Hi,
On 2023-09-27 15:42:15 +0100, Peter Eisentraut wrote:
> On 27.09.23 09:08, Michael Paquier wrote:
> > On Wed, Sep 27, 2023 at 08:41:55AM +0200, Jakub Wartak wrote:
> > > Attached patch adjusts pgstat_track_activity_query_size to be of
> > > size_t from int and fixes the issue.
> >
> > This cannot be backpatched, and using size_t is not really needed as
> > track_activity_query_size is capped at 1MB. Why don't you just tweak
> > the calculation done in pgstat_read_current_status() instead, say with
> > a cast?
>
> I think it's preferable to use the right type to begin with, rather than
> fixing it up afterwards with casts.
I don't think going for size_t is a viable path for fixing this. I'm pretty
sure the initial patch would trigger a type mismatch from guc_tables.c - we
don't have infrastructure for size_t GUCs.
Nor do I think either of the patches here is a complete fix - there will still
be overflows on 32bit systems.
Perhaps we ought to error out (in BackendStatusShmemSize() or such) if
pgstat_track_activity_query_size * MaxBackends >= 4GB?
Frankly, it seems like a quite bad idea to have such a high limit for
pgstat_track_activity_query_size. The overhead such a high value has will
surprise people...
Greetings,
Andres Freund