> ... I still don't think the UP
> implementations of epicsSpinTryLock() should ever be able to call
> cantProceed() even in this circumstance ...
Ok, so I was thinking about this from the context of epicsSpinLock(). The try lock does have an error path which can be used. I would argue that we shouldn't call this defined behavior as I don't think we want the overhead of detecting recursive calls in other implementations.
> ... I still don't think the UP
> implementations of epicsSpinTryLock() should ever be able to call
> cantProceed() even in this circumstance ...
Ok, so I was thinking about this from the context of epicsSpinLock(). The try lock does have an error path which can be used. I would argue that we shouldn't call this defined behavior as I don't think we want the overhead of detecting recursive calls in other implementations.