OpenVPN versions before 2.4.3 and before 2.3.17 are vulnerable to remote denial-of-service due to memory exhaustion caused by memory leaks and double-free issue in extract_x509_extension().
OpenVPN versions before 2.4.3 and before 2.3.17 are vulnerable to denial-of-service by authenticated remote attacker via sending a certificate with an embedded NULL character.
OpenVPN version 2.3.12 and newer is vulnerable to unauthenticated Denial of Service of server via received large control packet. Note that this issue is fixed in 2.3.15 and 2.4.2.
OpenVPN versions before 2.3.15 and before 2.4.2 are vulnerable to reachable assertion when packet-ID counter rolls over resulting into Denial of Service of server by authenticated attacker.
OpenVPN, when using a 64-bit block cipher, makes it easier for remote attackers to obtain cleartext data via a birthday attack against a long-duration encrypted session, as demonstrated by an HTTP-over-OpenVPN session using Blowfish in CBC mode, aka a "Sweet32" attack.
OpenVPN 2.x before 2.0.11, 2.1.x, 2.2.x before 2.2.3, and 2.3.x before 2.3.6 allows remote authenticated users to cause a denial of service (server crash) via a small control channel packet.
Unquoted Windows search path vulnerability in the ptservice service prior to PrivateTunnel version 3.0 (Windows) and OpenVPN Connect version 3.1 (Windows) allows local users to gain privileges via a crafted program.exe file in the %SYSTEMDRIVE% folder.
The openvpn_decrypt function in crypto.c in OpenVPN 2.3.0 and earlier, when running in UDP mode, allows remote attackers to obtain sensitive information via a timing attack involving an HMAC comparison function that does not run in constant time and a padding oracle attack on the CBC mode cipher.
Unspecified vulnerability in OpenVPN 2.1-beta14 through 2.1-rc8, when running on non-Windows systems, allows remote servers to execute arbitrary commands via crafted (1) lladdr and (2) iproute configuration directives, probably related to shell metacharacters.
OpenVPN 2.0.7 and earlier, when configured to use the --management option with an IP that is not 127.0.0.1, uses a cleartext password for TCP sessions to the management interface, which might allow remote attackers to view sensitive information or cause a denial of service.