Developping a configurable RTOS

First, this sounds like a misguided venture. But if you want to do it for the thrill and learning experience, go ahead. It's fun. Years I wrote two simple RTOSs from scratch and it was a nice exercise. I even got paid for it.

You might start by learning what you can about existing RTOSs. There are several low-cost or free versions, such as uC-OS, and eCOS. I first learned a lot by studying the internals of TIP-MX, a nice commercial realtime OS for the TI-9900 uP decades ago. Determine the tradeoffs and what features that are most important to you. Here are some issues off the top of my head:

fast context switch minimum value for maximum interrupt latency small memory footprint compatibility with existing tools and languages memory protection between tasks dynamic (vs static) configuration network support multiple processor support good error detection and reporting good development tools for the user good documentation easy to expand

Thanks a lot Thad,

It is not for fun (a little bit). I have a 5 persons team and we have to make a project in embedded systems but we have not an idea how to develop a RTOS (do you have some references or examples of small RTOSs that could be understoud easily).

Thanks

Moctar

Thad Smith wrote:

The development of a basic RTOS kernel and services from the scratch is a one man one week project.

Start with the uC/OS book by Jean Labrosse. It is a good tutorial on the general principles of any RTOS, not just mucos.

Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

He already gave you two very good examples. I suggest you study them.

ok i will study them and will give the result later.

Thnks a lot!!

Moctar Grant Edwards wrote:

When you do, please don't top-post. Read the following links.

Regards, Richard.

• formatting link
• formatting link
  for Cortex-M3, ARM7, ARM9, HCS12, H8S, MSP430 Microblaze, Coldfire, AVR, x86, 8051, PIC24 & dsPIC

But maintaining it, and adding a lot of "necessary" features may cost considerably more time. ;-)

Don't reinvent the wheel, unless your requirements are very special.