libblock real-time behavior

Valette Eric eric.valette at
Thu Apr 10 16:22:11 UTC 2003

Till Straumann wrote:

> And then, I can't see how this approach can provide protection
> on a multiprocessor system but I believe to have asked this
> in the past and must admit that I still haven't looked into RTEMS MP...
> -- Till.

If you want RTEMS SMP (different cache same physical memory, shread 
interrupt controller), then believe me, there is a rather huge amount of 
redesign needed :
	1) Today, most critical section locking are performed via interrupt 
disabling/reenabling which of course does not work in a MP system,
	2) The interrupt disable/reenable use level so they can be arbitrily 
nested without any notion of locking heirarchy so remplacing them by 
maskedspilock will probably lead to immediate deadlock,
	3) Some structure need to be per cpu (e.g currentask, ...) and there is 
no real provision for that

I do not even speak of cache coherency problem on system where bus 
snooping must be specifically enabled...

BTW : ecos 2.0 starts to have some SMP code in it...

   /  `                   	Eric Valette
  /--   __  o _.          	6 rue Paul Le Flem
(___, / (_(_(__         	35740 Pace

Tel: +33 (0)2 99 85 26 76	Fax: +33 (0)2 99 85 26 76
E-mail: eric.valette at

More information about the users mailing list