CSI error - explanation

Autonomous Machines Jetson & Embedded Systems Jetson Nano
1 
Hello.

I try to run adv7280-m chip on Nano platform.

As for now, I have some errors:

[ +0.201276] video4linux video0: frame start syncpt timeout!0
[ +0.005726] video4linux video0: TEGRA_VI_CSI_ERROR_STATUS 0x00000000
[ +0.000006] vi 54080000.vi: TEGRA_CSI_PIXEL_PARSER_STATUS 0x00000000
[ +0.004901] vi 54080000.vi: TEGRA_CSI_CIL_STATUS 0x00000010
[ +0.000004] vi 54080000.vi: TEGRA_CSI_CILX_STATUS 0x00000040

How to find description of these errors?
How to calculate register from TRM, related to these errors?

2

Do Nvidia manage drivers/media/platform/tegra/camera/csi/csi2_fops.c file?

How to interpret these status errors?

4

hello dimaz,

you may look into the CSI driver for general failures.
for example,

int tegra_csi_error(struct tegra_csi_channel *chan, int port_idx)
{
...
	val = pp_read(port, TEGRA_CSI_PIXEL_PARSER_STATUS);
	err |= val & 0x4000;
	pp_write(port, TEGRA_CSI_PIXEL_PARSER_STATUS, val);

	val = cil_read(port, TEGRA_CSI_CIL_STATUS);
	err |= val & 0x02;
	cil_write(port, TEGRA_CSI_CIL_STATUS, val);

	val = cil_read(port, TEGRA_CSI_CILX_STATUS);
	err |= val & 0x00020020;
	cil_write(port, TEGRA_CSI_CILX_STATUS, val);
5

Hello JerryChangis.

I know it.

In my case, TEGRA_CSI_CIL_STATUS 0x00000010

How do I understand the value 0x00000010?
Which is appropriate register in TRM with bit description?

6

please download TRM and checking [29.16.25 CSI_CSI_CIL_A_STATUS_0] and also [29.16.26 CSI_CSI_CILA_STATUS_0] for details.

7

There is no TRM inside download center among documents, related to Nano.

Screenshot from 2023-03-06 17-51-07
Screenshot from 2023-03-06 17-51-071261×872 119 KB

Could you please share it?

8

please check Tegra X1 TRM, I’ve already link the documentation in my previous post.

9

Thank you for such information.

About my issue.

This is explanation of my errors:

CILA_CTRL_ERR: Control Error. Set when CIL-A detects LP state 01 or 10 followed by a stop state (LP11)
instead of transitioning into the Escape mode or Turn Around mode (LP00).

CILA_DATA_LANE0_CTRL_ERR: Control Error. Set when CIL-A detects LP state 01 or 10 followed by a
stop state (LP11) instead of transitioning

What can be reason for it?

  1. HW issure?
  2. ADV7280-m configuration?
  3. Device tree binding?

Thank you a lot for your ideas.

10

hello dimaz,

this usually due to sensor side issues, you may examine the sensor configuration.
please dig into TRM, you may check THS_SETTLE to understand the settle time for data lane when moving from LP to HS (i.e. LP11->LP01->LP00)
camera stack attempts to auto-calibrate this by mclk_multiplier parameter, you may configured this manually by device tree property, cil_settletime. please see-also Property-Value Pairs for details.

11

Hi.

As a result, issue was related to the Nvidia SoM board, maybe it was damaged.
I’ve replaced it on another one and now it works.

After it, I ported driver and device tree to Xavier NX platform.

The image from analog input was corrupted. I’ve found, that bytesperline should be atom (64-Byte) aligned.
By default its value is 1440.

Where this value is stored? Device tree or kernel code?

12

hello dimaz,

it came from device tree property settings.
but… you may enable v4l2-ctl command to adjust the stride by --preferred_stride=<>.

13

Hello.

I can not find this property, could you please show me the place where it stored.

I attached modified device tree for adv7280-m.
tegra210-camera-rbpcv2-dual-adv7280m.dtsi (6.8 KB)

14

hello dimaz,

which Jetpack release you’re using? this control property, --preferred_stride should be included in the latest release. i.e. Jetpack-4.6.3.
you may running v4l2-ctl --all for checking.

15

Hello.

I use L4T 32.6.1 release.

But my question was, how to set this parameter via device tree?

16

And another one issue, related to CSI.

I’ve got the video, but it is “shaking” up and down


.

Could you please advise me, what it can be a reason for it?

17

hello dimaz,

--preferred_stride is the control property through command-line. it’s used to override the stride settings, which is calculated by sensor’s active width.

regarding to this “shaking” issue.
is your driver output the video frames as interleave or progressive?

18

Hello JerryChang.

I tried both video output: interleave and progressive, results are the same.

Please take a look to this video. I’ve activated border pattern on ADV side.
Maybe, it can help to understand the reason of issue.

19

hello dimaz,

may I know what’s the ground truth it is, or, what’s the sources image looks like.
is there a frame-drop? what’s the output fps results?

20

I don’t know, how source image looks like.
I’ve activated boundary box test pattern inside adv7180 chip.

Screenshot from 2023-03-14 10-41-45
Screenshot from 2023-03-14 10-41-451211×327 18.6 KB

How to check it?

I also don’t know how to check it, please advise.

21

hello dimaz,

may I know what’s your capture pipeline?
for example,
here’s gst pipeline to disable preview and shows frame-rate only.
$ gst-launch-1.0 nvarguscamerasrc sensor-id=0 ! 'video/x-raw(memory:NVMM),width=1920, height=1080, framerate=30/1, format= NV12' ! nvvidconv ! 'video/x-raw(memory:NVMM),format=I420' ! fpsdisplaysink text-overlay=0 video-sink=fakesink sync=0 -v

if you’re using gst pipeline,
please refer to above, by adding fpsdisplaysink to enable frame-rate checking.
thanks