[PATCH v6 07/18] nvme: Introduce FENCING and FENCED controller states

Mohamed Khalfella mkhalfella at purestorage.com
Mon Oct 19 14:40:07 PDT 2026


On Mon 2026-10-19 19:18:32 +0200, Daniel Wagner wrote:
> On Mon, Oct 19, 2026 at 07:01:38PM +0200, Daniel Wagner wrote:
> > > --- a/drivers/nvme/host/sysfs.c
> > > +++ b/drivers/nvme/host/sysfs.c
> > > @@ -542,6 +542,8 @@ static ssize_t nvme_sysfs_show_state(struct device *dev,
> > >  	static const char *const state_name[] = {
> > >  		[NVME_CTRL_NEW]		= "new",
> > >  		[NVME_CTRL_LIVE]	= "live",
> > > +		[NVME_CTRL_FENCING]	= "fencing",
> > > +		[NVME_CTRL_FENCED]	= "fenced",
> > >  		[NVME_CTRL_RESETTING]	= "resetting",
> > >  		[NVME_CTRL_CONNECTING]	= "connecting",
> > >  		[NVME_CTRL_DELETING]	= "deleting",
> > 
> > The ABI documentation needs to be updated.
> 
> And while you’re at it, what’s the impact on blktests? I can’t quite
> recall what you presented during ALPSS regarding blktests. Is this
> covered?

Both [1] and [2] trigger "fencing" state. [1] focuses on not seeing
corruption while [2] also inspects kernel logs to make sure that CCR and
CQT are triggered when they should. There is no explict check of the
controller state in these tests. We know that CCR/CQT will happen in
"fencing" state though.

> 
> Keith, Christoph, and I are mostly concerned about this patch.
> Experience tells us that anyone touching the state machine will
> introduce regressions. Because of that, I’d really like to make sure the
> upstream testing side is covered. I know you’ve run tests in your lab,
> but we need test cases that everyone can execute in their own
> environments, and blktests is the right vehicle for that.
> 

I documented part of the testing I did in lab in my ALPSS slides [3].

I addition to [1] and [2], what kind of blktest you would like to see?
If you describe the blktest in detail I will implement it and put it for
review.

1- https://lore.kernel.org/all/20260917020752.1672578-1-mkhalfella@purestorage.com/
2- https://lore.kernel.org/all/20260921205951.3023714-1-mkhalfella@purestorage.com/
3- https://drive.google.com/file/d/17lt_H3FxeoDSWD8q32diO7e03JYJaimn/view?usp=drive_link



More information about the Linux-nvme mailing list