From: Ronan Dunklau <ronan.dunklau@aiven.io>
To: Thomas Munro <thomas.munro@gmail.com>
Cc: Devrim =?ISO-8859-1?Q?G=FCnd=FCz?= <devrim@gunduz.org>,
 PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>,
 Fabien COELHO <coelho@cri.ensmp.fr>, Andres Freund <andres@anarazel.de>,
 Dmitry Dolgov <9erthalion6@gmail.com>
Subject: Re: LLVM 16 (opaque pointers)
Date: Mon, 16 Oct 2023 10:31:32 +0200
Message-ID: <2700884.mvXUDI8C0e@aivenlaptop>
In-Reply-To:
 <CA+hUKGLxVtLCo3xD1OEnMSQbq04MKx1G_vQ5SZQNMBSFPR_TpA@mail.gmail.com>
References:
 <CA+hUKGKNX_=f+1C4r06WETKTq0G4Z_7q4L4Fxn5WWpMycDj9Fw@mail.gmail.com>
 <5716086.DvuYhMxLoT@aivenlaptop>
 <CA+hUKGLxVtLCo3xD1OEnMSQbq04MKx1G_vQ5SZQNMBSFPR_TpA@mail.gmail.com>
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain; charset="UTF-8"
Archived-At: <https://www.postgresql.org/message-id/2700884.mvXUDI8C0e%40aivenlaptop>
Precedence: bulk

Le vendredi 13 octobre 2023, 22:32:13 CEST Thomas Munro a =C3=A9crit :
> On Wed, Oct 11, 2023 at 10:31=E2=80=AFPM Ronan Dunklau <ronan.dunklau@aiv=
en.io>=20
wrote:
> > Le mercredi 11 octobre 2023, 10:59:50 CEST Thomas Munro a =C3=A9crit :
> > > The back-patch to 12 was a little trickier than anticipated, but after
> > > taking a break and trying again I now have PG 12...17 patches that
> > > I've tested against LLVM 10...18 (that's 54 combinations), in every
> > > case only with the clang corresponding to LLVM.
> >=20
> > Thank you Thomas for those patches, and the extensive testing, I will r=
un
> > my own and let you know.
>=20
> Thanks!  No news is good news, I hope?  I'm hoping to commit this today.
>=20
> > > I've attached only the patches for master, but the 12-16 versions are
> > > available at https://github.com/macdice/postgres/tree/llvm16-$N in
> > > case anyone has comments on those.
> >=20
> > For PG13 and PG12, it looks like the ExecEvalBoolSubroutineTemplate is =
not
> > used anywhere, as ExecEvalBoolSubroutine was introduced in PG14 if I'm =
not
> > mistaken.
>=20
> Right, looks like I can remove that in those branches.

Oh sorry I thought I followed up. I ran the same stress testing involving=20
several hours of sqlsmith with all jit costs set to zero and didn't notice=
=20
anything with LLVM16.

Thank you !

=2D-
Ronan Dunklau