Does the R5F on the BeagleY-AI work as is? Is there a set of Firmware to run on?

I saw some people interested in the R5F on the BeagleY-AI a while back and I am not sure if anything transpired.

Has anyone been working on the R5F on the BeagleY-AI for motor advancement?

Has anyone been working on the R5F firmware and/or overlays to have them available for muxing to GPIO pins?

So, just a starter here and from what I can tell:

  1. Linux cannot load firmware to the 800MHz R5F remote cores (is this a fact?)
  2. It is u-boot in this case that loads the firmware to the R5F remote cores while Linux debugs it…
  3. and sysconfig is needed to get out of the remote cores to the real world.

I will keep trying, i.e. as 800MHz is a good remote core which is highly unlike the M4F that I used on a TI Launchpad years ago…

MCU+ SDK will not be updated for some time or I fell behind. Anyway, this is a start to anew.

Okay. I was on 9.x.x.x instead of 11.x.x.x and now I am starting to see that there are some ideas instead of barely anything relating to the R5F remote core that is located on the BeagleY-AI.

Technically, there is Zephyr support for R5s as well: BeagleY-AI — Zephyr Project Documentation

I saw that too. Nice. I got lost in the workings quickly when there was a Zephyr Port on the images for the BeagleY-AI.

I guess I will just stay patient. I see the TPM is what loads the firmware in SPL mode (I think that is correct). Anyway @ayush1325 , thank you for your reply. So far, I am more of a cosmetic feature person that is trying to port some M4F code logic to the R5F remote cores on the BeagleY-AI.

I may try the Arago image again. I am not sure, though. I am still trying to catch up.

Odd. I can load on the BBB the PB and PB2 m4 and pru, it has to be done in kernel. There are drivers for these in the kernel that loads the firmware. I would expect your box to be similar. That said though, an having computer problems here right now so can’t log into my beagles and see what’s available.

Okay about your computer issues. And yes, I see on the BeagleY-AI that some mcu_x and another are located in the fs. The wake-up_x file I thought needed to be done in SBL.

Maybe I am lost. I will update you once I try on a Debian box I have here. It could already be ported and ready… I see why you say, “Odd.” Maybe I am just heavily mistaken.

Iirc the m4 on the beagles is a watchdog device. So it needs to be regularly talked to or the machine reboots. Iirc in uboot a register is written, so at least the computer boots without being reset. Once the kernel is up it updates the software on the m4. What it rewrites it with, I don’t know the purpose, it’s just a binary blob that it loaded. I have never looked into what the code does. My guess it @RobertCNelson or Jason Kirinder would know. Anyway, if you can find the drivers for this in the kernel, it will give clues to what you need to do …

@DavidSummers anything outside of what u-boot sets up, all the other cores are a factory “blank” slate.. For the average user here, using Zephyr with your own custom build is what I/we recommend. :wink:

I thought with u-boot (SBL) and the r5f remote cores, the MCUPSDK would work but who knows now?

and yes and okay to @ayush1325 and @RobertCNelson with utilizing Zephyr. From my perspective, it is a shame to not be able to home in on the R5F remote cores via u-boot.

From what I have been reading with the Academy and MCUPSDK (neither are ready to start with their remote core R5F core(s)), there are a couple additions to make in u-boot and the kernel to get up and running with debugging. Then, one could use make and/or CCS to debug on the UART line. printf and things come in handy…

There are plenty of cores to go around, just didn’t touch the r5 used by u-boot as it’s running atf/optee

When I get back on my beagle farm, I’ll look into uboot and see what was set. If I manage to get my computer tomorrow, (which means digging out my my AMD E three hundred and fifty dating from 2007 …) I’ll check what is done in uboot.

@DavidSummers , I can guest give you (if that is what it is called) a computer if you have your own chips. I sold all the chips out of the computers a while back. I got ripped off but in kindness only.

I have some Win 11 era with, I think, TPM and some without the TPM versioning for newer models of Windows Server. You pay S & H and I will gladly send one to you. Seriously. No if, and, or buts about it. I acquired some hardware that people are not wanting to use these days. So anyway, if you say yay, fine. Nay is okay too. No big rush or any pressure. I just know sometimes people need support.

update

I have intel only. I forget the socket size but I can check. I put Debian GNU/Linux on the i7 Intel at super speed. It is blazing and with a Mesa Card, forget about it...

It is an old workstation. Graphics are pretty much out… But, the clock speed of the motherboard and i7 processors really feel comfortable when plugging things in and output is quick.

@RobertCNelson and anyone else looking to get data:

I am sure everyone wants to get their hands dirty with the R5F remote cores on the BeagleY-AI.

I am going to start to follow TI in this matter in case I can configure anything.

From what I can tell, breaking the communication between am67a (remoteproc and etc) and R5FSS and then altering the board state negatively can be done by both Linux and the R5FSS using the same memory at the same time.