STM32G070 software reset issue

[aravindpv]

I was doing reset through SPI command and expected results are not happening.

this is the line i was expecting to reset the MCU

SCB->AIRCR = (NVIC_AIRCR_VECTKEY | (SCB->AIRCR & (0x700)) | (1 << NVIC_SYSRESETREQ));

where

#define NVIC_AIRCR_VECTKEY (0x05FA << 16)

#define NVIC_SYSRESETREQ    2

also my watchdog is running and expected to reset the system.
code snippet

FLASH_Unlock();
FLASH_ErasePage(0x08013000);
FLASH_ProgramWord(0x08013000, (uint64_t)0x0000000000000000ULL);
FLASH_Lock();

FLASH_Unlock();
FLASH_ErasePage(0x08013008);
FLASH_Lock();

SCB->AIRCR = (NVIC_AIRCR_VECTKEY | (SCB->AIRCR & (0x700)) | (1 << NVIC_SYSRESETREQ));
__DSB();

while (1);

---

Edited to apply proper source code formatting - please see How to insert source code for future reference.

[Andrew Neil]

Please give some more details - see: How to write your question to maximize your chances to find a solution

---

@aravindpv wrote:

I was doing reset through SPI command and expected results are not happening..

---

So what, exactly, were you expecting to happen ?

What, exactly, is actually happening ?

What investigation/testing/debugging have you done to find the issue?

---

@aravindpv wrote:

this is the line i was expecting to reset the MCU.

---

Is that line ever executed?

Does it work without all the Flash stuff?

Without the SPI?

[aravindpv]

i was running a bootloader + main app setup on mcu where i expecting a particular command over SPI to capture one of my else condition and which is properly triggering. And on that condition i have added a reset line like mentioned above which is not happening or code executing but stuck at somewhere that kind of result I am getting. Actually my plan was to reset the mcu and set to start from 0x08000000 were my bootloader lies.

[mƎALLEm]

Hello,

Why complicate things to yourself? and why simply you don't use HAL_NVIC_SystemReset() to reset your MCU? You're already using HAL with the flash based on your shared code!

[Andrew Neil]

You didn't answer whether your reset line is ever getting executed?

---

@aravindpv wrote:

 i have added a reset line like mentioned above .

---

What do you mean, "like" ?

Why is that line doing a read-modify-write on SCB->AIRCR ?

Shouldn't it be just a write?

[aravindpv]

void Execute_write()
{
    for (uint8_t byteCounter = C_DATA_WR_OFFSET_BEGIN; byteCounter < C_DATA_ARRAY_LEN; byteCounter++)
    {
        uint8_t offset = byteCounter - C_DATA_WR_OFFSET_BEGIN;

        if (modified_app_data[offset] != 1)
            continue;

        uint8_t value = app_data[byteCounter];

        // RTC Write: set bits 0–5, and write if all are set
        if (byteCounter >= OFFSET_SET_TIME && byteCounter < OFFSET_SET_TIME + 6)
        {
            rtc_time_updated |= (1 << (byteCounter - OFFSET_SET_TIME));
            if (rtc_time_updated == 0x3F)
            {
                RTC_Write(&app_data[OFFSET_SET_TIME]);
                rtc_time_updated = 0;
            }
        }

        // PMIC Register Write
        else if (byteCounter >= OFFSET_PMIC_REG && byteCounter < OFFSET_PMIC_REG + 20)
        {
            PMIC_Write_Registers(byteCounter - OFFSET_PMIC_REG, value);
        }

        // Clear Tamper Data
        else if (byteCounter == OFFSET_CLR_TV_TAMPER_TIME)
        {
            Clear_DataArray(&app_data[OFFSET_LAST_TV_TAMPER_TIME], 6);
        }
        else if (byteCounter == OFFSET_CLR_CHASSIS_TAMPER_TIME)
        {
            Clear_DataArray(&app_data[OFFSET_LAST_CHASSIS_TAMPER_TIME], 6);
        }

        // Clear SOS Alarm
        else if (byteCounter == OFFSET_SOS_ALARM_CLR)
        {
            app_data[OFFSET_SOS_ALARM] = 0x00;
        }

