Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:8 as permitted sender) client-ip=2620:137:e000::3:8;
MIME-Version: 1.0
References: <cover.1697710527.git.viresh.kumar@linaro.org> <f709e9e273004be43efe3a2854a7e7b51a777f99.1697710527.git.viresh.kumar@linaro.org>
 <CAPDyKFqbnsdT0nqKwQhai875CwwpW_vepr816fL+i8yLh=YQhw@mail.gmail.com>
 <20231020034547.suggwoefx5kauek4@vireshk-i7> <CAPDyKFqMA9=qdz4L3Oqj0zRQmSj0bxrF1RzZu-pBcuj9__GSRw@mail.gmail.com>
 <20231020105658.4zjfuiawwyrtnssu@vireshk-i7>
In-Reply-To: <20231020105658.4zjfuiawwyrtnssu@vireshk-i7>
From:   Ulf Hansson <ulf.hansson@linaro.org>
Date:   Fri, 20 Oct 2023 13:09:13 +0200
Message-ID: <CAPDyKFqKVFO-p8Amu8aD7i0tT2dtpJw5rbDVxxhjOpaTYY=ZzA@mail.gmail.com>
Subject: Re: [PATCH 1/2] OPP: Use _set_opp_level() for single genpd case
To:     Viresh Kumar <viresh.kumar@linaro.org>
Cc:     Viresh Kumar <vireshk@kernel.org>, Nishanth Menon <nm@ti.com>,
        Stephen Boyd <sboyd@kernel.org>, linux-pm@vger.kernel.org,
        Vincent Guittot <vincent.guittot@linaro.org>,
        "Rafael J. Wysocki" <rafael@kernel.org>,
        Stephan Gerhold <stephan.gerhold@kernkonzept.com>,
        Konrad Dybcio <konrad.dybcio@linaro.org>,
        Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, 20 Oct 2023 at 12:57, Viresh Kumar <viresh.kumar@linaro.org> wrote:
>
> On 20-10-23, 12:02, Ulf Hansson wrote:
> > For the single PM domain case, consumer drivers are often not able to
> > use dev_pm_opp_set_config(). That's because the PM domain has already
> > been attached from some of the generic buses, through
> > dev_pm_domain_attach().
> >
> > In this case, as dev_pm_opp_set_config() ends up trying to attach
> > again, via dev_pm_domain_attach_by_name() it would receive
> > "ERR_PTR(-EEXIST)".
> >
> > Or maybe I didn't quite understand your point?
>
> So the thing is that I _really_ want to call dev_pm_opp_set_opp() for
> each OPP we want to configure, primary or required. For example, the
> required OPP may want to do more than just performance state and we
> aren't touching them right now.

I understand - and it makes perfect sense to me too!

>
> Now, in order to call dev_pm_opp_set_opp() for any device, we need a
> device pointer and an OPP table associated with it.
>
> I can take care of it for the multi genpd case as there are extra
> device structures (which we get from dev_pm_domain_attach_by_name()),
> but there is no clean way out for single PM domain devices, unless
> they also call dev_pm_opp_set_config() to get a virtual structure.
>
> This is why I had to get this hackish code in place to make it work
> with the recursive calls to dev_pm_opp_set_opp(), where I could just
> reuse the opp-level thing for the primary device.
>
> How do you suggest we take care of this now ?

Honestly, I don't know yet. But I am certainly willing to help. Allow
me to have a closer look and see if I can propose a way forward.

Kind regards
Uffe