function assert_failed() empty after STM32CubeMx regeneration

[gk]

Posted on April 25, 2014 at 10:09

Hello

Problem with code in assert_failed() in main.c when using STM32CubeMX for F4 The assert_failed() is not in protected USER CODE BEGIN .. USER CODE END block - it is intended? ---------------------------------------------------------------- file main.c generated with STM32Cube : ... /* USER CODE BEGIN 4 */ /* USER CODE END 4 */ &sharpifdef USE_FULL_ASSERT /**    * @brief Reports the name of the source file and the source line number    * where the assert_param error has occurred.    * @param file: pointer to the source file name    * @param line: assert_param error line source number    * @retval None    */ void assert_failed(uint8_t* file, uint32_t line) {   /* User can add his own implementation to report the file name and line number,     ex: printf(''Wrong parameters value: file %s on line %d\r\n'', file, line) */ } &sharpendif ---------------------------------------------------------------- another problem is with /* &sharpdefine USE_FULL_ASSERT    1 */ in stm32f4xx_hal_conf.h 1. If I uncomment &sharpdefine USE_FULL_ASSERT 1 in stm32f4xx_hal_conf.h  and implement code in USER CODE 4 block: /* USER CODE BEGIN 4 */ void assert_failed(uint8_t* file, uint32_t line) { ....} /* USER CODE END 4 */ then compiler raise error: redefinition of 'assert_failed' because of second empty declaration assert_failed in &sharpifdef USE_FULL_ASSERT .. &sharpendif block 2. If I implement code in &sharpifdef USE_FULL_ASSERT .. &sharpendif block, then after regeneration is empty (deleted any previous code) What is solution? #stm32cube-assert_failed

[jarno]

Posted on April 25, 2014 at 13:24 one workaround... uncomment from stm32f4xx_hal_conf.h #define USE_FULL_ASSERT    1 and add into main.c: /* USER CODE BEGIN 4 */   void assert_failed(uint8_t* file, uint32_t line) {         // implementation     }          #undef USE_FULL_ASSERT /* USER CODE END 4 */ worked for me.

[gk]

Posted on April 25, 2014 at 15:06

Thanks for workaround, it works.

Some improvement - add flag to compiler -D USE_FULL_ASSERT=1 and compile main.c as last one to avoid overwite stm32f4xx_hal_conf.h when regenerated. But I really wish to know if its intended or is it just bug. Now I see the problem is more general, for example wrinting anything in stm32f4xx_it.c after regeneration is lost. Is there any solution to overcome this problem? or just each time copy and pase ?

[neo911]

Posted on May 01, 2014 at 10:28

But I really wish to know if its intended or is it just bug.

Smells like a bug... Otherwise, I agree with you and would like to know the reason why. If my understanding of the User Manual (UM1718, Appendix B.1, ''STM32CubeMX generated C C code and user sections'') is correct, that part should be protected by a ''user section'' and a predefined one (by Mr ST), as the user cannot add new ones.

Only the user sections defined by STM32CubeMX are preserved. User created sections will be ignored and lost at next C code generation. (source: UM1718)

Meaning that if a USER SECTION is not planned at the right place, we are a little bit stuck...

Finally, the assert_failed code should probably look like:

void assert_failed(uint8_t* file, uint32_t line)

{ /* USER CODE BEGIN x */   /* User can add his own implementation to report the file name and line number,     ex: printf(''Wrong parameters value: file %s on line %d\r\n'', file, line) */ /* USER CODE END x */ } Rgds.

From: k.g

Posted: Friday, April 25, 2014 5:08 PMSubject: function assert_failed() empty after STM32CubeMx regeneration

Thanks for workaround, it works.

Some improvement - add flag to compiler -D USE_FULL_ASSERT=1 and compile main.c as last one to avoid overwite stm32f4xx_hal_conf.h when regenerated. But I really wish to know if its intended or is it just bug. Now I see the problem is more general, for example wrinting anything in stm32f4xx_it.c after regeneration is lost. Is there any solution to overcome this problem? or just each time copy and pase ?