not really a Re: problem about RTEMS support mutiprocessor!

[Joel Sherrill]

red-eye at 263.net wrote:
> 
> hello,
> how RTEMS support SMP? (tightly-coupled use shared memory)

Yes but you would end up with essentially two instances of
RTEMS running.  Each would have to have its own private
executive area.

RTEMS treats each CPU as an independent entity and essentially
uses an RPC technique to make calls go to another CPU when
necessary.  

> i study the mp samples test!
> i guess if they work like this:
> there is same code in all processors's ROM .every processor select instructions to execute by "if" and their node number.
> if not , how the RTEMS work with mutiprocessor?

That's how they work but ...

This is just a test construct so the same source can be used
for both nodes in an mp test.
 
> does RTEMS support system restructure and fault(include hardware fault and software fault) inspect and fault restore in execute time?

No.

> that is ,does RTEMS support task redistribution to processors in execute time?

No.  

> when there is fault ,such as one processor is broken(can not work),RTEMS can restructure
> system and make the system run ok.

And no. :)

It would be nice to look at some of the hot-swapping support to
see if that could come into play but RTEMS is designed for
statically assignment of tasks for applications that require
more than one CPU.   This is a fairly common arrangement in
MP deeply embedded systems.  One CPU may do IO and crunch
numbers (run a filter) and pass results on to another CPU.

> if RTEMS support those ,how RTEMS support?        thanks alot
> 
>       IP¿¨¡¢ÉÏÍø¿¨Á½ÕÛ    Ê·Å«±È¡¢À¯±ÊСÐÂÈ«¼Ò¸£                  ¼ÓÈëÊéÓÑ»á»ñ¾«ÃÀÀñÆ·

-- 
Joel Sherrill, Ph.D.             Director of Research & Development
joel at OARcorp.com                 On-Line Applications Research
Ask me about RTEMS: a free RTOS  Huntsville AL 35805
   Support Available             (256) 722-9985