From mboxrd@z Thu Jan  1 00:00:00 1970
From: hch@lst.de (Christoph Hellwig)
Date: Tue, 19 Apr 2016 20:25:45 +0200
Subject: [PATCH 2/5] nvme: introduce a controller state machine
In-Reply-To: <571543EA.6080700@grimberg.me>
References: <1460982893-11315-1-git-send-email-hch@lst.de>
 <1460982893-11315-3-git-send-email-hch@lst.de> <5715294A.3020309@grimberg.me>
 <20160418201710.GA4921@lst.de> <571543EA.6080700@grimberg.me>
Message-ID: <20160419182545.GA989@lst.de>

On Mon, Apr 18, 2016@11:30:34PM +0300, Sagi Grimberg wrote:
>
>>>> -	clear_bit(NVME_CTRL_RESETTING, &dev->flags);
>>>> +	if (!nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_LIVE)) {
>>>> +		dev_warn(dev->ctrl.device, "failed to mark controller live\n");
>>>> +		goto out;
>>>> +	}
>>>>    	return;
>>>>
>>>>     out:
>>>
>>> You're missing a state change before removing a
>>> dead controller...
>>
>> why?
>
> Just to have the state consistent with what the driver
> is doing. Not a must though...

Let's allocate states on an as needed basis and avoid that new
state for now.