PT-2024-35566 · Linux+1 · Linux Kernel+1
Published
2024-11-25
·
Updated
2026-01-13
·
CVE-2024-53102
None
No severity ratings or metrics are available. When they are, we'll update the corresponding info on the page.
Name of the Vulnerable Software and Affected Versions
Linux kernel versions prior to 6.6.65
Description
The issue arises from the nvme keep-alive operation, which could potentially cause a race between the fabric controller admin queue destroy code path and the hw/hctx queue dispatcher. This race could lead to a kernel crash. The problem occurs when the nvme keep-alive request sneaks in while shutting down the fabric controller, causing the admin->q usage counter to become zero and allowing the blk-mq destroy queue operation to delete the admin queue. However, the nvme keep-alive thread may still access the admin queue resource after it has been deleted, resulting in a crash. The fix implements the keep-alive operation as a synchronous operation, ensuring that the fabric shutdown code path does not destroy the fabric admin queue until the keep-alive request has finished execution.
Recommendations
To resolve the issue, update to Linux kernel version 6.6.65 or later. As a temporary workaround, consider disabling the nvme keep-alive operation until a patch is available. Restrict access to the admin queue to minimize the risk of exploitation. Avoid using the
admin->q usage counter variable in the affected code path until the issue is resolved. Found an issue in the description? Have something to add? Feel free to write us 👾
Related Identifiers
Affected Products
Linux Kernel
Red Os