HAL_RTC_Init() looses time

[thomask77]

Describe the set-up I'm using a STM32H743 on a custom board with 32768 Hz LSE crystal and battery-backed up RTC.

Describe the bug Every time HAL_RTC_Init() is called, the RTC looses some sub-seconds.

HAL_RTC_Init should check if the RTC is already initializied (year != 0) and/or matches the desired configuration, and skip unnecessary stopping of the clock. According to the data sheet 45.3.8, the library should also check the INITS flag in the RTC_ISR register.

This is big a problem, because CubeMX5 will emit code that calls HAL_RTC_Init on each reset, without checking if it's already initialized. On each reset, you will loose up to 1s.

How To Reproduce

1. Read time with HAL_RTC_GetTime &HAL_RTC_GetDate and print to console
2. Call HAL_RTC_Init
3. Repeat

While doing this, the clock will gradually loose seconds.

[ALABSTM]

Hi Thomask77,

Thank you for this one more contribution. This issue has also been reported by the ST Community. We are already working on it. We will let you updated as soon as there is something new from our development teams.

With regards,

[ALABSTM]

ST Internal Reference: 126471

[ALABSTM]

Hi Thomask77,

According to our development teams, the issue is at MX initialization-level not at HAL one. The suggested solution looks like the following:

• At user application-level, RTC initialization shall be conditioned:

+/* Do not re-initialize RTC in case it has already been to avoid loosing seconds at each reset */
+if ((RCC->BDCR & RCC_BDCR_RTCEN) == 0)
+{
  if(HAL_RTC_Init(&RtcHandle) != HAL_OK)
  {
    /* Initialization Error */
    Error_Handler(); 
  }
+}
+else
+{
+  HAL_PWR_EnableBkUpAccess();
+}

• These line of codes shall be added into the main() function of your example or application (as the one below, for instance): https://github.com/STMicroelectronics/STM32CubeH7/blob/3d3ffbe6e15fc7e4d8a4d275e9fb3a334e31c8a3/Projects/NUCLEO-H743ZI/Examples/RTC/RTC_Alarm/Src/main.c#L95-L99

We hope these details help you solve the issue from your side.

This issue has been reported to the CubeMX team. A fix will be available in a future CubeMX release. Unfortunately, we cannot share a date for the moment.

As this is a CubeMX-related issue and not directly related to the firmware published whithin this repository, please allow me to close it. Thank you for your comprehension and thank you again for your contribution.

With regards,

[ALABSTM]

Hi @thomask77,

After further discussions with our development teams, it has been agreed this point could be handled at driver's level. Hence, I will reopen this issue pending a fix is published.

With regards,

[ALABSTM]

Hi @thomask77,

As for the fix, it will consist in checking the RTC_ISR_INITS bit, which, as described in the reference manual:

• is equal to 0 if the calendar has not been initialized
• is equal to 1 if the calendar has been initialized

Thus, the fix should be implemented in hal_rtc.c rather than in main.c and would look something like this:

HAL_StatusTypeDef HAL_RTC_Init(RTC_HandleTypeDef *hrtc)
{
  ...
+    /* Check whether the calendar needs to be initialized */
+    if (RTC_ISR_INITS == 0U)
+    {
       /* Disable the write protection for RTC registers */
       __HAL_RTC_WRITEPROTECTION_DISABLE(hrtc);
       
       /* Enter Initialization mode */
       status = RTC_EnterInitMode(hrtc);
       ...
+    }
}

With regards,

[JudicaelK]

Hi,

Thanks for the solution, it works but the function getTime sometimes retuns 00:00:00.

Any idea?

[PG430]

Hi,

Thanks for providing solution. Applied same solution but still losing 1 second. every 5 second printing "Wakeup..." string on terminal and soft reset on every 1 minute Logs are attached.

[ALABSTM]

Hi @Judicaelkouadio and @PG430,

Do you confirm you implemented the fix in hal_rtc.c rather than in main.c? I just edited my previous comment to emphasize this important point.

Please keep me informed.

With regards,

[JudicaelK]

@ALABSTM I confirm that. The problem has been solved by: __HAL_RTC_WRITEPROTECTION_DISABLE(&hrtc); if (HAL_RTC_WaitForSynchro(&hrtc) != HAL_OK) { Error_Handler(); } __HAL_RTC_WRITEPROTECTION_ENABLE(&hrtc);

before the fonctions getDate and getTime.

Thanks.

[PG430]

hi @ALABSTM ,

I've made your suggested changes in stm32wlxx_hal_rtc.c. pls check attached snapshot of the source

[PG430]

@ASELSTM Please share any available update regarding this concern.

[ALABSTM]

Hi @PG430,

We are deploying the fix to the HAL RTC drivers of the different STM32Cube firmware packages. Hopefully, the fix will be integrated into the STM32CubeH7 firmware soon.

With regards,

[PG430]

@ALABSTM is this issue also fixed for STM32WLE5JC?

[kaunomedis]

2023.01.08, 1.8.4 still no fix, RTC still loosing seconds after power up.

[CedricRBR]

I noticed that other than HAL_RTC_Init() any interrupt that gets enabled has its counter reset to zero in MX_RTC_Init(). For me it was the call to HAL_RTCEx_WakeUpTimer_IT(). I carved the function and kept only what was necessary and called those functions in MX_RTC_Init() instead of the call to HAL_RTCEx_WakeUpTimer_IT()

[ALABSTM]

Hi,

Issue fixed in version 1.11.0 of the STM32cubeH7 firmware.

With regards,