Do Not Generate Function Call option for MX_TouchGFX_Init not working

[franck23]

Hi,

In CubeMx, Project Manager tab -> Advanced Settings, the "Do Not Generate Function Call" for MX_TouchGFX_Init is not working.

After selecting it and generating the code, the MX_TouchGFX_Init() is still present in the main.c and going back to the Advanced Settings, the previously selected check box got deselected.

This is a problem because when using the QuadSpi with TouchGFx, it needs to be mapped before TouchGFX initialization.

The mapping cannot be made in the MX_QUADSPI_Init because there is no USER CODE section at the end of this function (Hey ST, what about having a user code section at the end of EVERY HAL init function???).

Settings:

MCU: STM32H743

CubeMx: 6.1.1

CubeIDE: 1.5.1

Firmware Package: FW_H7 1.8.0

[Romain DIELEMAN]

Hi,

Thank you for reporting this. I am surprised by it as well as I could replicate the issue on projects on a STM32H743 eval board and on a H7b3 dk board. It seems that only the MX_TouchGFX_Init is not working, we will investigate more on that (might be just me not aware of a step to do). Could you add the topic "CubeMX" in the Topic tags of this post ? That way this will have more visibility.

Are you working on a custom board ? Did you start your project from scratch from CubeMX ? I am asking this because from what I understand from your message, you have the MX_TouchGFX_Init called before the MX_QUADSPI_Init ? This is not the case in the application template for the H743 eval board available on TouchGFX Designer. I believe the issue might come in your case from the "rank order" of your functions: you can see that in the Project Manager tab -> Advanced Settings --> Generated Function Calls. You can change the ranks of the function by selecting a function and pressing the "up or low arrows" on the top right (see picture)

/Romain

[franck23]

Hi Romain,

Thanks for the answer.

(CubeMX is now added to the topic).

Yes, I am working on a custom board, starting from scratch with CubeMX.

The MX_TouchGFX_Init is indeed called after the MX_QUADSPI_Init.

However, the bit of code to map the QSPI must be after MX_QUADSPI_Init and before MX_TouchGFX_Init.

The best place to add this code would be at the end of the MX_QUADSPI_Init function, but this is not possible because there is no "user code" section there.

What I have done now (and is working) is that I selected the "Do Not Generate Function Call" for the QSPI instead and moved the call to a user code section earlier in the main ("USER CODE BEGIN SysInit"). Then I do the QSPI mapping.

[Romain DIELEMAN]

So I did a bit more testing and this issue seems to happen for all functions from additional X-Cube packages (like X-Cube-TouchGFX, X-CubeBLE, X-CubeNFC4, ...).

I created an internal ticket to investigate and fix this, thank you for reporting this.

[franck23]

Thanks for checking on this issue and creating the ticket!

[KPort]

You wrote: "This is a problem because when using the QuadSpi with TouchGFx, it needs to be mapped before TouchGFX initialization."

I tried to make a STM32CubeIDE project for STM32H743 and then enabled QSPI functionality in CubeMX.

At the end (last 2 lines) of MX_QUADSPI_Init() function I can now see the user code section:

/* USER CODE BEGIN QUADSPI_Init 2 */

 /* USER CODE END QUADSPI_Init 2 */

If you put your QSPI mapping code there it will be called before MX_TouchGFX_Init() is called.

If you don't have these 2 lines, maybe you had deleted them, or you use older fw library (I used STM32Cube_FW_H7_V1.8.0).

[franck23]

Hi KPort,

Can you tell me which version of CubeMx you are using?

I do not have any USER CODE BEGIN in the MX_QUADSPI_Init() function (I do have it in the HAL_QSPI_MspInit, but that's not the right place for the mapping code).

