CVE-2026-31565: RDMA/irdma: Fix deadlock during netdev reset with active connections
In the Linux kernel, the following vulnerability has been resolved:
RDMA/irdma: Fix deadlock during netdev reset with active connections
Resolve deadlock that occurs when user executes netdev reset while RDMA applications (e.g., rping) are active. The netdev reset causes ice driver to remove irdma auxiliary driver, triggering devicedelete and subsequent client removal. During client removal, uverbsclient waits for QP reference count to reach zero while cmaclient holds the final reference, creating circular dependency and indefinite wait in iWARP mode. Skip QP reference count wait during device reset to prevent deadlock.
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade to a fixed release to a version that resolves this vulnerability.
Fixed in 6.6.134.1-2
Event History
Frequently Asked Questions
Who is exposed to this deadlock?
Systems using the Linux kernel irdma RDMA driver in iWARP mode are exposed when RDMA connections remain active during a network-device reset. The described active workload includes RDMA applications such as rping.
What action triggers the issue?
A local user must execute a netdev reset while active RDMA connections exist. The reset removes the irdma auxiliary driver, and client removal can then enter an indefinite wait on a QP reference count.
What is the operational impact?
The issue causes a circular dependency during device/client removal and can result in an indefinite wait, producing a denial-of-service condition. The CVSS vector indicates local access and low privileges are required, with no user interaction.
What can be done if the fix cannot be deployed immediately?
Avoid resetting affected network devices while iWARP RDMA applications have active connections. Coordinate shutdown or disconnection of RDMA workloads before initiating a netdev reset.