keep-alive timeout zero value

keep-alive timeout zero value

[Engel, Amit]

Hi,

keep-alive timeout is allowed (by spec) to be set to 0h
In this case, the timer will be started ('nvmet_start_keep_alive_timer') where ctrl->kato = 0.
This will lead to an immediate timer expiration

We thought that in case of a zero keep alive timeout,
ctrl->kato should be set to a default NVMET_DISC_KATO value
(core.c , 'nvmet_alloc_ctrl')
this way we will avoid stating the timer with a zero timeout value
(This is correct for both NVME_NQN_DISC and NVME_NQN_NVME)

What are your thoughts?

Thanks
Amit Engel


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme

Re: keep-alive timeout zero value

[hch]

On Mon, Aug 17, 2020 at 04:19:36PM +0000, Engel, Amit wrote:
> Hi,
> 
> keep-alive timeout is allowed (by spec) to be set to 0h
> In this case, the timer will be started ('nvmet_start_keep_alive_timer') where ctrl->kato = 0.
> This will lead to an immediate timer expiration
> 
> We thought that in case of a zero keep alive timeout,
> ctrl->kato should be set to a default NVMET_DISC_KATO value
> (core.c , 'nvmet_alloc_ctrl')
> this way we will avoid stating the timer with a zero timeout value
> (This is correct for both NVME_NQN_DISC and NVME_NQN_NVME)
> 
> What are your thoughts?

Yes.  Can you send a patch?


_______________________________________________
Linux-nvme mailing list
Linux-nvme@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-nvme