The /charms endpoint on a Juju controller lacked sufficient authorization checks, allowing any user with an account on the controller to upload a charm. Uploading a malicious charm that exploits a Zip Slip vulnerability could allow an attacker to gain access to a machine running a unit through the affected charm.
Certificate generation in juju/utils using the cert.NewLeaf function could include private information. If this certificate were then transferred over the network in plaintext, an attacker listening on that network could sniff the certificate and trivially extract the private key from it.
Sudo before 1.9.17p1 allows local users to obtain root access because /etc/nsswitch.conf from a user-controlled directory is used with the --chroot option.
When a non-x86 platform is detected, cloud-init grants root access to a hardcoded url with a local IP address. To prevent this, cloud-init default configurations disable platform enumeration.
cloud-init through 25.1.2 includes the systemd socket unit cloud-init-hotplugd.socket with default SocketMode that grants 0666 permissions, making it world-writable. This is used for the "/run/cloud-init/hook-hotplug-cmd" FIFO. An unprivileged user could trigger hotplug-hook commands.
A flaw was found in the temporary user record that authd uses in the pre-auth NSS. As a result, a user login for the first time will be considered to be part of the root group in the context of that SSH session.
Race condition in Canonical apport up to and including 2.32.0 allows a local attacker to leak sensitive information via PID-reuse by leveraging namespaces.
When handling a crash, the function `_check_global_pid_and_forward`, which detects if the crashing process resided in a container, was being called before `consistency_checks`, which attempts to detect if the crashing process had been replaced. Because of this, if a process crashed and was quickly replaced with a containerized one, apport could be made to forward the core dump to the container, potentially leaking sensitive information. `consistency_checks` is now being called before `_check_global_pid_and_forward`. Additionally, given that the PID-reuse race condition cannot be reliably detected from userspace alone, crashes are only forwarded to containers if the kernel provided a pidfd, or if the crashing process was unprivileged (i.e., if dump mode == 1).
In Ubuntu, gnome-control-center did not properly reflect SSH remote login status when the system was configured to use systemd socket activation for openssh-server. This could unknowingly leave the local machine exposed to remote SSH access contrary to expectation of the user.
Running DDoS on tcp port 22 will trigger a kernel crash. This issue is introduced by the backport of a commit regarding nft_lookup without the subsequent fixes that were introduced after this commit. The resolution of this CVE introduces those commits to the linux-bluefield package.