[ell-i-developers] Re: Context Switching

  • From: Ivan Raul <supra.material@xxxxxxxxx>
  • To: ell-i-developers@xxxxxxxxxxxxx
  • Date: Tue, 4 Feb 2014 12:22:00 +0200

Hi to All.

Regarding the Cortex-M4, is there already any choice or possible choice for
MCU, I was checking the ST page, looks that there is being a shift to a
high performance processor at Ell-i:


With DSP and FPU the possibilities for signal processing are quite open.

It would be nice that I can start reading the datasheets in parallel to
come up with a solution that fits both processors from the beginning.

With Warm Regards, Ivan Raul

On Tue, Feb 4, 2014 at 9:38 AM, Pekka Nikander <pekka.nikander@xxxxxx>wrote:

> > After checking the current implementation of the threading system, I can
> particularly see two important elements:
> >
> > First, the ARM-M0 is designed to simplify context switching (two stack
> pointers, interrupt and threading mode, partial context already stored at
> interrupt state, etc.)
> Correct :-)  Please have a look at Cortex-M4 as well, as we want to
> support also it.
> > Second, the context switching implementation seems quite functional, the
> scheduler is the one that is still in progress. It is like the basic layer
> for threading is already developed (maybe needs to be tested)
> >
> > What was the last state of that part, the oct9-demo branch when you stop
> coding for it?
> Honestly, I simply don't remember properly.  IIRC, it was functional, at
> least most of the time, sans perhaps some corner cases.  That is, I'm not
> aware of any bugs in the lowest level threading code in the oct9-demo
> branch, but there may be corner cases where it works wrong.
> > What would be the initial state for new development. In particular, I
> was thinking on defining the interfaces ASAP, in order to complete the API
> with reference to those parts and allow better matching when the threading
> part would be ready.
> Sounds like a plan.  :-)
> > I am preparing a document for the wiki, as suggested by Teemu, to
> explain and fully understand the current context switching implementation
> (and possible design decisions taken for that)
> Excellent!
> --Pekka

Other related posts: