Re: Question about OPP regulator for Panfrost Devfreq

From: Mark Brown <broonie@kernel.org>
To: Viresh Kumar <viresh.kumar@linaro.org>
Cc: "Clément Péron" <peron.clem@gmail.com>,
	"Marek Szyprowski" <m.szyprowski@samsung.com>,
	"Viresh Kumar" <vireshk@kernel.org>, "Nishanth Menon" <nm@ti.com>,
	"Stephen Boyd" <sboyd@kernel.org>,
	"open list:ALLWINNER CPUFREQ DRIVER" <linux-pm@vger.kernel.org>,
	"Steven Price" <steven.price@arm.com>
Subject: Re: Question about OPP regulator for Panfrost Devfreq
Date: Wed, 13 May 2020 12:00:39 +0100	[thread overview]
Message-ID: <20200513110039.GD4803@sirena.org.uk> (raw)
In-Reply-To: <20200513104015.oyiaimjpspablq5p@vireshk-i7>

[-- Attachment #1: Type: text/plain, Size: 1980 bytes --]

On Wed, May 13, 2020 at 04:10:15PM +0530, Viresh Kumar wrote:
> On 13-05-20, 11:18, Mark Brown wrote:
> > On Wed, May 13, 2020 at 02:49:22PM +0530, Viresh Kumar wrote:

> > > @Mark: Regarding enabling/disabling regulators from the OPP core, what

> > My name is spelt Mark.

> Hmm. But isn't that what I wrote ? I used @ as that's what people use
> to address questions to someone. :)

No, it isn't what people use to address a question to someone.  Some
social media uses it to prefix usernames but email is not one of those
systems.  If anything it makes it *less* likely that people will see
things as people mostly read by pattern matching the shapes of words
(especially when scanning) rather than spelling out individual letters.

> > As ever if you have requirements for the voltage of a regulator you
> > should use regulator_set_voltage() to tell the core about it.  The core
> > cannot be expected to infer these requirements without being told by the
> > users.  If you need the voltage to be a particular level when you enable
> > the regulator you should set that voltage.  Why can't the code do that
> > instead of trying to add these complex and fragile bodges?  Randomly
> > skipping applying configurations some of the time is not going to make
> > anything more robust or easier to understand.

> Right.

> Clément is facing a problem where his regulator isn't getting enabled
> and he was asking why the OPP core can't call regulator_enable(), when
> it handles everything else around them. Can you suggest to him on what
> he should be doing here ?

The OPP code can and should be calling regulator_enable().  If the OPP
code needs some particular voltage configuration it can and should be
calling regulator_set_voltage() to tell the core what it needs.  What
I'm saying is that if when the OPP code enables the regulator it needs
a particular voltage configuration it should tell the core about that
prior to enabling.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]