hi, I wanted to know in beaglebone ai, how the exchange of variables processed by the cortex m4 takes place, how can I read them with Linux in the cortex a?
sorry the question is how do I read variables in the m4 from to a15 and vice versa? Thank you
Hey
I had some success with this issue following these instructions.
PRU Cookbook (markayoder.github.io)
The instructions for the beaglebone black, you just need to adjust for the Beaglebone AI. there is a section for this one. Depending on what you want to do, and if youre willing to use “C”, or python, you probably can get away with it. Either the shared memory, or the remoteproc framework.
I believe the SDK also has some examples using Remoteproc. You need to download the sdk, and the examples for the am572x folder, there are some examples using this framwork.
Hope this helps.
There are probably several ways to do this (direct memory maps, etc), however as @mringelectrick said probably going thru remoteproc framework is easiest.
Once setup you will get /dev/<somename> character devices on the linux side … so sending something from linux will be as easy as echo "My Message" > /dev/<somename> and likewise reading back would be cat /dev/<somename>
The tricky part is getting rpmsg built into your m4 code…thats a bit beyond me at the moment but I’m sure searchable examples exist somewhere on TI’s site.
Another consideration is rpmsg may impose some latency above and beyond going the shared memory approach, however that may not matter given the application.
Hey Barry.
Yes. I believe the fastest and best way is to write to shared memory. By any chance, do you have an example using C++ on the ARM side?