[PATCH] nvmet-rdma: handle commands of established queue outside the CM handler

[Shin'ichiro Kawasaki]

On Oct 23, 2026 / 16:41, Keith Busch wrote:
...
> I think this might be a false positive. The handler_mutex taken in
> rdma_destroy_id() belongs to the listener, while the one held across
> ESTABLISHED belongs to the child cm_id. Is there ever an actual hang, or
> just the lockdep splat?

No, I just observed the WARN. So the WARN is likely a false positive as you
pointed out. I guess we can set different lockdep classes for the listener's
cm_ids and cm_ids for ESTABLISHED connections. Let me drop this patch and cook
something with the approach.