From: Tom Lane <tgl@sss.pgh.pa.us>
To: Noah Misch <noah@leadboat.com>
cc: Thomas Munro <thomas.munro@gmail.com>,
        Michael Paquier <michael@paquier.xyz>,
        Vik Fearing <vik@postgresfriends.org>,
 cary huang <hcary328@gmail.com>,
        pgsql-hackers@lists.postgresql.org
Subject: Re: Add support for AT LOCAL
In-reply-to: <20231016060510.72@rfd.leadboat.com>
References: 
 <CA+hUKGJE7rfnswHh021=vJmV3wAjmWfbb2BGp6CfeMk0rPyPLQ@mail.gmail.com>
 <CA+hUKGL04vme+yztreEFhGucv2d9ema+8_a+i43EUFdpYCPXfw@mail.gmail.com>
 <ZSyYqySa74x_RAH_@paquier.xyz>
 <CA+hUKGLYaVJ7df+Uu=ctUunRcqUsGY55feVQa-1pzz1=PT=z3w@mail.gmail.com>
 <2188264.1697425365@sss.pgh.pa.us>
 <CA+hUKGJSEOVonGsYCiuhZXsBFCxXtjNyHGYvn-w9jykGo2PgqQ@mail.gmail.com>
 <2191749.1697427017@sss.pgh.pa.us> <20231016045804.c0@rfd.leadboat.com>
 <20231016055051.3a@rfd.leadboat.com> <2208907.1697435663@sss.pgh.pa.us>
 <20231016060510.72@rfd.leadboat.com>
Comments: In-reply-to Noah Misch <noah@leadboat.com>
	message dated "Sun, 15 Oct 2023 23:05:10 -0700"
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-ID: <2659914.1697521218.1@sss.pgh.pa.us>
Content-Transfer-Encoding: quoted-printable
Date: Tue, 17 Oct 2023 01:40:18 -0400
Message-ID: <2659915.1697521218@sss.pgh.pa.us>
Archived-At: 
 <https://www.postgresql.org/message-id/2659915.1697521218%40sss.pgh.pa.us>
Precedence: bulk

Noah Misch <noah@leadboat.com> writes:
> On Mon, Oct 16, 2023 at 01:54:23AM -0400, Tom Lane wrote:
>> Ugh.  So if the failure is robust enough to persist across
>> several major xlc versions, why couldn't Thomas reproduce it?

> Beats me.  hornet wipes its starting environment down to OBJECT_MODE=3D3=
2_64
> PERL5LIB=3D/home/nm/sw/cpan64/lib/perl5 SPECIES=3Dxlc64 PATH=3D/usr/bin,=
 then
> applies all the environment settings seen in buildfarm logs.

I was able to reproduce the failure on cfarm111 after adopting
these settings from hornet's configuration:

export OBJECT_MODE=3D64
export CC=3D'xlc_r -D_LARGE_FILES=3D1 '
export CFLAGS=3D'-O2 -qmaxmem=3D33554432 -qsuppress=3D1500-010:1506-995 -q=
suppress=3D1506-010:1506-416:1506-450:1506-480:1506-481:1506-492:1506-944:=
1506-1264 -qinfo=3Dall:nocnd:noeff:noext:nogot:noini:noord:nopar:noppc:nor=
ea:nouni:nouse -qsuppress=3D1506-374:1506-419:1506-434:1506-438:1506-451:1=
506-452:1506-453:1506-495:1506-786'

and doing

./configure --enable-cassert --enable-debug --without-icu --prefix=3D/home=
/tgl/testversion

etc etc.

It is absolutely, gold-platedly, a compiler bug, because inserting
a debug printf into the loop

	while (result->time >=3D USECS_PER_DAY)
		result->time -=3D USECS_PER_DAY;

makes the failure go away.  Unfortunately, I've not yet found another
way to make it go away :-(.  My upthread idea of using a local variable
instead of result->time is no help, and some other random code
alterations didn't change the results either.

Not sure where we go from here.  While I don't have much hesitation
about blowing off xlc_r 12.1, it would be sad if their latest
toolchain doesn't work either.  (I didn't try permuting the code
while using the newer compiler, though.)

			regards, tom lane