Hi Tom=2C
> All the callers do=2C so I don't entirely see your point.
Well=2C I still confuse...
> The stack traces you show look to me like the fault is probably in
> the caller=2C ie=2C calling PQfinish twice on the same "conn".
patch from dbd_pgsql_close function:
275 if (dbd->conn !=3D NULL) {
276 #if defined(DEBUG) && !defined(_WIN32)
277 PQuntrace(dbd->conn)=3B
278 #endif
279 PQfinish(dbd->conn)=3B
280 dbd->conn =3D NULL=3B
281 }
282 FREE(dbd)=3B
283 dbd =3D NULL=3B
IHMO=2C the caller of PQfinish can=B4t call twice.
Best regards=2C
Ranier
> From: tgl@sss.pgh.pa.us
> To: ranier_gyn@hotmail.com
> CC: pgsql-bugs@postgresql.org
> Subject: Re: [BUGS] Use after free? in fe-connect.c:closePGconn
> Date: Wed=2C 15 Jun 2016 19:05:53 -0400
>=20
> Ranier VF <ranier_gyn@hotmail.com> writes:
> > In [d:\pginstaller.auto\postgres.windows\src\interfaces\libpq\fe-conne=
ct.c=2C
> > closePGconn(PGconn *conn):
> > Does not check if conn is NULL.
>=20
> All the callers do=2C so I don't entirely see your point.
>=20
> The stack traces you show look to me like the fault is probably in
> the caller=2C ie=2C calling PQfinish twice on the same "conn".
>=20
> regards=2C tom lane
=20
=09
=20
Livre de v=EDrus. www.avast.com. =09
=09
=
