incomplete logout in JupyterHub

JupyterHub is an open source multi-user server for Jupyter notebooks. In affected versions users who have multiple JupyterLab tabs open in the same browser session, may see incomplete logout from the single-user server, as fresh credentials (for the single-user server only, not the Hub) reinstated after logout, if another active JupyterLab session is open while the logout takes place. Upgrade to JupyterHub 1.5. For distributed deployments, it is jupyterhub in the _user_ environment that needs patching. There are no patches necessary in the Hub environment. The only workaround is to make sure that only one JupyterLab tab is open when you log out.

Cross organization admin control in Grafana

Grafana is an open-source platform for monitoring and observability. In affected versions when the fine-grained access control beta feature is enabled and there is more than one organization in the Grafana instance admins are able to access users from other organizations. Grafana 8.0 introduced a mechanism which allowed users with the Organization Admin role to list, add, remove, and update users’ roles in other organizations in which they are not an admin. With fine-grained access control enabled, organization admins can list, add, remove and update users' roles in another organization, where they do not have organization admin role. All installations between v8.0 and v8.2.3 that have fine-grained access control beta enabled and more than one organization should be upgraded as soon as possible. If you cannot upgrade, you should turn off the fine-grained access control using a feature flag.

Incomplete validation in `tf.summary.create_file_writer`

TensorFlow is an open source platform for machine learning. In affected versions if `tf.summary.create_file_writer` is called with non-scalar arguments code crashes due to a `CHECK`-fail. The fix will be included in TensorFlow 2.7.0. We will also cherrypick this commit on TensorFlow 2.6.1, TensorFlow 2.5.2, and TensorFlow 2.4.4, as these are also affected and still in supported range.

Account takeover when having only access to a user's short lived token in wire-server

Wire-server is the backing server for the open source wire secure messaging application. In affected versions it is possible to trigger email address change of a user with only the short-lived session token in the `Authorization` header. As the short-lived token is only meant as means of authentication by the client for less critical requests to the backend, the ability to change the email address with a short-lived token constitutes a privilege escalation attack. Since the attacker can change the password after setting the email address to one that they control, changing the email address can result in an account takeover by the attacker. Short-lived tokens can be requested from the backend by Wire clients using the long lived tokens, after which the long lived tokens can be stored securely, for example on the devices key chain. The short lived tokens can then be used to authenticate the client towards the backend for frequently performed actions such as sending and receiving messages. While short-lived tokens should not be available to an attacker per-se, they are used more often and in the shape of an HTTP header, increasing the risk of exposure to an attacker relative to the long-lived tokens, which are stored and transmitted in cookies. If you are running an on-prem instance and provision all users with SCIM, you are not affected by this issue (changing email is blocked for SCIM users). SAML single-sign-on is unaffected by this issue, and behaves identically before and after this update. The reason is that the email address used as SAML NameID is stored in a different location in the databse from the one used to contact the user outside wire. Version 2021-08-16 and later provide a new end-point that requires both the long-lived client cookie and `Authorization` header. The old end-point has been removed. If you are running an on-prem instance with at least some of the users invited or provisioned via SAML SSO and you cannot update then you can block `/self/email` on nginz (or in any other proxies or firewalls you may have set up). You don't need to discriminate by verb: `/self/email` only accepts `PUT` and `DELETE`, and `DELETE` is almost never used.

Improper Restriction of XML External Entity Reference (XXE) in Nokogiri on JRuby

Nokogiri is a Rubygem providing HTML, XML, SAX, and Reader parsers with XPath and CSS selector support. In Nokogiri v1.12.4 and earlier, on JRuby only, the SAX parser resolves external entities by default. Users of Nokogiri on JRuby who parse untrusted documents using any of these classes are affected: Nokogiri::XML::SAX::Parse, Nokogiri::HTML4::SAX::Parser or its alias Nokogiri::HTML::SAX::Parser, Nokogiri::XML::SAX::PushParser, and Nokogiri::HTML4::SAX::PushParser or its alias Nokogiri::HTML::SAX::PushParser. JRuby users should upgrade to Nokogiri v1.12.5 or later to receive a patch for this issue. There are no workarounds available for v1.12.4 or earlier. CRuby users are not affected.

In Eclipse Lyo versions 1.0.0 to 4.1.0, a TransformerFactory is initialized with the defaults that do not restrict DTD loading when working with RDF/XML. This allows an attacker to cause an external DTD to be retrieved.

