Screen and input devices hang on

Colibri T30, WCE7, BSP 2.0

Hello Toradex Team,

Today, with a Toradex board, the screen and keyboard + mouse completely hung up.

The system is also no longer in a normal state, only a reboot solves the problem.

The problem happened when I clicked through the folder folders on the SD card.

Small processes in the background continue.

What could be the problem?

I need the screen and a reboot is not allowed in my application.

Many thanks for your help.

Best regards

Saphymo

Which exact image is it? 2.0 final is currently not released. Is it 2.0 beta 4? Do you have any debug output available while this happend? Any way to reproduce the issue or it just happens once and can not be reproduced any longer?

Hello Samuel.tx.

The image is 2.0 Beta 4. I do not have a debug output.

The problem is not reproducible.

Best regards.

Saphymo

Hello Toradex Team,

I could reproduce the problem.

It seems as if the operating system manipulates individual sectors of the SD card.

A defect of the SD card as well as the controller could be launched, as different SD card types were tested by different manufacturers.

The SD card also has no defective sectors after checking.

When the bad sectors are accessed, the screen freezes.

No debugging information is written.

In my opinion it looks like a fault of the BSP / operating system.

Can you fix this?

Best Regards

Saphymo

Hello Toradex Team,

I have made new observations on phenomene.

If the T30 freezes the screen the I2C interfaces no longer operated.

I have written a test software, which detects the SD card speed every minute and reads the temperature of an I2C sensor.

If the screen freezes the SD card speed is still determined, but the temperature is not read anymore.

This means that components within the operating system are “blocked”.

The condition is permanent. Only one restart fixes the error.

Would be very nice if you could help me.

Best regards

Saphymo

Thanks for the additional information. We will check the details and will get back with our answer or further questions soon.

@Saphymo: Are you using the latest I2C Lib from the lib package 2.0? I remember we earlier had issues with the I2C Lib. The power management of the device is also done over I2C, if something goes wrong there, it could freeze the device.

One other thing you could do to check this situation: If you disable DFS, you should see the freezes less often. If this is the case and you are already using the latest libs, it is likely, there is a bug in the I2C library. Let me know your findings.

Hello Toradex Team,

Thank you for your answers.

I use the current DLL package 2.0

Best regards

Saphymo

In one of the previous posts your wrote, that the freeze happens if the SD gets corrupted. I this still true or did you see the freeze also independently?

If it is still true, do you see any output on the debug output in case of a freeze? Could it bee the application somehow blocks everything in case of a SD corruption, because it reads wrong data?

If this issue is an independent issue, could you once try to reproduce it a litte bit more deterministic as mentioned before. Is the freeze still there in case the DFS is disabled? (Disable DFS in the Colibri Monitor)

Dear Samuel.Tx,
According to current knowledge, it looks as follows:
The freezing of the operating system happens when a file on the SD card is corrupt and then a read access to this file takes place.
The files on the SD card corrupt through the I2C.dll of Toradex. If the I2C communication is reduced I have less problems. (actual .Dll)
Today, however, the whole operating system has been destroyed. Only a flashing could solve the problem.
Debug messages I can not see. (Communication via RS232).
The problem comes definitely from the operating system or components of Toradex.
DFS I now disabled. However, we need DFS because of the power consumption, this was one of the purchase decision for Toradex.

Best regards

Saphymo

@Saphymo: Thanks for this update. On our side, we also still try to identify the source of the issue. If you have any sample project which can be used to easily reproduce the issue we would be happy to get it to speed up tracking down the issue.

@Saphymo: Any update on this issue? Do you still see this issue or is there anything that resolved the issue on your side?