Experimental Physics and
Industrial Control System

"Ernest L. Williams Jr." <[email protected]> · Wed, 09 Nov 2005 11:17:43 -0500

On Wed, 2005-11-09 at 16:02 +0100, Ralph Lange wrote:
> Same results:
> 5 messages on Linux, 10 messages on HP-UX for the middle test.
What is the linux kernel version that you have for Debian 3.1?


Ernest


> 
> Ralph
> 
> 
> Marty Kraimer wrote: 
> > Ralph,
> > 
> > 
> > Can you turn on real time priorities and try again?
> > 
> > 
> > Look at configure/CONFIG_SITE
> > 
> > 
> > 
> > 
> > Marty
> > 
> > 
> > 
> > On Nov 8, 2005, at 6:55 AM, Eric Norum wrote:
> > 
> > > On Nov 8, 2005, at 6:22 AM, Ralph Lange wrote:
> > > 
> > > > On my linux box (Debian 3.1), epicsMessageQueueTestHost yields
> > > > unexpected results:
> > > > 
> > > > aragon: .../base/3-14-X > bin/linux-
> > > > x86/epicsMessageQueueTestHost 
> > > > Simple single-thread tests.
> > > > [...]
> > > > Single receiver, single sender tests.
> > > > Systems with priority-based scheduler should send only
> > > > 4 or 5 messages (sender priority > receiver priority).
> > > > Sent 5 messages.
> > > > Systems with priority-based scheduler should send 10
> > > > messages (sender priority < receiver priority).
> > > > Sent 5 messages.
> > > > All systems should send 10 messages (sender pauses
> > > > after sending each message).
> > > > Sent 10 messages.
> > > > 
> > > > Shouldn't the middle case (sender priority < receiver priority)
> > > > show 10 messages being sent instead of 5?
> > > 
> > > 
> > > You're seeing the fact that your linux box doesn't have a strict
> > > priority-based scheduler -- i.e. it's not a "real-time" operating
> > > system.
> > >

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System