Re: PgStat_HashKey padding issue when passed by reference

Hi,

On 2025-09-08 10:25:13 +0900, Michael Paquier wrote:
> On Sat, Sep 06, 2025 at 10:35:58AM +0900, Michael Paquier wrote:
> > One idea would be, for example, that keys used with simplehash should
> > never have any padding at all, and we could force a check in the shape
> > of a static assertion to force a failure when attempting to compile
> > code that attempts to do so.  That would give us a way to check in a
> > broader way if some code path do that currently, scaling better with
> > the expectations we could have in the whole tree or even out-of-core
> > extension code.
> 
> Doing some research here, I have noticed this one:
> https://en.cppreference.com/w/cpp/types/has_unique_object_representations.html
> 
> I was also wondering about some use of pg_attribute_packed() here, or
> perhaps enforce a check based on offsetof() and the structure size,
> but I doubt that any of that would be really portable across the
> buildfarm.
> 
> Another idea would be to make sure that the sizeof() of the structure
> matches with the sum of the sizeof() for the individual fields in it.
> That's cumbersome to rely on, still simpler.  Perhaps we could do
> something among these lines for pgstat_shmem.c, or just document that
> the structure should never have any padding.

I'd just add a comment mentioning that any padding bytes should be avoided.

Greetings,

Andres