How to properly use cortexm_sleep

Hey

Is there anyone using cortexm_sleep for a real application? And if yes, would you want to share how exactly that's done?

Let me say that I've never written a real application with RIOT, but I would like to. One of the important requirements is that the app must be in sleep (deep sleep) and only wake up to do some work. And another requirement is to use the watchdog.

That said, I'm familiar with the following Arduino pseudo code.

 while \(forever\) \{
     do stuff

     disable interrupts
     if \!certain condition
         \_\_WFI\(\);
     enable interrupts
 \}

That "certain condition" is mostly checking that no WDT or RTC interrupt occurred before disabling the interrupts. These interrupts are handled by ISR's and they set a flag. This is a fairly common thing to do, I think.

I don't see (or understand) how this can be achieved with RIOT-OS and cortexm_sleep. It is essential to check the condition AFTER disabling the interrupts.

Hi

We do. On STM32L0, STM32L1, nRF52, etc.

Checking for interrupt flags is absolutely useless here - 1) WFI checks for any relevant interrupt flags, 2) interrupt still can happen between check and WFI.

Hi Kees,

Some observations:

1- WFI will not send you to the deepest sleep states- clocks are gated, but many things remain powered.

2- If you are using the timer module and depending on you clock configuration you may experience more or less frequent wake ups. Whether this is acceptable is up to you.

3- Normally, if no thread is runnable (i.e. all are waiting/blocked) the idle thread is run and that has a loop which sends the MCU to the deepest state possible (considering the currently-enabled peripherals). This means that in many cases the explicit WFI is not needed if in your "while(forever)" loop you put some call that blocks waiting for the events you are interested in.

The interaction between (1) and (3) means that by doing a WFI in a thread it is possible that you don't sleep as deep as you could:

- The current thread will be blocked on the WFI instruction until an    event arrives. - In the meantime, control will not go back to the kernel. - This means the idle thread will not run (from the schedulers point of    view, you thread is still running). - The pm_whatever() call in the idle thread will not run.

In conclusion, try to use the default mechanism if you can.

Regards,

Juan.

Hey Oleg,

Are you using the watchdog?

Thanks Juan,

That's a lot of information to digest. It will take me a bit of time to go through. -- Kees

Hi.

Yes, but for emergency reboots only.

Implementations: https://github.com/unwireddevices/RIOT/blob/loralan-public/cpu/stm32_common/periph/iwdg.c and https://github.com/unwireddevices/RIOT/blob/loralan-public/cpu/nrf5x_common/periph/iwdg.c

For task scheduling on sleepy devices we use RTC-based timers with 1 ms resolution and 1 week max period (https://github.com/unwireddevices/RIOT/tree/loralan-public/sys/rtctimers-millis; on STM32L, we use regular HW RTC with SSR register, on nRF52 — “software” RTC implemented on top of HW RTT unit).

Nice,

Thanks for sharing

Hey everyone,

let me say a few words as Kaspar and I have been driving the implementation of this. In general there is no need to manually go to sleep, as the deepest sleep mode will always be selected from the idle thread depending on the deepest mode possible. Which mode is possible depends on the used periphs. For STM32 sleep was long not possible, as PM_BLOCKER_INITIAL was initialized with 1 for each mode, which resulted in never going to a sleep mode.

From what I can see is that it is still initialized with ones?

Nevertheless, to make us of it PM_BLOCKER_INITIAL should be initialized with 0. All periph modules should then individually block the sleep modes they can not operate in. For now the biggest problem was the shell/uart, as it would always block any sleep modes. An exception are some platforms, as they have a Uart which can wake them up from sleep modes.

I have modified the atmega1284p implementation to make use of it [0] and [1].

Best Regards Robert

[0]

[1]

Thanks