On Wed, Aug 25, 2021 at 07:03:34PM +0300, Vladimir Oltean wrote:
> On Wed, Aug 25, 2021 at 04:54:27PM +0100, Mark Brown wrote:

> > No problem with adding raw spinlocks to regmap, I think it's just nobody
> > needed them before.  I've not looked at the problem in sufficient detail
> > to figure out if that's actually the best solution here but from an initial
> > pass through it looks reasonableit looks reasonable

> The question becomes how will syscon be told that one of its consumers
> needs the regmap to use raw spinlock locking? The syscon regmap is
> initialized before any of its consumers probe, AFAIU.

I'd expect it to figure this out based on the compatible string for the
syscon device.