[PATCH 2/5] nvme: introduce a controller state machine

[Christoph Hellwig]

On Tue, Apr 19, 2016 at 09:35:20PM +0300, Sagi Grimberg wrote:
>> Let's allocate states on an as needed basis and avoid that new
>> state for now.
>
> Didn't mean a new state, just a transition to NVME_CTRL_DELETING
> if nvme_reset_work fails...

Oh, we can't actually do that - for PCIe deleting is only set
for calls from the PCI ->remove callback, which we'll invoke a
little later through the driver model from the workqueue we just
kicked off.