Re: [PATCH v2 09/11] media: i2c: max9286: Configure remote I2C speed from device tree

[Laurent Pinchart <laurent.pinchart@ideasonboard.com>]

Hi Jacopo,

On Sun, Jan 09, 2022 at 11:43:32AM +0100, Jacopo Mondi wrote:
> On Sat, Jan 01, 2022 at 08:28:04PM +0200, Laurent Pinchart wrote:
> > Read the maxim,i2c-clock-frequency DT property that specifies the speed
> > of the remote I2C bus, and configure the MAX9286 accordingly. The remote
> > serializers must all have a matching configuration.
> >
> > Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
> > ---
> >  drivers/media/i2c/max9286.c | 56 +++++++++++++++++++++++++++++++------
> >  1 file changed, 47 insertions(+), 9 deletions(-)
> >
> > diff --git a/drivers/media/i2c/max9286.c b/drivers/media/i2c/max9286.c
> > index 4b69bd036ca6..d88a4d8e63ab 100644
> > --- a/drivers/media/i2c/max9286.c
> > +++ b/drivers/media/i2c/max9286.c
> > @@ -143,6 +143,11 @@ struct max9286_format_info {
> >  	u8 datatype;
> >  };
> >
> > +struct max9286_i2c_speed {
> > +	u32 rate;
> > +	u8 mstbt;
> > +};
> > +
> >  struct max9286_source {
> >  	struct v4l2_subdev *sd;
> >  	struct fwnode_handle *fwnode;
> > @@ -176,6 +181,7 @@ struct max9286_priv {
> >  	/* The initial reverse control channel amplitude. */
> >  	u32 init_rev_chan_mv;
> >  	u32 rev_chan_mv;
> > +	u8 i2c_mstbt;
> >
> >  	struct v4l2_ctrl_handler ctrls;
> >  	struct v4l2_ctrl *pixelrate_ctrl;
> > @@ -250,6 +256,17 @@ static const struct max9286_format_info max9286_formats[] = {
> >  	},
> >  };
> >
> > +static const struct max9286_i2c_speed max9286_i2c_speeds[] = {
> > +	{ .rate =   8470, .mstbt = MAX9286_I2CMSTBT_8KBPS },
> > +	{ .rate =  28300, .mstbt = MAX9286_I2CMSTBT_28KBPS },
> > +	{ .rate =  84700, .mstbt = MAX9286_I2CMSTBT_84KBPS },
> > +	{ .rate = 105000, .mstbt = MAX9286_I2CMSTBT_105KBPS },
> > +	{ .rate = 173000, .mstbt = MAX9286_I2CMSTBT_173KBPS },
> > +	{ .rate = 339000, .mstbt = MAX9286_I2CMSTBT_339KBPS },
> > +	{ .rate = 533000, .mstbt = MAX9286_I2CMSTBT_533KBPS },
> > +	{ .rate = 837000, .mstbt = MAX9286_I2CMSTBT_837KBPS },
> > +};
> > +
> >  /* -----------------------------------------------------------------------------
> >   * I2C IO
> >   */
> > @@ -370,7 +387,7 @@ static int max9286_i2c_mux_init(struct max9286_priv *priv)
> >  static void max9286_configure_i2c(struct max9286_priv *priv, bool localack)
> >  {
> >  	u8 config = MAX9286_I2CSLVSH_469NS_234NS | MAX9286_I2CSLVTO_1024US |
> 
> Does the 'slave remote timeout' need to be changed according to speed
> as well ?

Possibly, but the documentation doesn't tell how, and the current values
for the setup/hold times and the timeout work fine in my setup, so I
haven't changed them.

