When to use NRF_LOG_FLUSH()

[masterLee]

Dear all,

I have noticed that NRF_LOG_FLUSH() function is used in many examples after each nrf_log print function. In some cases though, the program freezes when I try my own implementations.

So my question is when should I use NRF_LOG_FLUSH()  and where.

Thank you for your time

[Jørgen Holmefjord]

Hi,

NRF_LOG_FLUSH() is used to force processing of the entire log queue. It can be triggered in case you want to make sure that a log message is processed/output before a specific action, but in general it is best to keep the log processing to the main thread. Calling this function in an interrupt context will block other interrupts from being handled, which could cause timing issues in your application.

Best regards,
Jørgen

[masterLee]

Oh Ok! Now it makes sense why the things I wanted to print inside interrupts were not printed... I found out though that when I placed the   __SEV() together with  __WFE() in the main loop without changing the rest of my code the things I wanted, were printed with no problem. Why do you think this happens?

[Jørgen Holmefjord]

If you have configured the logger in deferred mode, you need to call NRF_LOG_PROCESS or NRF_LOG_FLUSH to print the log queue. Do you have one of these in your main loop? 

[masterLee]

According to the CMSIS configurator in the nrf-log folder NRF_LOG_DEFERRED is enabled indeed.But what does it mean? 

Yes, NRF_LOG_FLUSH() exists in my code in several places and has caused me much trouble so far i would say

[Jørgen Holmefjord]

You can read about the difference between in-place and deferred mode in the documentation. In general, I would recommend only processing the logs in the main loop, before going to sleep. This will avoid the processing of logs to block other event handling.

[masterLee]

Ok great! Thank you for the good tips!