        // Handle Buzzer Timings
        else if (byteCounter == OFFSET_BUZZ_DUR || byteCounter == OFFSET_BUZZ_DUR + 1)
        {
            uint16_t duration = (app_data[OFFSET_BUZZ_DUR + 1] << 8) | app_data[OFFSET_BUZZ_DUR];
            if (duration < 15000)
            {
                buzzer_systick_ctr = duration;
                turn_buzzer_off = 0;
            }
        }
        else if (byteCounter == OFFSET_BUZZ_ONTIME || byteCounter == OFFSET_BUZZ_ONTIME + 1)
        {
            uint16_t duration = (app_data[OFFSET_BUZZ_ONTIME + 1] << 8) | app_data[OFFSET_BUZZ_ONTIME];
            if (duration < 5000)
                buzzer_ontime_ctr = duration;
        }
        else if (byteCounter == OFFSET_BUZZ_OFFTIME || byteCounter == OFFSET_BUZZ_OFFTIME + 1)
        {
            uint16_t duration = (app_data[OFFSET_BUZZ_OFFTIME + 1] << 8) | app_data[OFFSET_BUZZ_OFFTIME];
            if (duration < 15000)
                buzzer_offtime_ctr = duration;
        }

        // LED Blink Durations
        else if (byteCounter == OFFSET_LED1_BLINK_DUR || byteCounter == OFFSET_LED1_BLINK_DUR + 1)
        {
            led1_duration = (app_data[OFFSET_LED1_BLINK_DUR + 1] << 8) | app_data[OFFSET_LED1_BLINK_DUR];
        }
        else if (byteCounter == OFFSET_LED2_BLINK_DUR || byteCounter == OFFSET_LED2_BLINK_DUR + 1)
        {
            led2_duration = (app_data[OFFSET_LED2_BLINK_DUR + 1] << 8) | app_data[OFFSET_LED2_BLINK_DUR];
        }
        else if (byteCounter == OFFSET_ENTER_OTA && value == 0x5E)
        {

            FLASH_Unlock();
            FLASH_ErasePage(C_PRG_FORCE_OTA); // newly added

            FLASH_ProgramWord(C_PRG_FORCE_OTA, (uint64_t)0x0000000000000000ULL);

            FLASH_Lock();
            FLASH_Unlock();
            FLASH_ErasePage(C_PRG_FORCE_RPI_POWER_CYCLE);
            FLASH_Lock();

            dbglog(0x2B);
            dbglog(0x2E);
//SCB->AIRCR = (0x5FAUL << SCB_AIRCR_VECTKEY_Pos) | SCB_AIRCR_SYSRESETREQ_Msk;
//HAL_NVIC_SystemReset();
            SCB->AIRCR  = (NVIC_AIRCR_VECTKEY | (SCB->AIRCR & (0x700)) | (1 << NVIC_SYSRESETREQ));
            __DSB();
            while (1);
        }

        modified_app_data[offset] = 0;
    }
}

this is my code base. I already tried these all ways which didn't worked.

First, to answer whether the reset line is getting executed: yes, it reaches that SCB->AIRCR write (confirmed by toggling a GPIO just before it in debug). After the write + __DSB(), the code hits the while(1) and just hangs there – no actual reset happens. Removing the flash erase/program parts doesn't change anything, the reset still doesn't trigger.

Regarding the read-modify-write on AIRCR: I copied that pattern from older examples. But i also tried this HAL_NVIC_SystemReset() which also didnt helped.

observations: - The watchdog is running (IWDG), which was initially resetting the whole program due to less heap allocation which solved allocating enough Heap

Thanks..

[mƎALLEm]

---

@aravindpv wrote:

Regarding the read-modify-write on AIRCR: I copied that pattern from older examples. But i also tried this HAL_NVIC_SystemReset() which also didnt helped.

---

How did you know if the reset wasn't performed? Did you set a breakpoint at HAL_NVIC_SystemReset()?

Did you probe the NRST with an oscilloscope and check if the reset was/wasn't asserted?

[aravindpv]

There are printf() calls very early in main() (e.g., "*****BOOTLOADER*****", some meta data's, etc.). - These are redirected via __io_putchar() to USART1 (custom low-level UART_Init()). - After software reset from the app (or even from bootloader itself), **no UART output appears**. - But on cold power-on (full power cycle), the printf logs show up fine.
also,

I2C not re-enumerating on Raspberry Pi: - Bootloader calls I2C_Init() (clock enable, reset via RCC, timing, OAR1 set to 0x30<<1, interrupts enabled). - But after reset, the RPi no longer sees the slave at 0x30 (i2cdetect shows nothing).
That's how I confirmed the control didn't reaching back to bootloader.

[mƎALLEm]

I suggest to check if there was a pulse on NRST pin after calling HAL_NVIC_SystemReset(). Frankly I don't rely on printf/traces..

Check if you really calling HAL_NVIC_SystemReset(), toggle a pin just before that call, and toggle another pin just at the beginning of your program to check if the reset has been performed.

[TDK]

If HAL_NVIC_SystemReset doesn't reset the chip perhaps you have NRST tied high. Can't think of any other reason for it not working.

> But after reset, the RPi no longer sees the slave at 0x30

So it is resetting just not doing what you expect after reset?