Problems with custom bootloader on STM32F4

[Roberto Fortunato]

Posted on June 01, 2017 at 17:34

Hello,

I have some problems to implement a bootloader in a STM32F4. I searched on internet and I found coherent informations and examples, but I 'm not able to make simple test work.

Here my simple code:

bootloader (writed on 0x08000000 address):

&sharpdefine APPLICATION_ADDRESS        0x08008000

typedef void (*pFunction)(void);

pFunction appEntry;uint32_t appStack, test, i;

int main(void){

    /* Get the application stack pointer (First entry in the application vector table) */

    appStack = (uint32_t) *((__IO uint32_t*)APPLICATION_ADDRESS);

    /* Get the application entry point (Second entry in the application vector table) */

    appEntry = (pFunction) *(__IO uint32_t*) (APPLICATION_ADDRESS + 4);

    /* Reconfigure vector table offset register to match the application location */

    SCB->VTOR = APPLICATION_ADDRESS;

    /* Set the application stack pointer */

    __set_MSP(appStack);

    /* Start the application */

    appEntry();

    while(1);

    return 0;

}

program (writed on 0x08008000 address):

int main(void){

    register uint32_t tl;

    GPIO_InitTypeDef GPIO_InitStructure;

    ErrorStatus HSEStartUpStatus;    RCC_DeInit();    // Enable HSE    RCC_HSEConfig(RCC_HSE_ON);    // Wait till HSE is ready    HSEStartUpStatus = RCC_WaitForHSEStartUp();    if(HSEStartUpStatus == SUCCESS){        // HCLK = SYSCLK impostato a 168 tick        RCC_HCLKConfig(RCC_SYSCLK_Div1);        // PCLK2 = HCLK/2 perch� massimo � 84Mhz        RCC_PCLK2Config(RCC_HCLK_Div2);        // PCLK1 = HCLK/4 perch� massimo � 42Mhz        RCC_PCLK1Config(RCC_HCLK_Div4);        // PLLCLK = 168 MHz        RCC_PLLConfig(RCC_PLLSource_HSE, 12, 336, 2, 7);        // PLLCLK = 144 MHz        //RCC_PLLConfig(RCC_PLLSource_HSE, 12, 288, 2, 6);        // Enable PLL        RCC_PLLCmd(ENABLE);        // Wait till PLL is ready        while(RCC_GetFlagStatus(RCC_FLAG_PLLRDY) == RESET)        {        }        // Select PLL as system clock source        RCC_SYSCLKConfig(RCC_SYSCLKSource_PLLCLK);        // Wait till PLL is used as system clock source        while(RCC_GetSYSCLKSource() != 0x08){}        SystemCoreClockUpdate();    }    RCC_AHB1PeriphClockCmd(RCC_AHB1Periph_GPIOG, ENABLE);    GPIO_InitStructure.GPIO_Pin = GPIO_Pin_4;    GPIO_InitStructure.GPIO_Mode = GPIO_Mode_OUT;    GPIO_InitStructure.GPIO_OType = GPIO_OType_PP;    GPIO_InitStructure.GPIO_Speed = GPIO_Speed_50MHz;    GPIO_InitStructure.GPIO_PuPd = GPIO_PuPd_UP;    GPIO_Init(GPIOG, &GPIO_InitStructure);

    GPIO_SetBits(GPIOG, GPIO_Pin_4);

    while(1){

                for(tl=0; tl < 0xFFFFFF; tl++);        // about 0.6 seconds                    GPIO_ToggleBits(GPIOG, GPIO_Pin_4);    }    

    return 0;

}I write program to 0x08000000 address for test and it works, so I'm sure there is no bugs.

When I use bootloader system doesn't work. I try to understand with debugger and I see that instruction appEntry(); has no effect and next command is while(1);

Can anyone tell me where it is wrong?

ThanksRoberto#bootloader #stm-32f407

[Bogdan Golab]

Posted on January 15, 2018 at 18:25

If use debugger try to see if the interrupts are enabled. If you have right VTOR value the interrupt might be disabled. I understand that program does not crash i.e. does not go to the Hard_Fault handler...

[Carmine Gianni]

Posted on January 16, 2018 at 09:47

SOLVED!

I found that what kept my code from working was that I was trying to launch the application from an ISR. To solve the issue I simply returned from the ISR and launched the application from the main code

[Bogdan Golab]

Posted on January 16, 2018 at 10:53

Great! I just implemented my bootloader and was interesting what kind of new issues we can have - different than I came across

[varani luciano]

May be you have solved the problem, but...

The question is ( I think): when the application starts at APPLICATION_ADDRESS+4,calls the Reset_Handler that activate the function SystemInit () . This function (as you can read) do

SCB->VTOR = FLASH_BASE | VECT_TAB_OFFSET; (to set vector table address)

but in the application project you have

#define VECT_TAB_OFFSET 0x00 (in system_stm32f4xx.c) and

#define FLASH_BASE           0x08000000U (in my case in stm32f429xx.h).

So now the vector table is at 0x8000000 (this is booloader! not application....)

The boootloader with

 SCB->VTOR =APP_ADDRESS;

do a useless thing.

The solution I find? may be inserting in the application code NVIC_SetVectorTable(NVIC_VectTab_FLASH, 0x8000);

or

 SCB->VTOR =APP_ADDRESS;

or (I do that) modifyng the file system_stm32f4xx.c

//#define VECT_TAB_OFFSET 0x00

#define VECT_TAB_OFFSET  0x8008000 (or other application APP_ADDRESS).

I hope that a new option in cubeMX (may be there is), ask me the starting address of the application and modify the .ld file and the system_stm32f4xx.c (or the Reset_Handler code).

Ciao!

[Tesla DeLorean]

You know you can just use the symbol table entry for the vector table that the linker sets to the base of the image as it fixes it for a specific address. No smoke and mirrors, it knows exactly where the image is built to function.