From mboxrd@z Thu Jan  1 00:00:00 1970
From: Mike Turquette <mturquette@linaro.org>
Subject: Re: [RESEND 2/2] clk: Add driver for Palmas clk32kg and clk32kgaudio clocks
Date: Tue, 01 Jul 2014 21:33:38 -0700
Message-ID: <20140702043338.23338.31600@quantum>
References: <1403848871-26519-1-git-send-email-peter.ujfalusi@ti.com>
 <1403848871-26519-3-git-send-email-peter.ujfalusi@ti.com>
 <CABxcv=metB4=eoT=fAh7S-b_mhAvHcJtn2wrV8v8koSyKMxNNw@mail.gmail.com>
 <53B0FC27.5040002@ti.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: QUOTED-PRINTABLE
Return-path: <linux-doc-owner@vger.kernel.org>
In-Reply-To: <53B0FC27.5040002@ti.com>
Sender: linux-doc-owner@vger.kernel.org
To: Peter Ujfalusi <peter.ujfalusi@ti.com>, Javier Martinez Canillas <javier@dowhile0.org>
Cc: Grant Likely <grant.likely@linaro.org>, "devicetree@vger.kernel.org" <devicetree@vger.kernel.org>, Linux Kernel <linux-kernel@vger.kernel.org>, "linux-omap@vger.kernel.org" <linux-omap@vger.kernel.org>, ldewangan@nvidia.com, linux-doc@vger.kernel.org, Nishanth Menon <nm@ti.com>, Rob Herring <robh+dt@kernel.org>
List-Id: devicetree@vger.kernel.org

Quoting Peter Ujfalusi (2014-06-29 22:56:55)
> Hi Javier,
>=20
> On 06/27/2014 09:23 PM, Javier Martinez Canillas wrote:
> > Hello Peter,
> >=20
> > On Fri, Jun 27, 2014 at 8:01 AM, Peter Ujfalusi <peter.ujfalusi@ti.=
com> wrote:
> >> Palmas class of devices can provide 32K clock(s) to be used by oth=
er devices
> >> on the board. Depending on the actual device the provided clocks c=
an be:
> >> CLK32K_KG and CLK32K_KGAUDIO
> >> or only one:
> >> CLK32K_KG (TPS659039 for example)
> >>
> >> Use separate compatible flags for the two 32K clock.
> >> A system which needs or have only one of the 32k clock from
> >> Palmas will need to add node(s) for each clock as separate section
> >> in the dts file.
> >> The two compatible property is:
> >> "ti,palmas-clk32kg" for clk32kg clock
> >> "ti,palmas-clk32kgaudio" for clk32kgaudio clock
> >>
> >> Apart from the register control of the clocks - which is done via
> >> the clock API there is a posibility to enable the external sleep
> >> control. In this way the clock can be enabled/disabled on demand b=
y the
> >> user of the clock.
> >>
> >> See the documentation for more details.
> >>
> >> Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
> >> Reviewed-by: Nishanth Menon <nm@ti.com>
>=20
> >> +static unsigned long palmas_clks_recalc_rate(struct clk_hw *hw,
> >> +                                            unsigned long parent_=
rate)
> >> +{
> >> +       return 32768;
> >> +}
> >=20
> > I see that other clock drivers using a constant rate return 0 if th=
e
> > clock has not been enabled.
>=20
> and there are examples when similar fixed clock drivers returns only =
the clock
> value, like clk-max77686. I can not find clear guidelines neither in =
the
> documentation or around the header/c files for this.
> Mike, what is the appropriate way of handling the recalc_rate?

You are right that there are no guidelines stating, "don't do that", bu=
t
please, "don't do that" ;-)

clk_enable and clk_set_rate are entirely unrelated operations from the
perspective of the Linux clock framework, and mixing these two classes
of operations is a recipe for pain.

>=20
> > So maybe is more correct to have something
> > like the following?
> >=20
> > if (__clk_is_enabled(hw->clk))
> >         return 32768;
> > else
> >         return 0;

So what happens here if this is gateable clock and later on we call
clk_enable on it? The clocks rate will still be zero since
clk_enable/clk_disable do not touch the rate at all.

Regards,
Mike

> >=20
> > Best regards,
> > Javier
> >=20
>=20
>=20
> --=20
> P=C3=A9ter