On 2017-06-04 19:14:06 -0400, Tom Lane wrote:
> Andres Freund <andres@anarazel.de> writes:
> > At the moment a number of flag variables set in signal handlers have
> > 'volatile bool' as type, others have 'volatile sig_atomic_t'. That's
> > kinda confusing. I think either is safe, but I think we should
> > standardize one of them.
>
> sig_atomic_t is more standards-conforming, I should think. I'm not sure
> if there are any current platforms where a store to a char variable
> wouldn't be atomic, but why live dangerously?
Well, we already have some variables that aren't actually booleans,
although I think all of them are only read not manipulated in signal
handlers (InterruptHoldoffCount etc). So one could argue that there's
no safety benefit in sig_atomic_t, because we're already using in other
places. We also already rely on int32 stores being atomic in other
parts of the code, although that's between processes not between signal
/ normal path of execution.
> I'd be inclined to let the code continue to treat the variables as
> if they were bool, ie store "true" and "false" not "1" and "0"
> into them. That should be perfectly safe.
Indeed.
Greetings,
Andres Freund