> > -		    MAX9286_I2CMSTBT_105KBPS;
> > +		    priv->i2c_mstbt;
> >
> >  	if (localack)
> >  		config |= MAX9286_I2CLOCACK;
> > @@ -1320,6 +1337,8 @@ static int max9286_parse_dt(struct max9286_priv *priv)
> >  	struct device_node *node = NULL;
> >  	unsigned int i2c_mux_mask = 0;
> >  	u32 reverse_channel_microvolt;
> > +	u32 i2c_clk_freq = 105000;
> > +	unsigned int i;
> >
> >  	/* Balance the of_node_put() performed by of_find_node_by_name(). */
> >  	of_node_get(dev->of_node);
> > @@ -1410,6 +1429,23 @@ static int max9286_parse_dt(struct max9286_priv *priv)
> >  	}
> >  	of_node_put(node);
> >
> > +	of_property_read_u32(dev->of_node, "maxim,i2c-clock-frequency",
> > +			     &i2c_clk_freq);
> > +	for (i = 0; i < ARRAY_SIZE(max9286_i2c_speeds); ++i) {
> > +		const struct max9286_i2c_speed *speed = &max9286_i2c_speeds[i];
> > +
> > +		if (speed->rate == i2c_clk_freq) {
> > +			priv->i2c_mstbt = speed->mstbt;
> > +			break;
> > +		}
> > +	}
> > +
> > +	if (i == ARRAY_SIZE(max9286_i2c_speeds)) {
> > +		dev_err(dev, "Invalid %s value %u\n",
> > +			"maxim,i2c-clock-frequency", i2c_clk_freq);
> > +		return -EINVAL;
> > +	}
> > +
> 
> I never got a real answer to this question in years: should we assume
> dts are correct or should we validate them ?

That's a very good question :-) Given the DT-is-an-ABI dogma, I tend to
do validation in drivers, to avoid accepting incorrect device trees that
would work by chance and find myself unable to fix the driver later.

> >  	/*
> >  	 * Parse the initial value of the reverse channel amplitude from
> >  	 * the firmware interface and convert it to millivolts.
> > @@ -1484,10 +1520,16 @@ static int max9286_probe(struct i2c_client *client)
> >  	priv->client = client;
> >  	i2c_set_clientdata(client, priv);
> >
> > +	ret = max9286_parse_dt(priv);
> 
> Any reason why moving this so up ? Doing so just before i2c_config()
> would be enough ?

Just to avoid having to move it up again in the future if we add more
parameters to the device tree that would be needed before
max9286_configure_i2c().

> > +	if (ret)
> > +		goto err_cleanup_dt;
> > +
> >  	priv->gpiod_pwdn = devm_gpiod_get_optional(&client->dev, "enable",
> >  						   GPIOD_OUT_HIGH);
> > -	if (IS_ERR(priv->gpiod_pwdn))
> > -		return PTR_ERR(priv->gpiod_pwdn);
> > +	if (IS_ERR(priv->gpiod_pwdn)) {
> > +		ret = PTR_ERR(priv->gpiod_pwdn);
> > +		goto err_cleanup_dt;
> > +	}
> >
> >  	gpiod_set_consumer_name(priv->gpiod_pwdn, "max9286-pwdn");
> >  	gpiod_set_value_cansleep(priv->gpiod_pwdn, 1);
> > @@ -1514,10 +1556,6 @@ static int max9286_probe(struct i2c_client *client)
> >  	if (ret)
> >  		goto err_powerdown;
> >
> > -	ret = max9286_parse_dt(priv);
> > -	if (ret)
> > -		goto err_cleanup_dt;
> > -
> >  	ret = max9286_get_poc_supplies(priv);
> >  	if (ret)
> >  		goto err_cleanup_dt;
> > @@ -1528,10 +1566,10 @@ static int max9286_probe(struct i2c_client *client)
> >
> >  	return 0;
> >
> > -err_cleanup_dt:
> > -	max9286_cleanup_dt(priv);
> >  err_powerdown:
> >  	gpiod_set_value_cansleep(priv->gpiod_pwdn, 0);
> > +err_cleanup_dt:
> > +	max9286_cleanup_dt(priv);
> >
> >  	return ret;
> >  }

-- 
Regards,

Laurent Pinchart