RDMA/mlx5: Fix the recovery flow of the UMR QP
Linux
Affected versions
| Product | Affected | Fixed in |
|---|---|---|
| Linux Linux |
|
Linux
| Product | Affected | Fixed in |
|---|---|---|
| Linux Linux |
|
d8f7bff9a42627d37f4ecffeb01e44db42167175| No fix yet |
| Linux Linux |
|
|
In the Linux kernel, the following vulnerability has been resolved:
RDMA/mlx5: Fix the recovery flow of the UMR QP
This patch addresses an issue in the recovery flow of the UMR QP, ensuring tasks do not get stuck, as highlighted by the call trace [1].
During recovery, before transitioning the QP to the RESET state, the software must wait for all outstanding WRs to complete.
Failing to do so can cause the firmware to skip sending some flushed CQEs with errors and simply discard them upon the RESET, as per the IB specification.
This race condition can result in lost CQEs and tasks becoming stuck.
To resolve this, the patch sends a final WR which serves only as a barrier before moving the QP state to RESET.
Once a CQE is received for that final WR, it guarantees that no outstanding WRs remain, making it safe to transition the QP to RESET and subsequently back to RTS, restoring proper functionality.
From the CVE record · Linux
A race in the Linux kernel’s RDMA/mlx5 UMR queue-pair recovery can lose completion events and leave tasks stuck. This can occur during recovery if the queue pair is reset before outstanding work requests complete; the reported hang occurred in a local RDMA verbs operation. The input does not establish whether authentication is required.
Operators running Linux kernels with the RDMA/mlx5 driver should check their kernel’s fix status; affected versions are not specified.