RTOS selection for STM32H7R/S

[tdecker2]

We are starting to develop a new hardware platform and are currently evaluating the STM32H7R/S series for the main controller.

The question now is which RTOS is best suited here in terms of long-term support from ST. We're going to use a lot of communication interfaces (USB, Ethernet, external WiFi/BT module, CAN), an RTOS with an integrated communication stack would make our work easier.

Zephyr: High initial training effort, but you get a great framework. The two available H7RS-Eval boards are already supported, but in the video Zephyr 4.1: Status of STM32 support – Erwan Gouriou // Zephyr Meetup Grenoble Mar. 26, 2025 by @Erwan Gouriou from March of this year you can see that Zephyr support for the H7RS is not yet complete. Can anyone give an estimate of the extent to which you can already work with it or how much effort is involved in porting drivers from the H7 to the H7RS that are not yet available? I'm new to Zephyr and can't estimate the effort involved.

ThreadX: Support for Ethernet, USB and flash file systems is included, the AIROC radio modules from Infineon are also supported. But ST seems to be turning away from ThreadX and is focusing on FreeRTOS again. See the post from @Maxime_MARCHETTO here: New strategic directions for STM32Cube 
One advantage is the CubeMX integration, so that you get a code template here.

FreeRTOS: I won't touch ST's lwIP port for FreeRTOS again. So maybe I have to look at the FreeRTOS-Plus-TCP stack? According to the "New stratetig direction" ST is working on a port of the UsbX and FileX modules for FreeRTOS. But not for NetX(duo). The CubeMX integration is also an advantage here.

Any opinion on this is welcome and will help us make a decision.

[mihe]

Same question here: We are also evaluating a STM32H7S as the main controller for a new hardware platform and we're undecided what's the best suited RTOS for this controller. 

I'm also completely new to Zephyr and can't estimate the learning curve.

@tdecker2 : Which way did you go? Any positive/negative points with your decision?

[tdecker2]

We decided to go the Zephyr-way. Steep learning curve, but you get a lot of features. The support for H7RS-family is not complete yet, but it gets better every month. And you can learn a lot about Zephyr internas when contributing missing features.

[mihe]

That's also my current impression: The learning curve seems steep, but the features you get look massive! Thanks for your reply.

[Vidar]

Hi,

i have been working with the H7S3L8 for some time now, mainly with RTOS.

For context, my requirements have always been speed and reliability. I work with multiple DMA implementations as well as Ethernet.

 

I tried out FreeRTOs as well as ThreadX, and I tried out ThreadX whilst it was the main RTOS supported by ST. The Issue I got with ThreadX is the big overhead in terms modules you need to install and therefore which could break. But nothing beats FileX and USBX, which I dont need.

I switched to FreeRTOS with LWIP finally and currently use it with Percepio, and it works like a beast. DMA implementation works great and I achieve solid performance (0.3 ms between start of data gathering and successful sending via Ethernet). If you look at my post history, i had a interesting past with Ethernet on the H7RS, i glued together this POS of a driver 3 diffrent times, I did it each time again as the driver update came, back at 1.1 the support was aweful, but i would sayy I got a very good implementation fo LWIP with FreeRTOS with DHCP, working ARP and so on. 

 

Why would you never touch LWIP again btw.?

[mihe]

Hi Vidar,

sounds good, i'll have a look at your previous posts regarding the H7S3L8 with FreeRTOS implementation. :thumbs_up:

[Vidar]

If you have dedicated questions regarding LWIP, task management and high data flow, and some common issues regarding DMA, feel free to mention me in your post, then I will see it faster