Linux NVMe Fabric Controller Kernel Crash Fix via Sync Keep-Alive

From this webpage screenshot, we can extract the following key information about the vulnerability: 1. Vulnerability Description: - Issue Description: The NVMe keep-alive operation may "leak" during the shutdown of a Fabric controller, causing a race condition between the Fabric controller's admin queue and the hw/hctx queue dispatcher. - Potential Impact: This race condition may result in the admin queue of the Fabric controller being deleted, leading to a kernel crash. 2. Mitigation Measures: - Solution: Implement the keep-alive operation as a synchronous operation, ensuring that the (admin queue usage counter) is decremented only after the keep-alive operation completes. - Specific Implementation: - In the function, ensure that is decremented only after the keep-alive operation completes. - In the function, ensure that relevant operations are performed only after the keep-alive operation completes. 3. Code Changes: - File: - Changes Made: - Added the function. - Modified the function to ensure is decremented only after the keep-alive operation completes. 4. Review and Sign-off: - Reviewer: Christoph Hellwig - Sign-offs: Nilay Shroff and Keith Busch This information indicates that the vulnerability involves a potential issue with the NVMe keep-alive operation during Fabric controller shutdown, which is resolved by synchronizing the operation and making appropriate code modifications.

Goal Reached Thanks to every supporter — we hit 100%!

Linux NVMe Fabric Controller Kernel Crash Fix via Sync Keep-Alive