RE: [PATCH 0/3] Add power domain driver support for i.mx8m family

From: Jacky Bai <ping.bai@nxp.com>
To: Sudeep Holla <sudeep.holla@arm.com>
Cc: Lucas Stach <l.stach@pengutronix.de>,
	Aisheng Dong <aisheng.dong@nxp.com>,
	"robh+dt@kernel.org" <robh+dt@kernel.org>,
	"mark.rutland@arm.com" <mark.rutland@arm.com>,
	"shawnguo@kernel.org" <shawnguo@kernel.org>,
	"s.hauer@pengutronix.de" <s.hauer@pengutronix.de>,
	"kernel@pengutronix.de" <kernel@pengutronix.de>,
	"festevam@gmail.com" <festevam@gmail.com>,
	dl-linux-imx <linux-imx@nxp.com>,
	"devicetree@vger.kernel.org" <devicetree@vger.kernel.org>,
	"linux-arm-kernel@lists.infradead.org"
	<linux-arm-kernel@lists.infradead.org>,
	"jacky.baip@gmail.com" <jacky.baip@gmail.com>
Subject: RE: [PATCH 0/3] Add power domain driver support for i.mx8m family
Date: Thu, 18 Apr 2019 01:54:18 +0000	[thread overview]
Message-ID: <VI1PR0402MB35196D5E9D4E261C0153556D87260@VI1PR0402MB3519.eurprd04.prod.outlook.com> (raw)

> 
> On Wed, Apr 17, 2019 at 02:30:53PM +0000, Jacky Bai wrote:
> >
> > Hi Lucas,
> >
> > Either SCMI or SCPI, I think it is initially designed for SOC that has
> > dedicated SCP to manage the SOC resource, but for i.MX8M, we don't
> > have such support. so just for the power domain support, do you have
> > other suggestion for it.
> >
> 
> I disagree, I thought it was clear from my earlier email. All OSPM like Linux
> kernel cares is conformance to the interface. And SCMI/SCPI kind of interface
> doesn't mandate how to implement either at the hardware (you may or may
> not have dedicated processor) or software level(can do it in TF-A through SMC,
> or some remote process through mailbox)
> 

For the current SCMI/SCPI driver in mainline kernel, no SCMI/SCPI over SMC support, right?
If we want to make SCMI over SMC as a standard support, does ARM have some suggestion
or plan? :)

> I am clearly against the custom SMC:
> 
> 1. It will never end and it's put together without much design thoughts
>    just to address specific need of the hour.
> 2. It gets obsolete very soon, mainly because of (1) 3. Very platform and
> vendor specific and soon we will find ourselves
>    in ocean of such custom calls
> 
> And I can definitely continue the list, but I have made the point.
> I think we already have enough of those custom SMC already now and it's
> high time to stop that non-sense for OSPM.
> 

>From the ARM SMC calling convention, the SIP service call is designed to
provide SOC specific service, for example, Secure platform initialization,
configuration and some power control. I am confused if upstream kernel will never
accept SIP service patch.

>From the SCMI spec, it provide the clock management for enable/disable/set_rate,
But don't have set_parent support, for i.MX8M platform, we have the requirement to
set clock parent explicitly. If we move to use SCMI framework, how do handle
the set parent?

> --
> Regards,
> Sudeep