Re: [rtl] RT and X problem

[Volker Zahn <volker@NeRo.Uni-Bonn.DE>]

Hi RT folks,

my problem was reported twice (see below), but nobody seems to have
any clues. So here another report with some details.

When I'm running my RT task, using hardware interrupts and shared
memory, the X server sometimes hangs for some 10-30 seconds.
The RT task is running correct and reliable. Just the X inputs
by mouse and keybord are not executed, only stored. So all events
happend while hanging are executed after X server is active again. 
This happens often (but not reproducable), when lots of X events are
proccessed, e.g. a slider is moved or lots of X outputs are made.

It seems, that something goes wrong with the grabbing of the events. 
Perhaps it depends on the window manager (fvwm95) ? I'm now testing
another window manager, but haven't enough tested to verify.

Any clues ???

Volker Zahn




> From owner-rtl@rtlinux.cs.nmt.edu Mon May 18 04:54 MET 1998
> Date: Sun, 17 May 1998 21:08:10 -0400
> From: cecil ogg <cecilogg@erinet.com>
> MIME-Version: 1.0
> To: rtl@rtlinux.cs.nmt.edu
> CC: JosÈ Daniel MuÒoz FrÌas <daniel@srvnt1.upco.es>
> Subject: Re: [rtl] RT and X problem
> Content-Transfer-Encoding: 8bit
> 
> JosÈ Daniel MuÒoz FrÌas wrote:
> > 
> > Hello!
> > 
> > I have a RT-task connected to a hardware interrupt and when this task is
> > running and I open a window in X, the X window server seems to stop (no
> > mouse movements...) and after some seconds (5, 20... it depends) X
> > continue to work normally (and gets a lot of events suddenly (mouse
> > movements and clikcs that I have done while X was sleeping) ). Somebody
> > have similar problems?
> > 
> > Im using a Pentium 200 Mhz, Debian 1.3.1 Kernel 2.0.33 RT-linux 0.6
> > 
> > Thanks in advance
> > 
> > Dani.
> > 
> > PS. Sorry, my English is not so good.
> > --
> > ______________________________________________________________
> > JosÈ Daniel MuÒoz FrÌas                Daniel@dea.icai.upco.es
> > 
> > Universidad Pontificia Comillas
> > ICAI. Depto. ElectrÛnica y Autom·tica
> > 
> > Alberto Aguilera 23
> > 28015 Madrid, Spain
> > 
> > Telf. 34-1-5422800                           Fax: 34-1-5596569
> > ______________________________________________________________
> > --- [rtl] ---
> > For more information on Real-Time Linux see:
> > http://rtlinux.cs.nmt.edu/
> 
> I've noticed similar events on my system. In my case, I'm not using any
> interrupts. I suspect that the periodic task runs with a period of 0
> while the module continues to load and finally define the correct
> period. I'm guessing; I think I have some timestamps to support my
> conjecture. Eventually, the module runs the RT-task correctly but for a
> while, the keyboard and mouse are locked. Any clues? I can't reliably
> reproduce this yet.
> 
> Cecil Ogg
> 
> --- [rtl] ---
> For more information on Real-Time Linux see:
> http://rtlinux.cs.nmt.edu/
> 
--- [rtl] ---
For more information on Real-Time Linux see:
http://rtlinux.cs.nmt.edu/