Experimental Physics and
Industrial Control System

"Zimoch Dirk \(PSI\) via Core-talk" <core-talk at aps.anl.gov> · Wed, 2 Jun 2021 13:50:13 +0000

Checking taskLock for differences now....

> -----Ursprüngliche Nachricht-----
> Von: Michael Davidsaver <mdavidsaver at gmail.com>
> Gesendet: Mittwoch, 2. Juni 2021 15:45
> An: Zimoch Dirk (PSI) <dirk.zimoch at psi.ch>
> Cc: 'core-talk at aps.anl.gov' <core-talk at aps.anl.gov>
> Betreff: Re: Deadlock on VxWorks 6.7
> 
> On 6/2/21 2:23 AM, Zimoch Dirk (PSI) via Core-talk wrote:
> > Hi,
> >
> > When I use EPICS 7.0.5.1 or 7.0.4.1 on mv mv5100 running VxWorks 6.7, I
> see shortly after booting:
> >
> > 0x132f380 (timerQueue): epicsSpinLock(0x1fda0b79): Deadlock.
> > Recursive lock, missed unlock or block when locked.Thread timerQueue
> (0x132f380) can't proceed, suspending.
> >
> > Any idea?
> 
> Not immediately, though I'm more familiar with the RTEMS side.
> I don't see any recursive call in the stack trace you provided.
> Is it possible that the behavior of taskLock() has changed wrt. disabling
> interrupts?
> 
> > Why spinlocks on a single core CPU machine?
> 
> The RTEMS and vxWorks implementations don't actually spin.
> 
> https://github.com/epics-base/epics-
> base/blob/b777233efb06fa4e988c4f0738b0270dd3d095a3/modules/libcom/
> src/osi/os/vxWorks/osdSpin.c#L64-L86

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System