Bug#1136156: postgresql-common: Please ignore testsuite results on m68k

Hi,

On Sun, May 10, 2026 at 03:33:42PM +0200, John Paul Adrian Glaubitz wrote:
> On Sun, 2026-05-10 at 15:06 +0200, Michael Banck wrote:
> > > > > 2026-05-09 22:54:05.199 UTC client backend[3478032] [unknown] PA=
NIC:  stuck spinlock detected at LWLockWaitListLock, ./build/../src/backen=
d/storage/lmgr/lwlock.c:888
> > > > > qemu: uncaught target signal 6 (Aborted) - core dumped
> > > > > 2026-05-09 22:54:05.436 UTC postmaster[3477436] LOG:  client bac=
kend (PID 3478032) was terminated by signal 6: Aborted
> > > > > 2026-05-09 22:54:05.436 UTC postmaster[3477436] LOG:  terminatin=
g any other active server processes
> > > > > 2026-05-09 22:54:05.446 UTC postmaster[3477436] LOG:  all server=
 processes terminated; reinitializing
> > > >=20
> > >=20
> > > That's why I wrote I built the package locally with gcc-16 where
> > > that issue is fixed. I just uploaded the locally built package.
> >=20
> > The above is from the build log you cited, but maybe I am just confuse=
d
> > then about why you posted a link to it.
>=20
> Yes, you're right, of course. But that's part of the testsuite, no?

Well yes, but it implies the binary package is likely broken as well, it
is not just a matter of some quirky different behaviour.

However, what we just committed for Hurd is this:
https://salsa.debian.org/postgresql/postgresql-common/-/commit/9e57389e878=
45f61843627bac473a140c7d394df

This helped not getting the VM hanging due to too much concurrency, you
could try to see whether it helps on m68k as well.


Michael