untested functions in freertos

[jumjum123]

Now that driver are more and more complete, let me ask for that couple of untested functions in freertos.
They are mainly croutine, notification, task resume and some special queue.
Are these already on a todo list for next future ?
In my case notification works(on same cpu), but it is still untested.

[www.freertos.org]

Can you clarify what you mean when you say this functionality is not tested? FreeRTOS developers test the code they distribute, including these functions.

[ESP_Sprite]

We modified FreeRTOS for multi-core use, and in the process we modified a fair few functions of it. Most of the FreeRTOS functions are tested for multicore-compatibility by now, but some are not and they are marked by an assert; that is what the OP means.

We've been planning to test these functions and validate/fix them. With a bit of luck, we'll get around to doing this before the 2.0 release.

[guthsmash]

I'm using some untested functions because I like to live dangerously. I'd like to share the results of my testing efforts, is there a list if what should be tested to mark them as tested?

[ESP_Sprite]

Not yet. In general we need to see if these functions still work as advertised. More specifically for multicore, we need to test the behaviour of these functions when called from both CPUs at the same time; there may be some muxing race conditions involved. I would definitely be interested in hearing which functions you tested, how, and what the result was. Feel free to post in this topic or mail to jeroen at espressif dot com.

[jumjum123]

I'm using notification, task resume and task suspend in my app.
Good news, it works fine.
Bad news, both tasks are running on same core, so this is not the test you need.

[guthsmash]

I didn't really test them exclusively, I simply used some of the process listing functions. I guess making a dozen threads on two cpus and continuously calling some of the utility functions could be used as a kind of explorative testing, but I'm not familiar with systematically looking for problems with deadlocks and race conditions. I'll let you know if I run into an issue