CVE’s

Opencast before 8.1 and 7.6 allows almost arbitrary identifiers for media packages and elements to be used. This can be problematic for operation and security since such identifiers are sometimes used for file system operations which may lead to an attacker being able to escape working directories and write files to other locations. In addition, Opencast's Id.toString(…) vs Id.compact(…) behavior, the latter trying to mitigate some of the file system problems, can cause errors due to identifier mismatch since an identifier may unintentionally change. This issue is fixed in Opencast 7.6 and 8.1.

Opencast before 8.1 stores passwords using the rather outdated and cryptographically insecure MD5 hash algorithm. Furthermore, the hashes are salted using the username instead of a random salt, causing hashes for users with the same username and password to collide which is problematic especially for popular users like the default `admin` user. This essentially means that for an attacker, it might be feasible to reconstruct a user's password given access to these hashes. Note that attackers needing access to the hashes means that they must gain access to the database in which these are stored first to be able to start cracking the passwords. The problem is addressed in Opencast 8.1 which now uses the modern and much stronger bcrypt password hashing algorithm for storing passwords. Note, that old hashes remain MD5 until the password is updated. For a list of users whose password hashes are stored using MD5, take a look at the `/user-utils/users/md5.json` REST endpoint.

Opencast before 8.1 and 7.6 allows unauthorized public access to all media and metadata by default via OAI-PMH. OAI-PMH is part of the default workflow and is activated by default, requiring active user intervention of users to protect media. This leads to users unknowingly handing out public access to events without their knowledge. The problem has been addressed in Opencast 7.6 and 8.1 where the OAI-PMH endpoint is configured to require users with `ROLE_ADMIN` by default. In addition to this, Opencast 9 removes the OAI-PMH publication from the default workflow, making the publication a conscious decision users have to make by updating their workflows.

Feedgen (python feedgen) before 0.9.0 is susceptible to XML Denial of Service attacks. The *feedgen* library allows supplying XML as content for some of the available fields. This XML will be parsed and integrated into the existing XML tree. During this process, feedgen is vulnerable to XML Denial of Service Attacks (e.g. XML Bomb). This becomes a concern in particular if feedgen is used to include content from untrused sources and if XML (including XHTML) is directly included instead of providing plain tex content only. This problem has been fixed in feedgen 0.9.0 which disallows XML entity expansion and external resources.

Cross-site scripting in SimpleSAMLphp before version 1.18.4. The www/erroreport.php script allows error reports to be submitted and sent to the system administrator. Starting with SimpleSAMLphp 1.18.0, a new SimpleSAMLUtilsEMail class was introduced to handle sending emails, implemented as a wrapper of an external dependency. This new wrapper allows us to use Twig templates in order to create the email sent with an error report. Since Twig provides automatic escaping of variables, manual escaping of the free-text field in www/errorreport.php was removed to avoid double escaping. However, for those not using the new user interface yet, an email template is hardcoded into the class itself in plain PHP. Since no escaping is provided in this template, it is then possible to inject HTML inside the template by manually crafting the contents of the free-text field.

Log injection in SimpleSAMLphp before version 1.18.4. The www/erroreport.php script, which receives error reports and sends them via email to the system administrator, did not properly sanitize the report identifier obtained from the request. This allows an attacker, under specific circumstances, to inject new log lines by manually crafting this report ID. When configured to use the file logging handler, SimpleSAMLphp will output all its logs by appending each log line to a given file. Since the reportID parameter received in a request sent to www/errorreport.php was not properly sanitized, it was possible to inject newline characters into it, effectively allowing a malicious user to inject new log lines with arbitrary content.

In Django User Sessions (django-user-sessions) before 1.7.1, the views provided allow users to terminate specific sessions. The session key is used to identify sessions, and thus included in the rendered HTML. In itself this is not a problem. However if the website has an XSS vulnerability, the session key could be extracted by the attacker and a session takeover could happen.

In PrivateBin versions 1.2.0 before 1.2.2, and 1.3.0 before 1.3.2, a persistent XSS attack is possible. Under certain conditions, a user provided attachment file name can inject HTML leading to a persistent Cross-site scripting (XSS) vulnerability. The vulnerability has been fixed in PrivateBin v1.3.2 & v1.2.2. Admins are urged to upgrade to these versions to protect the affected users.

Opencast before 7.6 and 8.1 enables a remember-me cookie based on a hash created from the username, password, and an additional system key. This means that an attacker getting access to a remember-me token for one server can get access to all servers which allow log-in using the same credentials without ever needing the credentials. This problem is fixed in Opencast 7.6 and Opencast 8.1

In uftpd before 2.11, it is possible for an unauthenticated user to perform a directory traversal attack using multiple different FTP commands and read and write to arbitrary locations on the filesystem due to the lack of a well-written chroot jail in compose_abspath(). This has been fixed in version 2.11