[PATCH] nvmet-rdma: Suppress a class of lockdep complaints

[Max Gurtovoy]

On 09/05/2023 16:56, Bart Van Assche wrote:
> Waiting in nvmet_rdma_queue_connect() for previous teardowns to complete
> (flush_workqueue(nvmet_wq)) confuses lockdep. Clear up this confusion by
> by using dynamic lockdep keys instead of static lockdep keys for
> queue->release_work.
> 

The concept of adding a code to a working driver only because of some 
confusion of some other SW competent is not clear to me.

Can we fix the lockdep not to get confused or ignore it if its not a 
real issue ?