I did a test by writing some garbage in this function and refreshing the CubeMx project. It does rewrite the MX_QUADSPI_Init(), but still no user code section

/* QUADSPI init function */
void MX_QUADSPI_Init(void)
{
 
 hqspi.Instance = QUADSPI;
 hqspi.Init.ClockPrescaler = 1;
 hqspi.Init.FifoThreshold = 4;
 hqspi.Init.SampleShifting = QSPI_SAMPLE_SHIFTING_HALFCYCLE;
 hqspi.Init.FlashSize = 23;
 hqspi.Init.ChipSelectHighTime = QSPI_CS_HIGH_TIME_1_CYCLE;
 hqspi.Init.ClockMode = QSPI_CLOCK_MODE_0;
 hqspi.Init.FlashID = QSPI_FLASH_ID_1;
 hqspi.Init.DualFlash = QSPI_DUALFLASH_DISABLE;
 if (HAL_QSPI_Init(&hqspi) != HAL_OK)
 {
 Error_Handler();
 }
 
}

[franck23]

BTW, my firmware package is FW_H7 V1.8.0.

[franck23]

OK, I found out why I don't have a user code section.

It depends if the peripheral initialization is done in the main.c or in separated files.

In CubeMx Project Manager->CodeGenerator->Generated files, if the "Generate peripheral initialization as pair '.c/.h' files per peripheral" is selected, the MX_QUADSPI_Init function does not include the user code sections!

[KPort]

Hi franck23,

I'using the latest STM32CubeIDE 1.5.1 and CubeMX 6.1.1 and FW_H7 V1.8.0.

The problem seems to be that you have selected in CubeMX the option "Generate peripheral initialization as a pair of '.c/.h' files per peripheral".

With this option it generates separate .c/.h files for each peripheral AND you don't get all the user code sections.

Without this option selected the generated code for MX_QUADSPI_Init() has all needed user code sections (but the code is inside main.c).

You could do the following: define a macro in main.c just before the HAL initializations like this:

 /* USER CODE BEGIN SysInit */

#define MX_QUADSPI_Init() MX_QUADSPI_Init(); map_my_qspi()

 /* USER CODE END SysInit */

...where map_my_qspi() is your custom code, run right after MX_QUADSPI_Init() (also declare it in section /* USER CODE BEGIN PFP */).

If you generate the code without the "Generate peripheral initialization..." flag selected, then it all becomes so much easier...

This is what gets generated without the flag (look at... /* USER CODE BEGIN QUADSPI_Init 2 */):

static void MX_QUADSPI_Init(void)

{

 /* USER CODE BEGIN QUADSPI_Init 0 */

 /* USER CODE END QUADSPI_Init 0 */

 /* USER CODE BEGIN QUADSPI_Init 1 */

 /* USER CODE END QUADSPI_Init 1 */

 /* QUADSPI parameter configuration*/

 hqspi.Instance = QUADSPI;

 hqspi.Init.ClockPrescaler = 255;

 hqspi.Init.FifoThreshold = 1;

 hqspi.Init.SampleShifting = QSPI_SAMPLE_SHIFTING_NONE;

 hqspi.Init.FlashSize = 1;

 hqspi.Init.ChipSelectHighTime = QSPI_CS_HIGH_TIME_1_CYCLE;

 hqspi.Init.ClockMode = QSPI_CLOCK_MODE_0;

 hqspi.Init.FlashID = QSPI_FLASH_ID_1;

 hqspi.Init.DualFlash = QSPI_DUALFLASH_DISABLE;

 if (HAL_QSPI_Init(&hqspi) != HAL_OK)

 {

   Error_Handler();

 }

 /* USER CODE BEGIN QUADSPI_Init 2 */

 /* USER CODE END QUADSPI_Init 2 */

}

[franck23]

Thanks for the macro tip, looks safer than using the "Do Not Generate Function Call" for the QSPI as I was doing.

I wouldn't have this issue if I was not using the "Generate peripheral initialization as pair '.c/.h'" option, but I find that it makes my code easier to maintain when using it.

Just got me in trouble this one time...

[JCici.1]

Hi,

I'm having the same problem. When I have selected the option "Generate peripheral initialization as a pair of '.c/.h' files per peripheral" in CubeMX, the function "MX_QUADSPI_Init()" in the file "quadspi.c" doesn't have an USER CODE section.

Is this behavior a bug?

Settings:

STM32CubeIDE 1.5.1

CubeMX 6.1.1

STM32CubeF7 Firmware Package V1.16.0

Thanks!