ThinkPHP6.0.13 Deserialization Vulnerability Analysis
I’ve been a little idle recently, and I feel uncomfortable if I don’t find something to do. I plan to find some loopholes to analyze, so I plan to take a look at some loopholes in TP. ThinkPHP6.0.13 is the latest version of TP. In August, a master submitted one The issue pointed out that TP has a deserialization problem. Some experts on the Internet have analyzed it, but there are many breakpoints, and some methods do not clarify their uses, so I also tried to analyze it in detail. Below is the POC
analysis
First look at the starting point of the POC
We found that the starting point is in the Psr6Cache class. We entered this class, but we did not find common deserialization starting magic such as __destruct or __wakeup. Method, it is speculated that it should be in the abstract class of its parent class AbstractCache. Follow the AbstractCache class
as shown in the figure, and successfully find the starting class of this deserialization chain. Here we can control the autosave attribute to false to enter the save method.
Go back to the Psr6Cache class to view this method
You can find that we can control both the pool attribute and the key attribute. Therefore, there may be two routes to call methods with the same name (getItem) of different classes. Or try to trigger the __call method directly. Let's take a look at how the POC author allows deserialization to proceed.
The author passed in exp using the constructor method, and exp is actually instantiating the Channel class. We enter the Channel class to see
There is a __call method in the Channel class, so the author chooses to trigger __call to continue the chain. This call method accepts two parameters, the method is hard-coded (getItem), and the parameters are controllable (that is, the previously controllable key attributes)
Follow the log View the method. It accepts three parameters (but it is actually useless for subsequent chains). Pass in the record method
followed by the record method
Go back and check the author's POC and find that its control lazy attribute is false, let the function enter the last if branch to execute the save method
Then The save method should be a more critical method. Following the save method, there are three points that may be exploited. Which one did the author choose?
According to the POC, it is not difficult to find that the author chose to control the logger attribute, use the constructor to assign a value to it, and make it an object of the Socket class
In this class we find a complex method of the same name with a large number of operations.
# Let's continue to see how the author constructs it. The author controls the config attribute and assigns it an array value. The array has the following content
The key lies in these two key values. The author controls the config and allows the program to run to the branch that calls the invoke method
At the same time, the app attribute is controllable. The author makes the app attribute an object of the App class. Let’s enter the App class.
Let’s first look at the exists of the App class. In the case of a method, this method was found in its parent class
Moving on, here is the only operation performed on the App class, which controls the value of the instances attribute. The purpose of controlling its value here is to enter the Request class and execute the url method
The author makes the only manipulation of the Request class here, It is to control the value of the url attribute. It can be seen that if the url attribute exists, then the first branch will be entered, and its value is equal to itself.
At the same time, we noticed that the complete value we passed in before was true. Therefore, the final returned result is $this->domain().$url. We have controlled the url, so what does the domain method return?
OK, we don’t need to look at this. After analyzing so much, we got the final value of $currentUri, which is:
http://localhost/
currentUri is passed in as an array Invoked, according to the length of the chain, reaching invoke, our deserialization journey is almost over
Check invoke, the App class cannot find this method, in other This method was found in the parent class
You can see it here. There are three branches in this function, so where will it eventually go? According to what we passed in $config['format_head'] before, first of all, the object we passed in is not an instance or subclass of Closure, and it does not meet the conditions of the second branch
So enter the third branch. We follow up with the invokeMethod() method. The $callab passed in here is [new \think\view\driver\Php,'display'], and $vars is ['http://localhost/']
Note that the $method we passed in is an array, so we enter the first branch. Assign new \think\view\driver\Php (i.e. object) to $class, and assign 'display' (i.e. method name) to new $method.
Then a judgment is made below. If $class is an object, then its value is itself. Because we are passing in an object, there is no change here. Then enter the most critical code
You can see that the object new \think\view\driver\Php and the method display are passed into ReflectionMethod.
At the end, call the invokeArgs method, passing in the new \think\view\driver\Php object, and passing in $args
So what are args?
After we followed up, we found out that it is a processing function. Because I am lazy and have almost finished the analysis at this point, I will not read it hard and give the conclusion directly. In short The key parts of the $vars we passed in, that is, ['http://localhost/'], are retained and entered into subsequent parameters.
Looking further, this function (invokeArgs) can be simply compared to call_user_func(), so the final key code is actually only these two lines
, that is,
$reflect = new ReflectionMethod(new \think\view\driver\Php,’display’); return $reflect->invokeArgs(new \think\view\driver\Php,’ ’)
Friends who often watch TP deserialization will know that it is over! After all, the display method is called. But what exactly is the above operation of calling the ReflectionMethod class? We can demonstrate this with the help of the following example. So this thing is very similar to call_user_func
The last is the display method, there is nothing more to say, the content is passed into the display method, and eval executes the command
Conclusion
The chain of TP is as interesting (and complex) as ever, especially the usage of the last ReflectionMethod class. If you don’t understand this class and the combination of methods in the class to achieve functions similar to the call_user_func function, it is easy to miss this. A wonderful exploit.
[Related tutorial recommendations: thinkphp framework]
The above is the detailed content of ThinkPHP6.0.13 Deserialization Vulnerability Analysis. For more information, please follow other related articles on the PHP Chinese website!

