On 9/14/21 9:06 AM, Sagi Grimberg wrote:
>>>> @@ -361,11 +366,13 @@ static inline void nvme_end_req(struct request
>>>> *req)
>>>> void nvme_complete_rq(struct request *req)
>>>> {
>>>> + struct nvme_ctrl *ctrl = nvme_req(req)->ctrl;
>>>> +
>>>> trace_nvme_complete_rq(req);
>>>> nvme_cleanup_cmd(req);
>>>> - if (nvme_req(req)->ctrl->kas)
>>>> - nvme_req(req)->ctrl->comp_seen = true;
>>>> + if (ctrl->kas)
>>>> + ctrl->comp_seen = true;
>>>> switch (nvme_decide_disposition(req)) {
>>>> case COMPLETE:
>>>> @@ -377,6 +384,15 @@ void nvme_complete_rq(struct request *req)
>>>> case FAILOVER:
>>>> nvme_failover_req(req);
>>>> return;
>>>> + case AUTHENTICATE:
>>>> +#ifdef CONFIG_NVME_AUTH
>>>> + if (nvme_change_ctrl_state(ctrl, NVME_CTRL_RESETTING))
>>>> + queue_work(nvme_wq, &ctrl->dhchap_auth_work);
>>>
>>> Why is the state change here and not in nvme_dhchap_auth_work?
>>>
>> Because switching to 'resetting' is an easy way to synchronize with the
>> admin queue.
>
> Maybe fold this into nvme_authenticate_ctrl? in case someone adds/moves
> this in the future and forgets the ctrl state serialization?
Yeah; not a bad idea. Will be looking into it.
Cheers,
Hannes
--
Dr. Hannes Reinecke Kernel Storage Architect
hare@xxxxxxx +49 911 74053 688
SUSE Software Solutions Germany GmbH, Maxfeldstr. 5, 90409 Nürnberg
HRB 36809 (AG Nürnberg), GF: Felix Imendörffer
