Re: [PATCH v3 1/4] media: dt-bindings: media: i2c: Switch to assigned-clock-rates

From: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
To: "Lad, Prabhakar" <prabhakar.csengg@gmail.com>
Cc: Maxime Ripard <maxime@cerno.tech>,
	Mark Rutland <mark.rutland@arm.com>,
	"open list:OPEN FIRMWARE AND FLATTENED DEVICE TREE BINDINGS" 
	<devicetree@vger.kernel.org>,
	Pengutronix Kernel Team <kernel@pengutronix.de>,
	Ezequiel Garcia <ezequiel@collabora.com>,
	Sascha Hauer <s.hauer@pengutronix.de>,
	Magnus Damm <magnus.damm@gmail.com>,
	Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>,
	LKML <linux-kernel@vger.kernel.org>,
	Linux-Renesas <linux-renesas-soc@vger.kernel.org>,
	Rob Herring <robh+dt@kernel.org>,
	Geert Uytterhoeven <geert@linux-m68k.org>,
	Sakari Ailus <sakari.ailus@linux.intel.com>,
	Shawn Guo <shawnguo@kernel.org>,
	Mauro Carvalho Chehab <mchehab@kernel.org>,
	Fabio Estevam <festevam@gmail.com>,
	linux-media <linux-media@vger.kernel.org>,
	LAK <linux-arm-kernel@lists.infradead.org>,
	NXP Linux Team <linux-imx@nxp.com>
Subject: Re: [PATCH v3 1/4] media: dt-bindings: media: i2c: Switch to assigned-clock-rates
Date: Tue, 24 Mar 2020 18:12:21 +0200	[thread overview]
Message-ID: <20200324161221.GA27805@pendragon.ideasonboard.com> (raw)
In-Reply-To: <CA+V-a8tMVoJOdgM_S0sJ0WEGhwBirCC4mi-TtxLCn1SKVXXiBQ@mail.gmail.com>

Hi Prabhakar,

On Tue, Mar 24, 2020 at 04:04:43PM +0000, Lad, Prabhakar wrote:
> On Tue, Mar 24, 2020 at 3:40 PM Maxime Ripard <maxime@cerno.tech> wrote:
> > On Thu, Mar 19, 2020 at 01:17:51PM +0000, Lad, Prabhakar wrote:
> > > On Thu, Mar 19, 2020 at 1:04 PM Laurent Pinchart wrote:
> > > > On Thu, Mar 19, 2020 at 01:44:52PM +0100, Maxime Ripard wrote:
> > > > > On Fri, Mar 13, 2020 at 09:12:31PM +0000, Lad Prabhakar wrote:
> > > > > > Use assigned-clock-rates to specify the clock rate. Also mark
> > > > > > clock-frequency property as deprecated.
> > > > > >
> > > > > > Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
> > > > > > ---
> > > > > >  Documentation/devicetree/bindings/media/i2c/ov5645.txt | 5 +++--
> > > > > >  1 file changed, 3 insertions(+), 2 deletions(-)
> > > > > >
> > > > > > diff --git a/Documentation/devicetree/bindings/media/i2c/ov5645.txt b/Documentation/devicetree/bindings/media/i2c/ov5645.txt
> > > > > > index 72ad992..e62fe82 100644
> > > > > > --- a/Documentation/devicetree/bindings/media/i2c/ov5645.txt
> > > > > > +++ b/Documentation/devicetree/bindings/media/i2c/ov5645.txt
> > > > > > @@ -8,7 +8,7 @@ Required Properties:
> > > > > >  - compatible: Value should be "ovti,ov5645".
> > > > > >  - clocks: Reference to the xclk clock.
> > > > > >  - clock-names: Should be "xclk".
> > > > > > -- clock-frequency: Frequency of the xclk clock.
> > > > > > +- clock-frequency (deprecated): Frequency of the xclk clock.
> > > > > >  - enable-gpios: Chip enable GPIO. Polarity is GPIO_ACTIVE_HIGH. This corresponds
> > > > > >    to the hardware pin PWDNB which is physically active low.
> > > > > >  - reset-gpios: Chip reset GPIO. Polarity is GPIO_ACTIVE_LOW. This corresponds to
> > > > > > @@ -37,7 +37,8 @@ Example:
> > > > > >
> > > > > >                     clocks = <&clks 200>;
> > > > > >                     clock-names = "xclk";
> > > > > > -                   clock-frequency = <24000000>;
> > > > > > +                   assigned-clocks = <&clks 200>;
> > > > > > +                   assigned-clock-rates = <24000000>;
> > > > > >
> > > > > >                     vdddo-supply = <&camera_dovdd_1v8>;
> > > > > >                     vdda-supply = <&camera_avdd_2v8>;
> > > > >
> > > > > clock-frequency is quite different from assigned-clock-rates though,
> > > > > semantically speaking. clock-frequency is only about what the clock
> > > > > frequency is, while assigned-clock-rates will change the rate as well,
> > > > > and you have no idea how long it will last.
> > > >
> > > > The driver currently reads the clock-frequency property and then calls
> > > > clk_set_rate(). I agree tht assigned-clock-rates isn't a panacea, but I
> > > > think it's less of a hack than what we currently have.
> > > >
> > > > As discussed on IRC, maybe the best option in this specific case is to
> > > > drop clock-frequency and assigned-clok-rates, and call clk_set_rate()
> > > > with a hardcoded frequency of 24MHz in the driver, as that's the only
> > > > frequency the driver supports.
> > > >
> > > Does this mean any driver which has a fixed clock requirement shouldn't be a
> > > DT property and should be just handled by the drivers internally ?
> >
> > It's hard to give a generic policy, but here, the hardware is pretty
> > flexible since it can deal with anything between 6MHz to 50-something
> > MHz, it's the driver that chooses to enforce a 24MHz and be pedantic
> > about it, so it's up to the driver to enforce that policy, not to the
> > DT since it's essentially a software limitation, not a hardware one.
>
> Thank you for the clarification, Ill drop patches 1-4 from this series.

That's the whole series... :-) I think you should keep patch 1/4 but
just remove the clock-frequency from the bindings, then remove it from
the DT files, and patch the driver to set the clock rate to 24MHz
unconditionally in patch 4/4.

-- 
Regards,

Laurent Pinchart