Hot AI Tools

Undress AI Tool
Undress images for free

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Clothoff.io
AI clothes remover

Video Face Swap
Swap faces in any video effortlessly with our completely free AI face swap tool!

Hot Article

Hot Tools

Notepad++7.3.1
Easy-to-use and free code editor

SublimeText3 Chinese version
Chinese version, very easy to use

Zend Studio 13.0.1
Powerful PHP integrated development environment

Dreamweaver CS6
Visual web development tools

SublimeText3 Mac version
God-level code editing software (SublimeText3)

Hot Topics

The core method of building social sharing functions in PHP is to dynamically generate sharing links that meet the requirements of each platform. 1. First get the current page or specified URL and article information; 2. Use urlencode to encode the parameters; 3. Splice and generate sharing links according to the protocols of each platform; 4. Display links on the front end for users to click and share; 5. Dynamically generate OG tags on the page to optimize sharing content display; 6. Be sure to escape user input to prevent XSS attacks. This method does not require complex authentication, has low maintenance costs, and is suitable for most content sharing needs.

To realize text error correction and syntax optimization with AI, you need to follow the following steps: 1. Select a suitable AI model or API, such as Baidu, Tencent API or open source NLP library; 2. Call the API through PHP's curl or Guzzle and process the return results; 3. Display error correction information in the application and allow users to choose whether to adopt it; 4. Use php-l and PHP_CodeSniffer for syntax detection and code optimization; 5. Continuously collect feedback and update the model or rules to improve the effect. When choosing AIAPI, focus on evaluating accuracy, response speed, price and support for PHP. Code optimization should follow PSR specifications, use cache reasonably, avoid circular queries, review code regularly, and use X

1. Maximizing the commercial value of the comment system requires combining native advertising precise delivery, user paid value-added services (such as uploading pictures, top-up comments), influence incentive mechanism based on comment quality, and compliance anonymous data insight monetization; 2. The audit strategy should adopt a combination of pre-audit dynamic keyword filtering and user reporting mechanisms, supplemented by comment quality rating to achieve content hierarchical exposure; 3. Anti-brushing requires the construction of multi-layer defense: reCAPTCHAv3 sensorless verification, Honeypot honeypot field recognition robot, IP and timestamp frequency limit prevents watering, and content pattern recognition marks suspicious comments, and continuously iterate to deal with attacks.

User voice input is captured and sent to the PHP backend through the MediaRecorder API of the front-end JavaScript; 2. PHP saves the audio as a temporary file and calls STTAPI (such as Google or Baidu voice recognition) to convert it into text; 3. PHP sends the text to an AI service (such as OpenAIGPT) to obtain intelligent reply; 4. PHP then calls TTSAPI (such as Baidu or Google voice synthesis) to convert the reply to a voice file; 5. PHP streams the voice file back to the front-end to play, completing interaction. The entire process is dominated by PHP to ensure seamless connection between all links.

PHP does not directly perform AI image processing, but integrates through APIs, because it is good at web development rather than computing-intensive tasks. API integration can achieve professional division of labor, reduce costs, and improve efficiency; 2. Integrating key technologies include using Guzzle or cURL to send HTTP requests, JSON data encoding and decoding, API key security authentication, asynchronous queue processing time-consuming tasks, robust error handling and retry mechanism, image storage and display; 3. Common challenges include API cost out of control, uncontrollable generation results, poor user experience, security risks and difficult data management. The response strategies are setting user quotas and caches, providing propt guidance and multi-picture selection, asynchronous notifications and progress prompts, key environment variable storage and content audit, and cloud storage.

PHP ensures inventory deduction atomicity through database transactions and FORUPDATE row locks to prevent high concurrent overselling; 2. Multi-platform inventory consistency depends on centralized management and event-driven synchronization, combining API/Webhook notifications and message queues to ensure reliable data transmission; 3. The alarm mechanism should set low inventory, zero/negative inventory, unsalable sales, replenishment cycles and abnormal fluctuations strategies in different scenarios, and select DingTalk, SMS or Email Responsible Persons according to the urgency, and the alarm information must be complete and clear to achieve business adaptation and rapid response.

PHP provides an input basis for AI models by collecting user data (such as browsing history, geographical location) and pre-processing; 2. Use curl or gRPC to connect with AI models to obtain click-through rate and conversion rate prediction results; 3. Dynamically adjust advertising display frequency, target population and other strategies based on predictions; 4. Test different advertising variants through A/B and record data, and combine statistical analysis to optimize the effect; 5. Use PHP to monitor traffic sources and user behaviors and integrate with third-party APIs such as GoogleAds to achieve automated delivery and continuous feedback optimization, ultimately improving CTR and CVR and reducing CPC, and fully implementing the closed loop of AI-driven advertising system.

PHPisstillrelevantinmodernenterpriseenvironments.1.ModernPHP(7.xand8.x)offersperformancegains,stricttyping,JITcompilation,andmodernsyntax,makingitsuitableforlarge-scaleapplications.2.PHPintegrateseffectivelyinhybridarchitectures,servingasanAPIgateway
