CVE-2024-27435 is a medium-severity vulnerability in the Linux kernel affecting the NVMe over RDMA subsystem. The issue arises from a deadlock condition during the reconnection process of the NVMe admin queue (admin_q). Specifically, the vulnerability is caused by an ABBA deadlock related to tag allocation. In NVMe over RDMA, tags are used to track outstanding commands. The admin queue and fabric queue share a tagset, with a limited number of reserved tags. During a controller reset, the admin queue is quiesced, causing keep-alive requests to be held and marked idle, preventing their processing until the reset completes. However, when attempting to reconnect the remote target, a tag is required for the connect command. If the only reserved tag is held by a keep-alive command waiting inside the admin queue, the reconnection attempt fails indefinitely because no tags are available to proceed. This results in a permanent failure to reconnect the admin queue despite the remote target and network being operational. The fix involves reserving two tags for the admin queue to avoid this deadlock scenario. The vulnerability requires local privileges (PR:L), no user interaction, and has a limited attack vector (local access), with no impact on confidentiality or integrity but causing a denial of service (availability impact). No known exploits are reported in the wild at this time. The affected versions are specific Linux kernel commits identified by their hashes, indicating that this is a recent and targeted fix in the kernel source.

For European organizations, this vulnerability primarily poses a risk of denial of service in environments using NVMe over RDMA storage configurations on Linux systems. NVMe over RDMA is commonly deployed in high-performance computing, data centers, and enterprise storage solutions where low-latency and high-throughput storage access is critical. A failure to reconnect the NVMe admin queue could lead to storage subsystem unavailability, potentially disrupting critical applications, data processing, and services reliant on these storage systems. This could affect cloud service providers, financial institutions, research centers, and large enterprises with advanced storage infrastructures. Although the vulnerability does not compromise data confidentiality or integrity, the availability impact could cause operational downtime and service degradation. Given the requirement for local privileges to exploit, the threat is more relevant in scenarios where an attacker or malfunctioning process already has some level of access to the host system. Therefore, the impact is significant in environments where Linux servers with NVMe over RDMA are used and where local access controls might be insufficient.

To mitigate this vulnerability, organizations should promptly apply the Linux kernel patches that address CVE-2024-27435. Since the fix involves kernel-level changes, updating to the latest stable kernel versions containing the patch is essential. Additionally, organizations should: 1) Audit and monitor systems using NVMe over RDMA to detect any abnormal admin queue reconnection failures or storage subsystem issues. 2) Restrict local access to trusted users and processes to reduce the risk of exploitation, as local privileges are required. 3) Implement robust access controls and system hardening to prevent unauthorized local access. 4) Test kernel updates in staging environments to ensure compatibility with existing NVMe over RDMA deployments before production rollout. 5) Consider fallback or redundancy mechanisms for critical storage paths to minimize downtime in case of admin queue failures. 6) Engage with hardware and software vendors for any firmware or driver updates that complement the kernel patch. These steps go beyond generic advice by focusing on the specific NVMe over RDMA context and the nature of the deadlock issue.