Re: System username in pg_stat_activity

On Fri, Jan 19, 2024 at 1:43 PM Julien Rouhaud <rjuju123@gmail.com> wrote:
>
> Hi,
>
> On Thu, Jan 18, 2024 at 11:01 PM Magnus Hagander <magnus@hagander.net> wrote:
> >
> > I did. Here it is, and also including that suggested docs fix as well
> > as a rebase on current master.
>
> +    if (MyClientConnectionInfo.authn_id)
> +        strlcpy(lbeentry.st_auth_identity,
> MyClientConnectionInfo.authn_id, NAMEDATALEN);
> +    else
> +        MemSet(&lbeentry.st_auth_identity, 0,
> sizeof(lbeentry.st_auth_identity));
>
> Should we use pg_mbcliplen() here?  I don't think there's any strong
> guarantee that no multibyte character can be used.  I also agree with
> the nearby comment about MemSet being overkill.

Hm. Good question. I don't think there is such a guarantee, no. So
something like attached v5?

Also, wouldn't that problem already exist a few lines down for the SSL parts?

> +       value as the identity part in <xref linkend="system-user" />, or NULL
> I was looking at
> https://www.postgresql.org/docs/current/auth-username-maps.html and
> noticed that this page is switching between system-user and
> system-username.  Should we clean that up while at it?

Seems like something we should clean up yes, but not as part of this patch.

--
 Magnus Hagander
 Me: https://www.hagander.net/
 Work: https://www.redpill-linpro.com/