PowerPC FP handling weakness.

[Eric Valette]

Sergei Organov wrote:
> 
> Eric Valette <valette at crf.canon.fr> writes:
> > Joel Sherrill wrote:
> >
> > > > I'm affraid deffered FPU context switching has never been tested on
> > > > 60x/7xx. Maybe I can be wrong...
> > >
> > > I think that in a system where all tasks are FP deferred switching is
> > > equivalent to on every switch.  Think about it.  All deferred switching
> > > says is that we won't save the context until another FP task is switched
> > > in.  This is every time.  Not as clear as it should be but technically
> > > OK.
> >
> > No I disagree. FP context switching is performed when the newly tasks
> > really use the FPU. Saying a thread is a FP task does not mean it will
> > use the FP each time it is scheduled...
> >
> > All I wanted to say is that deferred FP context saving remove the need
> > of hardcoding whether a thread is FP or not and perform a better job by
> > saving FP context only when needed.
> >
> > Did I miss something?
> 
> Yes. RTEMS doesn't have "on-demand" FP context switch. Only "deferred" context
> switch. I.e. switch context only when new task is marked to be "floating
> point" and doesn't already own the FP context.
> 

Yes but on-demand FP context switch is rather easy to implement since we
now can handle the FP unavail exception. We just need to know to wich
thread the current FP context belong and store/restore it. I just have
no interest in coding it yet since :
	- I'm not working on FP applications,
	- I have more urgent RTEMS related things to do,

Anyway thanks for the discussion,

-- 
   __                 
  /  `                   	Eric Valette - Canon CRF
 /--   __  o _.          	Canon Development Europe Team Leader
(___, / (_(_(__         	Rue de la touche lambert
				35517 Cesson-Sevigne  Cedex
				FRANCE
Tel: +33 (0)2 99 87 68 91	Fax: +33 (0)2 99 84 11 30
E-mail: valette at crf.canon.fr	http://www.crf.canon.fr