OpenVPN before 2.6.11 does not santize PUSH_REPLY messages properly which an attacker controlling the server can use to inject unexpected arbitrary data ending up in client logs.
OpenVPN from 2.6.0 through 2.6.10 in a server role accepts multiple exit notifications from authenticated clients which will extend the validity of a closing session
tap-windows6 driver version 9.26 and earlier does not properly
check the size data of incomming write operations which an attacker can
use to overflow memory buffers, resulting in a bug check and potentially
arbitrary code execution in kernel space
The interactive service in OpenVPN 2.6.9 and earlier allows the OpenVPN service pipe to be accessed remotely, which allows a remote attacker to interact with the privileged OpenVPN interactive service.
The interactive service in OpenVPN 2.6.9 and earlier allows an attacker to send data causing a stack overflow which can be used to execute arbitrary code with more privileges.
OpenVPN plug-ins on Windows with OpenVPN 2.6.9 and earlier could be loaded from any directory, which allows an attacker to load an arbitrary plug-in which can be used to interact with the privileged OpenVPN interactive service.
The PKCS#7 parser in OpenVPN 3 Core Library versions through 3.8.3 did not properly validate the parsed data, which would result in the application crashing.
The OpenVPN GUI installer before version 2.6.9 did not set the proper access control restrictions to the installation directory of OpenVPN binaries when using a non-standard installation path, which allows an attacker to replace binaries to run arbitrary executables.
The nodejs framework in OpenVPN Connect 3.0 through 3.4.3 (Windows)/3.4.7 (macOS) was not properly configured, which allows a local user to execute arbitrary code within the nodejs process context via the ELECTRON_RUN_AS_NODE environment variable
OpenVPN Connect version 3.0 through 3.4.6 on macOS allows local users to execute code in external third party libraries using the DYLD_INSERT_LIBRARIES environment variable