nvme-tcp poll queue crash

[Sagi Grimberg]

> While I am working on the unification of the ctrl state machine I run into
> the crash below. It gets triggered as soon a poll queue is involved in the
> setup, e.g
> 
>    nvme connect-all -P 2
> 
> An older downstream kernel works fine, thus I looked what has changed between
> the older version and current upstream in this area. One commit which
> caught my eye is
> 
>    3e08773c3841 ("block: switch polling to be bio based")
> 
> and a bisect test also points to this commit.
> 
> Not really sure what's wrong yet.

This was reported and is discussed in thread:
nvme-tcp: kernel NULL pointer dereference, address: 0000000000000034