Re: [PATCH v2 2/2] media: i2c: IMX296 camera sensor driver

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Laurent,

On Tue, Dec 21, 2021 at 05:56:54PM +0200, Laurent Pinchart wrote:

,,,

> > > +static int imx296_ctrls_init(struct imx296 *sensor)
> > > +{
> > > +	struct v4l2_fwnode_device_properties props;
> > > +	unsigned int hblank;
> > > +	int ret;
> > > +
> > > +	ret = v4l2_fwnode_device_parse(sensor->dev, &props);
> > > +	if (ret < 0)
> > > +		return ret;
> > > +
> > > +	v4l2_ctrl_handler_init(&sensor->ctrls, 9);
> > > +
> > > +	v4l2_ctrl_new_std(&sensor->ctrls, &imx296_ctrl_ops,
> > > +			  V4L2_CID_EXPOSURE, 1, 1048575, 1, 1104);
> > > +	v4l2_ctrl_new_std(&sensor->ctrls, &imx296_ctrl_ops,
> > > +			  V4L2_CID_ANALOGUE_GAIN, IMX296_GAIN_MIN,
> > > +			  IMX296_GAIN_MAX, 1, IMX296_GAIN_MIN);
> > > +
> > > +	/*
> > > +	 * Horizontal blanking is controlled through the HMAX register, which
> > > +	 * contains a line length in INCK clock units. The INCK frequency is
> > > +	 * fixed to 74.25 MHz. The HMAX value is currently fixed to 1100,
> > 
> > It seems the driver supports other values, too. Shouldn't this be the
> > actual frequency?
> 
> That's not clear to me from the documentation I have access to :-( It's
> quite convoluted, there are a few examples from which I tried to infer
> what was going on, but no clear explanation. My board uses a fixed clock
> frequency of 37.125MHz so I can't test other values.
> 
> Can we start with this and update it later if we can figure out more
> (assuming there's an issue, it may actually be correct already) ?

Sounds reasonable. I was just wondering.

> 
> > > +	 * convert it to a number of pixels based on the nominal pixel rate.
> > > +	 */
> > > +	hblank = 1100 * 1188000000ULL / 10 / 74250000
> > > +	       - IMX296_PIXEL_ARRAY_WIDTH;
> > > +	sensor->hblank = v4l2_ctrl_new_std(&sensor->ctrls, &imx296_ctrl_ops,
> > > +					   V4L2_CID_HBLANK, hblank, hblank, 1,
> > > +					   hblank);
> > > +	if (sensor->hblank)
> > > +		sensor->hblank->flags |= V4L2_CTRL_FLAG_READ_ONLY;
> > > +
> > > +	sensor->vblank = v4l2_ctrl_new_std(&sensor->ctrls, &imx296_ctrl_ops,
> > > +					   V4L2_CID_VBLANK, 30,
> > > +					   1048575 - IMX296_PIXEL_ARRAY_HEIGHT,
> > > +					   1, 30);
> > > +	/*
> > > +	 * The sensor calculates the MIPI timings internally to achieve a bit
> > > +	 * rate between 1122 and 1198 Mbps. The exact value is unfortunately not
> > > +	 * reported, at least according to the documentation. Report a nominal
> > > +	 * rate of 1188 Mbps as that is used by the datasheet in multiple
> > > +	 * examples.
> > > +	 */
> > > +	v4l2_ctrl_new_std(&sensor->ctrls, NULL, V4L2_CID_PIXEL_RATE,
> > > +			  1122000000 / 10, 1198000000 / 10, 1, 1188000000 / 10);
> > 
> > What about the link frequency?
> > 
> > Is this value constant for the sensor? Or should there be a list of
> > hardware supported link frequencies?
> 
> It seems to be constant, but again the documentation is fairly unclear.

Ack.

...

> > > +static int __maybe_unused imx296_runtime_resume(struct device *dev)
> > > +{
> > > +	struct i2c_client *client = to_i2c_client(dev);
> > > +	struct v4l2_subdev *subdev = i2c_get_clientdata(client);
> > > +	struct imx296 *sensor = to_imx296(subdev);
> > > +
> > > +	return imx296_power_on(sensor);
> > > +}
> > > +
> > > +static int __maybe_unused imx296_runtime_suspend(struct device *dev)
> > > +{
> > > +	struct i2c_client *client = to_i2c_client(dev);
> > > +	struct v4l2_subdev *subdev = i2c_get_clientdata(client);
> > > +	struct imx296 *sensor = to_imx296(subdev);
> > > +
> > > +	imx296_power_off(sensor);
> > > +
> > > +	return 0;
> > 
> > I'd merge these two with imx296_power_o{n,ff}.
> 
> That would require calling imx296_runtime_resume() and
> imx296_runtime_suspend() in probe() and remove(), which I don't really
> like. I'd prefer keeping the functions separate.

You could keep calling the functions imx296_power_o{n,ff}. There's really
no need for two pairs of functions doing the same things.

...

> > > +		dev_warn(&adapter->dev,
> > > +			 "I2C-Adapter doesn't support I2C_FUNC_SMBUS_BYTE\n");
> > > +		return -EIO;
> > > +	}
> > > +
> > > +	sensor = devm_kzalloc(&client->dev, sizeof(*sensor), GFP_KERNEL);
> > > +	if (!sensor)
> > > +		return -ENOMEM;
> > > +
> > > +	sensor->dev = &client->dev;
> > > +
> > > +	mutex_init(&sensor->lock);
> > 
> > You could simplify error handling a little by moving mutex init later. Up
> > to you.
> 
> That's right, but if you don't mind I'd prefer keeping it here, to have
> ass the "static" initialization of "generic" members at the top.

Sure.

-- 
Regards,

Sakari Ailus



[Index of Archives]     [Linux Input]     [Video for Linux]     [Gstreamer Embedded]     [Mplayer Users]     [Linux USB Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [Yosemite Backpacking]

  Powered by Linux