Buggy SystemClock_Config()

[nicolas]

I use STM32CubeIDE 1.13.0 to generate code for NUCLEO-L476RG.

The generated SystemClock_Config() put flash latency configuration after the clock configuration. I don't know why it works for the first call, but if I call it after exiting stop mode, it make some weird error.

If i put the flash latency configuration after the voltage configuration and before clock configuration, it works correctly.

Is there a reason to put the flash latency configuration after the clock configuration ?

Would it be better to generate code that put the flash latency configuration before the clock configuration so we can use the same function for exiting stop mode ?

[Sarra.S]

Hello @nicolas, 

>>but if I call it after exiting stop mode, it makes some weird error.

Could you please specify/show the error 

Actually, when changing the CPU frequency, depending on increasing or decreasing it, there is a software sequence that must be applied in order (check RM0411, the Read access latency section also PWR Control section 5.1.8)

[nicolas]

IIRC, sometime it was hardfaulting, sometime closing my debugger.

I didn't found the RM0411, but i found the RM0351 instead.

There is also a sequence chapter 3.3.3. But the problem is the code generated by ST tools doesn't follows those sequence so it can sometime fail.

[nicolas]

Apparently my error is similar to this one

I was wrongly thinking it was from flash config. My program crash only when i'm using a clock of 64 MHz or higher but works with 56 MHz.

[nicolas · Accepted solution]

My problem was coming by using HAL_PWR_EnterSLEEPMode(PWR_LOWPOWERREGULATOR_ON, PWR_SLEEPENTRY_WFI) without lowering the frequency, it was messing my SystemClock_Config() after exiting stop mode.

The problem is resolved by using the PWR_MAINREGULATOR_ON for the sleep mode.