Experimental Physics and
Industrial Control System

Eric Norum <[email protected]> · Thu, 31 May 2012 10:20:50 -0700

On May 31, 2012, at 9:17 AM, Andrew Johnson wrote:

>> 
> 
> You wrote "device support interrupt callback or read method", which I 
> originally read as the ISR, but it would be just as bad to set the TIME field 
> in the port thread or a callback routine without first locking the lockset.
> 
> If you meant that TIME gets set by the dset->read_xxx() routine then that's 
> fine, because it gets called by the record's process routine with the lockset 
> locked.  I wasn't really thinking about the limitations of interrupt context, 
> it's that accessing a record's fields from any thread that has not first 
> locked the record's lockset is not safe.

Right -- the actual setting of the .TIME field is done by the various devEpics  dset->read_xxx() routines.
Sorry to have implied anything different.

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System