In Mahara before 20.04.5, 20.10.3, 21.04.2, and 21.10.0, the account associated with a web services token is vulnerable to being exploited and logged into, resulting in information disclosure (at a minimum) and often escalation of privileges.

AEM Forms Improper Restriction of XML External Entity Reference

AEM Forms Cloud Service offering, as well as version 6.5.10.0 (and below) are affected by an XML External Entity (XXE) injection vulnerability that could be abused by an attacker to achieve RCE.

XML eXternal Entity (XXE) in OBDA systems’ Mastro 1.0 allows remote attackers to read system files via custom DTDs.

SAP BusinessObjects Business Intelligence Platform (Crystal Reports) - versions 420, 430, allows an unauthenticated attacker to exploit missing XML validations at endpoints to read sensitive data. These endpoints are normally exposed over the network and successful exploitation can enable the attacker to retrieve arbitrary files from the server.

Billion Laughs

Apache OpenOffice has a dependency on expat software. Versions prior to 2.1.0 were subject to CVE-2013-0340 a "Billion Laughs" entity expansion denial of service attack and exploit via crafted XML files. ODF files consist of a set of XML files. All versions of Apache OpenOffice up to 4.1.10 are subject to this issue. expat in version 4.1.11 is patched.

A vulnerability has been identified in Teamcenter V12.4 (All versions < V12.4.0.8), Teamcenter V13.0 (All versions < V13.0.0.7), Teamcenter V13.1 (All versions < V13.1.0.5), Teamcenter V13.2 (All versions < 13.2.0.2). The application contains a XML External Entity Injection (XXE) vulnerability. This could allow an attacker to view files on the application server filesystem.

Knot Resolver before 5.3.2 is prone to an assertion failure, triggerable by a remote attacker in an edge case (NSEC3 with too many iterations used for a positive wildcard proof).

Sensitive Cookie in HTTPS Session Without 'Secure' Attribute in ledgersmb/ledgersmb

LedgerSMB does not set the 'Secure' attribute on the session authorization cookie when the client uses HTTPS and the LedgerSMB server is behind a reverse proxy. By tricking a user to use an unencrypted connection (HTTP), an attacker may be able to obtain the authentication data by capturing network traffic. LedgerSMB 1.8 and newer switched from Basic authentication to using cookie authentication with encrypted cookies. Although an attacker can't access the information inside the cookie, nor the password of the user, possession of the cookie is enough to access the application as the user from which the cookie has been obtained. In order for the attacker to obtain the cookie, first of all the server must be configured to respond to unencrypted requests, the attacker must be suitably positioned to eavesdrop on the network traffic between the client and the server *and* the user must be tricked into using unencrypted HTTP traffic. Proper audit control and separation of duties limit Integrity impact of the attack vector. Users of LedgerSMB 1.8 are urged to upgrade to known-fixed versions. Users of LedgerSMB 1.7 or 1.9 are unaffected by this vulnerability and don't need to take action. As a workaround, users may configure their Apache or Nginx reverse proxy to add the Secure attribute at the network boundary instead of relying on LedgerSMB. For Apache, please refer to the 'Header always edit' configuration command in the mod_headers module. For Nginx, please refer to the 'proxy_cookie_flags' configuration command.

Improper Restriction of XML External Entity Reference in stanfordnlp/corenlp

corenlp is vulnerable to Improper Restriction of XML External Entity Reference

External Control of File Name or Path in netristv/ws-scrcpy

ws-scrcpy is vulnerable to External Control of File Name or Path

Improper Restriction of XML External Entity Reference in dbeaver/dbeaver

dbeaver is vulnerable to Improper Restriction of XML External Entity Reference

Ruby-MySQL Gem Client File Read

A malicious MySQL server can request local file content from a client using ruby-mysql prior to version 2.10.0 without explicit authorization from the user. This issue was resolved in version 2.10.0 and later.

ceph: RGW unauthenticated denial of service

A flaw was found in the Red Hat Ceph Storage RGW in versions before 14.2.21. When processing a GET Request for a swift URL that ends with two slashes it can cause the rgw to crash, resulting in a denial of service. The greatest threat to the system is of availability.

avahi: reachable assertion in avahi_s_host_name_resolver_start when trying to resolve badly-formatted hostnames

A flaw was found in avahi 0.8-5. A reachable assertion is present in avahi_s_host_name_resolver_start function allowing a local attacker to crash the avahi service by requesting hostname resolutions through the avahi socket or dbus methods for invalid hostnames. The highest threat from this vulnerability is to the service availability.