Re: [Xenomai-help] Page fault in real time task causes lockup

[Philippe Gerum <rpm@xenomai.org>]

On Wed, 2010-08-18 at 18:05 -0500, Steve Deiters wrote:
> > It turns out my problem was caused by an interrupt storm.  I 
> > had set up the interrupt to propagate to the Linux domain.  
> > When my rt task transferred to the Linux domain from the page 
> > fault it wasn't able to clear the device interrupt flag.  The 
> > interrupt was reenabled at the PIC level after Linux was done 
> > with it, and as soon as that happened it got interrupted again.
> > 
> > My fix was to disable the interrupt at the device level as 
> > soon as rt_intr_wait returns, and reenable it before calling 
> > rt_intr_wait.  I'm still not sure why I was getting that exception.
> 
> I'm still getting exceptions like I was getting before.  With the
> interrupt fix I had in there, the system stays responsive, just that
> task gets killed.  I'm still trying to track down the problem.
> 
> I'm using rt_intr_wait so I am synchronized with an external FPGA, but
> it is just periodic.  If I replace the rt_intr_wait with a timed wait
> with rt_wait_period it does not crash.  There seems to be some
> interaction with the rt_intr_wait I still do not understand.
> 
> I'm trying to make sense of the exception numbers it prints in messages
> like the following.  Maybe this will give me some better insight to what
> is happening.
> 
> [   24.480199] Xenomai: Switching dsp_task to secondary mode after
> exception #1792 in kernel-space at 0xc0062f48 (pid 595)
> 
> I tried turning on the I-pipe tracer to get some more information, but
> it crashes on startup.

Make sure to have CONFIG_IPIPE_TRACE_VMALLOC enabled. You may also want
to check whether disabling CONFIG_IPIPE_TRACE_IRQSOFF helps.

-- 
Philippe.