Nucleo-H7S3L8 Booting to Application via Bootloader

[bashira]

Hi all,

I have a Nucelo-H7S3L8 board running the demo application. The project has 2 applications (boot, and application). The boot is supposed to jump to the application, which is located in the external MX25UM25645GXDI00 memory.

However when debugging the boot application the BOOT_Application() function returns "BOOT_ERROR_MAPPEDMODEFAIL". My configuration for the MX25UM25645GXDI00 is attached to this post.

Any hints or guidance will be greatly appreciated.

[KDJEM.1]

Hello @bashira and welcome to the Community 🙂,

May Template_XIP example can help you.

For that, I advise you to take a look at this example and get inspired from the .ioc file and the readme.

I hope this help you.

Thank you.

Kaouthar

[bashira]

Hi Kaouthar,

Thanks for your replay.

I have had a chance to match my settings in the ioc file with the "Template_XIP" ioc settings. Unfortunately, I have had no luck with my problem. I am debugging the external memory with the  "ExtMem_manager" application but no luck.

[KDJEM.1]

Hi @bashira ,

Could you please check "MX25UW25645GXDI00" memory configuration.

Are you able to run this example Template_XIP ?

Note that, this project is targeted to run on STM32H7S3xx device on NUCLEO-H7S3L8 board from STMicroelectronics. 

This project runs from the external Flash memory. It is launched from a first boot stage and inherits from this boot project configuration (caches, MPU regions [regions 0 and 1], system clock at 600 MHz and external memory interface at the highest speed). Note that the boot part is automatically downloaded from the IDE environment via the board project Templates/Template_XIP/Binary/Boot_XIP.hex file.

Thank you.

Kaouthar

 

[Malvi2024]

I must say that I have found the same problem with Template_XIP.
Debugging boot project, the function BOOT_Application() fails with BOOT_ERROR_MAPPEDMODEFAIL.
I have seen that this comes from HAL_XSPI_Command() making some checks, then setting hxspi->ErrorCode = HAL_XSPI_ERROR_INVALID_SEQUENCE and returning HAL_ERROR.
Additionally I have also tried to replace hxspi2.Init.MemorySize = HAL_XSPI_SIZE_32GB with HAL_XSPI_SIZE_256MB (256Mbits), but no luck.

[bashira]

I can confirm this @Malvi2024 

The function in question is:
HAL_StatusTypeDef HAL_XSPI_Command(XSPI_HandleTypeDef *hxspi, XSPI_RegularCmdTypeDef *const pCmd, uint32_t Timeout)

The state gets set to 512 somehow. I have not had time to debug this further, but I know that state is NOT valid after the init function MX_EXTMEM_MANAGER_Init()

@KDJEM.1I just made a new project for the Nucleo-H7S3L8 board. Below is how everything is configured. I have not changed anything.

Thank you for the help.

[bashira]

@Malvi2024I just had a little debugging session and I found out that the invalid state is set at this point:

inside the HAL_XSPI_Read() function there is a XSPI_Command() followed by XSPI_Recieve(). The recieve function seems to timeout.

Here is my clock config.

Not sure, how to move forwards.

Thanks for the help guys.

[atosun]

I have the same board and encountered the same problem @bashira described.

Help would be massively appreciated.

[KDJEM.1 · Accepted solution]

Hello @bashira, @Malvi2024 and @atosun ,

Thank you for this clarification.

Could you please try to use the STM32CubeProgrammer to check the value of XSPI2-HSLV option byte.

To achieve higher frequency the XSPI2-HSLV must be "Checked" as mentioned in the readme.

Please let me know if the issue is solved or not.

Thank you for your contribution.

Kaouthar

[bashira]

Hi Kaouthar,

That definitely seemed to have solved the problem for me. I will do some intensive testing later on.

Thank you very much for the help 🙂