Interrupt not getting triggered everytime in Zephyr project

RE: Interrupt not getting triggered everytime in Zephyr project

Brian (Geek) — Mon, 17 Jan 2022 03:30:16 GMT

Hello,

[quote userid="111520" url="~/f/nordic-q-a/83592/interrupt-not-getting-triggered-everytime-in-zephyr-project/347542#347542"]Have you tried comparing power consumption when calling k_sem_take(&my_sem,K_FOREVER) to when calling k_cpu_idle?[/quote]

I don't think i have tried to check power consumption using semaphore signalling method alone. I have checked current consumption using both k_cpu_idling and semaphore signalling together.

RE: Interrupt not getting triggered everytime in Zephyr project

Einarh — Thu, 13 Jan 2022 09:48:51 GMT

Yes, using a semaphore to wait for your ISR should give priority to the idle thread, which should have the same result as calling k_cpu_idle.

Have you tried comparing power consumption when calling k_sem_take(&my_sem,K_FOREVER) to when calling k_cpu_idle?

You can read more about how this works in Zephyr here:

http://developer.nordicsemi.com/nRF_Connect_SDK/doc/latest/zephyr/reference/power_management/index.html

https://docs.zephyrproject.org/latest/reference/kernel/scheduling/index.html

https://docs.zephyrproject.org/latest/reference/kernel/synchronization/semaphores.html

https://docs.zephyrproject.org/latest/reference/kernel/other/cpu_idle.html

Best regards,

Einar

RE: Interrupt not getting triggered everytime in Zephyr project

Brian (Geek) — Thu, 13 Jan 2022 08:17:50 GMT

so you suggesting me to just use semaphore for the signalling between ISR and while loop in main function. And as you suggested, if i don't idle forcefully will it go to idle state if semaphore is not available ?

Because if i don't call k_cpu_idle than my power consumption is much higher as compared to when calling k_cpu_idle from main loop.

RE: Interrupt not getting triggered everytime in Zephyr project

Einarh — Wed, 12 Jan 2022 16:52:06 GMT

Hello,

It seems redundant to use both the semaphore, flag and irq_lock at once in your code, and I suspect this complexity is also what could lead to your program reaching a race condition at some point.

Instead of manually idling, you can write k_sem_take(&my_sem,K_FOREVER), and your program will idle until it receives the semaphore from your interrupt.

Best regards,

Einar