On Thu, Sep 22, 2022 at 4:50 PM kuroda.hayato@fujitsu.com
<kuroda.hayato@fujitsu.com> wrote:
>
> Hi Amit,
>
> > > I checked other flags that are set by signal handlers, their datatype seemed to
> > be sig_atomic_t.
> > > Is there any reasons that you use normal bool? It should be changed if not.
> > >
> >
> > It follows the logic similar to ParallelMessagePending. Do you see any
> > problem with it?
>
> Hmm, one consideration is:
> what will happen if the signal handler HandleParallelApplyMessageInterrupt() is fired during
"ParallelApplyMessagePending= false;"?
> IIUC sig_atomic_t has been needed to avoid writing to same data at the same time.
>
But we do call HOLD_INTERRUPTS() before we do
"ParallelApplyMessagePending = false;", so that should not happen.
However, I think it would be better to use sig_atomic_t here for the
sake of consistency.
I think you can start a separate thread to check if we can change
ParallelMessagePending to make it consistent with other such
variables.
--
With Regards,
Amit Kapila.
