On Thu, Sep 29, 2022 at 1:43 AM Jacob Champion <jchampion@timescale.com> wrote:
>
> On Tue, Sep 27, 2022 at 6:14 PM Masahiko Sawada <sawada.mshk@gmail.com> wrote:
> > No. Since cluster_name is PGC_POSTMATER, we leak a little postmaster
> > memory only once when starting up. application_name is PGC_USERSET but
> > since we normally allocate memory in PortalMemoryContext we eventually
> > can free it.
>
> Oh, I see; thank you for the correction. And even if someone put an
> application_name into their postgresql.conf, and then changed it a
> bunch of times, we'd free the leaked memory from the config_cxt that's
> created in ProcessConfigFile().
Right.
>
> Is there a reason we don't provide a similar temporary context during
> InitializeGUCOptions()? Naively it seems like that would suppress any
> future one-time leaks, and maybe cut down on some Valgrind noise. Then
> again, maybe there's just not that much demand for pallocs during GUC
> hooks.
While this seems a future-proof idea, I wonder if it might be overkill
since we don't need to worry about accumulation of leaked memory in
this case. Given that only check_cluter_name is the case where we
found a small memory leak, I think it's adequate to fix it.
Fixing this issue suppresses the valgrind's complaint but since the
boot value of cluster_name is "" the memory leak we can avoid is only
1 byte.
Regards,
--
Masahiko Sawada
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com
