Issue with srand() causing exit in STM32H747 Project with FreeRTOS and lwIP

[OskarP]

Hello everyone,

I’m encountering an issue with the srand() function in my STM32H747 project, and I would appreciate your insights on what might be causing this problem.

Hardware:

• MCU: STM32H747IIT6
• Custom board

Software setup:

• CubeMX-generated project
• Cube CLT version: 1.17
• Firmware version: 1.12.1

Problem Description

Whenever I call the srand() function, my code immediately ends up in the exit handler.

To troubleshoot, I created a minimal project with no additional middleware (bare minimum setup), and in that setup, the srand() function works without any issues.

However, in my main project where I need both lwIP and FreeRTOS, calling srand() causes an exit. The project was created following this guide:
How to Create a Project for STM32H7 with Ethernet and lwIP Stack.

I made one modification to the guide: instead of using the default OS wrapper, I selected CMSIS_V2.

Currently, I am only using the M7 core.

What I’ve Tried So Far

• Minimal project without lwIP and FreeRTOS -> srand(12345) works fine.
• Project with lwIP and FreeRTOS -> srand(12345) causes an exit.

I’ve attached an image showing the callstack where the issue occurs. I hope this helps in diagnosing the problem.

Here’s a link to my code: GitHub Repository

Does anyone have any ideas on what might be causing this issue? Could it be related to the use of CMSIS_V2, a conflict with FreeRTOS, or possibly a memory issue?

Thank you in advance for your help!

Best regards,
Oskar

[SofLit]

Hello @OskarP and welcome to the community,

1- I didn't see any srand() call in the project you shared in github.

2- Could you please try with STM32CubeIDE and check if you have the issue?

[KnarfB]

__assert_func should (try to) write soemthing to stderr. You may catch the output by setting a breakpoint in your/default _write syscall implementation.

my guess:

NOTES

<<rand>> and <<srand>> are unsafe for multi-threaded applications.

<<rand_r>> is thread-safe and should be used instead.

PS: Does ST provide easily obtainable (and buildable) official libc sources to reproduce such potential libc issues?

I use GitHub - bminor/newlib: Unofficial mirror of sourceware newlib repository. Updated daily. which should be close.

hth

KnarfB

[OskarP]

Hello SofLit,

Thank you for your quick response!

I call the srand() function in the main.c file of the CM7, specifically at line 362.
I tried building the project using Cube IDE with the same .ioc file. With this setup, the srand() function works, but I encounter a hard fault, as shown in the attached images.

I noticed that the flash script is slightly different. There is a second definition of ._user_heap_stack added in the flash.ld:

._user_heap_stack :  
{  
  . = ALIGN(8);  
  PROVIDE ( end = . );  
  PROVIDE ( _end = . );  
  . = . + _Min_Heap_Size;  
  . = . + _Min_Stack_Size;  
  . = ALIGN(8);  
} > RAM_D1

After copying the new flash.ld file, the srand() function call works in the project generated by CubeMX as well.

[OskarP]

Hello KnarfB,

Thank you for your suggestions. I believe lwIP uses the rand() function, so I need to call srand() to initialize the seed value. I will make sure to call it before starting FreeRTOS. That should resolve the issue with being thread unsafe at least for the srand() call.

Best regards,

OskarP

[SofLit]

---

@OskarP wrote:

I call the srand() function in the main.c file of the CM7, specifically at line 362.

Yes because I regenerated your code and you put srand() outside USER CODE START / USER CODE END location.

Anyway ..

Running your example with CubeIDE I got that hardfault but nothing related to srand() call. The hardfault is caused by  MX_SDMMC1_SD_Init() and may be the subsequent calls.

I commented out these lines and no more hardfalt occured again:

MX_SDMMC1_SD_Init();
MX_USB_OTG_HS_PCD_Init();
MX_FATFS_Init();

[OskarP]

Sorry SofLit, that one's on me. However, even if I comment out the three mentioned function calls, I still hit the hard fault at the same location. Since this is a new topic for me, do you have any suggestions on how I can debug this further?

Best regards

OskarP.

[SofLit]

---

@OskarP wrote:

However, even if I comment out the three mentioned function calls, I still hit the hard fault at the same location.

---

Did you use CubeIDE for that?

Just made a change in CubeMx to run the example on Nucleo board. Decreased the system clock to 400MHz, and selected SMPS as power config and I'm reaching the osDelay(1) in:

void StartDefaultTask(void *argument)
{
  /* init code for LWIP */
	  srand(12345);
  MX_LWIP_Init();
  /* USER CODE BEGIN 5 */
  /* Infinite loop */
  for(;;)
  {
    osDelay(1);
  }
  /* USER CODE END 5 */
}

I don't think these changes impact the functionality.

I'am attaching the project where I reverted the modifications for you (I just kept the system at 400MHz). Please test again with CubeIDE.