha384-xBuQ/xzmlsLoJpyjoggmTEz8OWUFM0/RC5BsqQBDX2v5cMvDHcMakNTNrHIW2I5f” crossorigin=”anonymous”></script>

Notice that there is a new crossorigin attribute as well. This is set to anonymous to allow CORS to work correctly. The CDN must have CORS set up to allow the integrity check to occur.

If you want to test the integrity check out, add another script tag to the page (after the above tag) that looks like the following:

<script>alert(window.jQuery);</script>

When the page loads, it should alert with some jQuery information. Now modify the Integrity value (I removed the last character) and reload the page. You should see a message that says “undefined”. This means that the resource was not loaded.

Browser support is still not complete. At this time, only Chrome, Opera, and Firefox support this feature.

Handling Failures

What do you do if the integrity check fails? You don’t want to break your site, right? Using the code snippet we tested with above, we could check to make sure it loaded, and if not, load it from a local resource. This gives us the benefit of using the CDN most of the time and falling back to a local resource only when necessary. The following may be what the updated script looks like:

<script src=”https://ajax.googleapis.com/ajax/libs/jquery/3.2.1/jquery.min.js” integrity=”sha384-xBuQ/xzmlsLoJpyjoggmTEz8OWUFM0/RC5BsqQBDX2v5cMvDHcMakNTNrHIW2I5f” crossorigin=”anonymous”></script>
<script> window.jQuery || document.write(‘<script src=”/jquery-.min.js”><\/script>’)</script>

When the integtrity check fails, you can see the local resource being loaded in the image below:

SRI-1

If you are using resources hosted on external networks, give some thought about implementing SRI and how it may benefit you. It is still in its early stages and not supported by all browsers, but it can certainly help reduce some of the risk of malicious files delivered through these networks.

Jardine Software helps companies get more value from their application security programs. Let’s talk about how we can help you.

James Jardine is the CEO and Principal Consultant at Jardine Software Inc. He has over 15 years of combined development and security experience. If you are interested in learning more about Jardine Software, you can reach him at james@jardinesoftware.com or @jardinesoftware on twitter.