Re: [RFC PATCH v5 13/14] media: tegra-video: Add CSI MIPI pads calibration

[Sowjanya Komatineni <skomatineni@nvidia.com>]

On 7/28/20 3:30 AM, Dmitry Osipenko wrote:
> 27.07.2020 23:57, Sowjanya Komatineni пишет:
>> +	/*
>> +	 * TRM has incorrectly documented to wait for done status from
>> +	 * calibration logic after CSI interface power on.
>> +	 * As per the design, calibration results are latched and applied
>> +	 * to the pads only when the link is in LP11 state which will happen
>> +	 * during the sensor stream-on.
>> +	 * CSI subdev stream-on triggers start of MIPI pads calibration.
>> +	 * Wait for calibration to finish here after sensor subdev stream-on
>> +	 * and in case of sensor stream-on failure, cancel the calibration.
>> +	 */
>>   	subdev = on ? src_subdev : csi_subdev;
>>   	ret = v4l2_subdev_call(subdev, video, s_stream, on);
>> -	if (ret < 0 && ret != -ENOIOCTLCMD)
>> +	if (ret < 0 && ret != -ENOIOCTLCMD) {
> I assume -ENOIOCTLCMD means that camera wasn't turned ON, so why
> -ENOIOCTLCMD is special?
No -ENOIOCTLCMD mean subdev don't have s_stream ops
>
>> +		if (on && csi_chan->mipi)
>> +			tegra_mipi_cancel_calibration(csi_chan->mipi);
>>   		return ret;
>> +	}
>> +
>> +	if (on && csi_chan->mipi) {
> Does finish_calibration() really need to be called for ret=-ENOIOCTLCMD?
>
> Shouldn't it be cancel_calibration( for the -ENOIOCTLCMD?

start calibration happens during csi sensor streaming which happens 
prior to this point.

In case if sensor subdev does not have s_stream ops, then either 
finish/cancel calibration should happen to disable the clock.

>
>> +		ret = tegra_mipi_finish_calibration(csi_chan->mipi);
>> +		if (ret < 0)
>> +			dev_err(csi_chan->csi->dev,
>> +				"MIPI calibration failed: %d\n", ret);
> Doesn't v4l2_subdev_call(OFF) need to be invoked here on error?

Not required as on error streaming fails and runtime PM will turn off 
power anyway.

Also we only did csi subdev s_stream on and during sensor subdev 
s_stream on fail, actual stream dont happen and on tegra side frame 
capture by HW happens only when kthreads run.
>> +		return ret;
>> +	}
>>   
>>   	return 0;
>>   }