s2idle breaks on machines without cpuidle support

[Sudeep Holla]

On Thu, Feb 09, 2023 at 12:19:39AM +0900, Kazuki wrote:
> On Wed, Feb 08, 2023 at 03:03:06PM +0000, Sudeep Holla wrote:

[...]

> > Well, if we are allowing to boot on such a system, then we must allow
> > adding a platform specific idle driver. It may be useful once we info
> > to add deeper than WFI states.
>
> Hmmm, I thought for arm64, non-PSCI idle drivers were prohibited? Or am
> I mistaken here?

I don't know. I am not against it especially now that we have allowed a
non-PSCI based production system to boot the kernel.

> > Are we ? I thought were making changes to enable it. Or are you saying
> > we allow to enter into such a state and render the system unusable, if
> > so we need to fix it.
>
> Both as I mentioned in my first email. Apologies if it turned out to
> be confusing.

Sorry still confusing. Are you saying you can enter s2idle and crash or
hang the system without changes(especially around this s2idle code) ?

If yes, then it is a bug. If the hang/crash is only after your changes,
we need to check.

-- 
Regards,
Sudeep