Hi,
Now we have a technical explanation, which explains the problem.
And not a judgment about knowledge.
Let's hope that always SPGIST_CONFIG_PROC hit first.
Best regards.
Ranier Vilela
________________________________________
De: Andres Freund <andres@anarazel.de>
Enviado: quarta-feira, 13 de novembro de 2019 18:03
Para: Ranier Vilela
Cc: pgsql-bugs@lists.postgresql.org
Assunto: Re: [BUG] Uninitializaed configOut.leafType used.
Hi,
On 2019-11-13 17:55:40 +0000, Ranier Vilela wrote:
> Ok, so all these commands are for what?
"commands"?
> case SPGIST_CONFIG_PROC:
> ok = check_amproc_signature(procform->amproc, VOIDOID, true,
> 2, 2, INTERNALOID,
INTERNALOID);
> configIn.attType = procform->amproclefttype;
> memset(&configOut, 0, sizeof(configOut));
>
> OidFunctionCall2(procform->amproc,
> PointerGetDatum(&configIn),
> PointerGetDatum(&configOut));
>
> configOutLefttype = procform->amproclefttype;
> configOutRighttype = procform->amprocrighttype;
>
> /*
> * When leaf and attribute types are the same, compress
> * function is not required and we set corresponding bit in
> * functionset for later group consistency check.
> */
> if (!OidIsValid(configOut.leafType) ||
> configOut.leafType == configIn.attType)
> {
>
> When case SPGIST_CONFIG_PROC, OidIsValid(configOut.leafType) is tested,
> but when case SPGIST_COMPRESS_PROC is fired, OidIsValid(configOut.leafType) is not necessary and configOut.leafType
> happy accessed?
Even if that were a problem - and I don't see why - that'd still not
make configOut.leafType be uninitialized. The SPGIST_CONFIG_PROC case is
always hit before SPGIST_COMPRESS_PROC, therefore configOut is always
initialized (c.f. memset(0) and the call to amproc to initialize the
contents).
And the OidIsValid() call is about whether a leafType is set or not, the
call to check_amproc_signature() doesn't need that.
- Andres