Re: CVS commit: src/sys/dev/i2c

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Tue, Aug 29, 2017 at 02:59:58PM -0300, Jared McNeill wrote:
> I'm not opposed to having a compile time option or sysctl -- but if we do I
> think we should try to make it MI. In theory an ACPI capable x86 board could
> have this problem as well..

Sure. The problem is more which poweroff callback to choose when there are
several available.

> 
> For axp specifically I think the right place to put it would be as a
> properly for the PMIC node in the device tree to disable poweroff.

Yes, I though about this too, and I agree.

> This way
> we can set it on a per-board basis. I didn't see one in the current binding
> docs, so we'd want to coordinate with the linux-sunxi folks on that.

sure.

> > For a personnal project (for those curious, available at
> > https://github.com/mbouyer/marine_chartplotter) I embeeded a olimex lime2
> > in a case, along with a power/interface board. I did develop a driver for
> > the external power controller, and in such a case I don't want the
> > power hook of the axp20x to be called. I don't think this use case is
> > uncommon ...
> 
> Looking at your driver:
> 
> https://github.com/mbouyer/marine_chartplotter/blob/master/software/NetBSD/driver/picbmc.c#L137
> 
> By taking over cpu_powerdown_address this will bypass the FDT hooks, so you
> should be ok here.

But it depends on the order in which the devices are probed/attached,
isn't it ?

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--