From: Tom Lane <tgl@sss.pgh.pa.us>
To: Andrew Gierth <andrew@tao11.riddles.org.uk>
cc: Jeremy Evans <code@jeremyevans.net>,
 Michael Paquier <michael@paquier.xyz>,
        pgsql-bugs@lists.postgresql.org
Subject: Re: BUG #15367: Crash in pg_fe_scram_free when using foreign tables
In-reply-to: <12293.1536591166@sss.pgh.pa.us>
References: <153626613985.23143.4743626885618266803@wrigleys.postgresql.org>
 <20180906215836.GP2726@paquier.xyz>
 <20180906222843.GF17425@jeremyevans.local>
 <20180907165103.GH17425@jeremyevans.local>
 <20180907175518.GI17425@jeremyevans.local>
 <20180907181522.GB32058@paquier.xyz> <20016.1536351535@sss.pgh.pa.us>
 <20180907204403.GC32058@paquier.xyz> <16558.1536407783@sss.pgh.pa.us>
 <18398.1536410055@sss.pgh.pa.us> <87o9d8rv08.fsf@news-spur.riddles.org.uk>
 <22025.1536521375@sss.pgh.pa.us> <87zhwqqwz7.fsf@news-spur.riddles.org.uk>
 <24041.1536523851@sss.pgh.pa.us> <12134.1536550702@sss.pgh.pa.us>
 <12293.1536591166@sss.pgh.pa.us>
Comments: In-reply-to Tom Lane <tgl@sss.pgh.pa.us>
	message dated "Mon, 10 Sep 2018 10:52:46 -0400"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <16983.1536596764.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Mon, 10 Sep 2018 12:26:04 -0400
Message-ID: <16984.1536596764@sss.pgh.pa.us>
Precedence: bulk

I wrote:
> AFAICT, this platform's "ld" doesn't have any simple equivalent of
> a version script, but it does have "-B symbolic", which appears to
> fix the problem in a quick test.

BTW, I discovered while rummaging through the git history that we
used to use -Bsymbolic on most platforms --- but only for ODBC,
which had exactly the problem we're fighting now of an intended-to-
be-local reference being resolved outside the library, but in a place
where the malfunction was very obvious.  When we removed ODBC from the
core distro, the Makefile support for -Bsymbolic eventually went away
too.  But back around 7.2 we had this:

$ grep -r shlib_symbolic /home/postgres/REL7_2/pgsql
/home/postgres/REL7_2/pgsql/src/interfaces/odbc/GNUmakefile: 54: LINK.shar=
ed +=3D $(shlib_symbolic)
/home/postgres/REL7_2/pgsql/src/makefiles/Makefile.bsdi: 19: shlib_symboli=
c =3D -Wl,-Bsymbolic
/home/postgres/REL7_2/pgsql/src/makefiles/Makefile.freebsd: 6: shlib_symbo=
lic =3D -Wl,-Bsymbolic -lc
/home/postgres/REL7_2/pgsql/src/makefiles/Makefile.hpux: 33: shlib_symboli=
c =3D -Bsymbolic
/home/postgres/REL7_2/pgsql/src/makefiles/Makefile.irix5: 4: shlib_symboli=
c =3D -Wl,-B,symbolic
/home/postgres/REL7_2/pgsql/src/makefiles/Makefile.linux: 4: shlib_symboli=
c =3D -Wl,-Bsymbolic
/home/postgres/REL7_2/pgsql/src/makefiles/Makefile.netbsd: 6: shlib_symbol=
ic =3D -Wl,-Bsymbolic -lc
/home/postgres/REL7_2/pgsql/src/makefiles/Makefile.openbsd: 6: shlib_symbo=
lic =3D -Wl,-Bsymbolic
/home/postgres/REL7_2/pgsql/src/makefiles/Makefile.sco: 3: shlib_symbolic =
=3D -Wl,-Bsymbolic
/home/postgres/REL7_2/pgsql/src/makefiles/Makefile.solaris: 11: shlib_symb=
olic =3D -Wl,-Bsymbolic
/home/postgres/REL7_2/pgsql/src/makefiles/Makefile.unixware: 6: shlib_symb=
olic =3D -Wl,-Bsymbolic

This suggests that we're gonna need -Bsymbolic on Solaris too,
which agrees with your remark upthread.  Unfortunately, with
all the Solaris buildfarm members being hors de C99 right now,
there's no easy way to test.  I'm inclined to go fix the problem
that's blocking castoroides, at least.

			regards, tom lane