[Pkg-samba-maint] failing salsa reprotest job

[Michael Tokarev]

03.04.2022 11:59, Mathieu Parent wrote:
> On Sun, Apr 3, 2022 at 8:53 AM Michael Tokarev <mjt at tls.msk.ru> wrote:
>>
>> I'm seeing a failing reprotest job on salsa.
>> It fails at configure (!) stage when the build
>> is happening the second time.
> 
> I have no idea. But the 2nd build works at:
> https://tests.reproducible-builds.org/debian/rb-pkg/unstable/amd64/samba.html

There the build is running under fakeroot (as shown in the logs).
This thing definitely works under fakeroot, since the syscalls
(such as setreuid()) are intercepted and handled by fakeroot
itself.

Here on reprotest salsa-ci job, it is run as real root, so we
see more or less real syscall behaviour. More is because yes,
it is actually handled by the kernel, and less is because the
second run has complications such as unshare, setarch, taskset
and something else.  First run is done without all this stuff.

Actually, it seems, this reprotest salsa-ci job is the only
place where samba code is testing actual syscall behaviour,
instead of either testing something emulated by fakeroot or
not testing at all because the calling user is non-root.

Heh.  I didn't expect *real* setuid autoconf test, the one
that checks the actual *behaviour* of the syscall and not
just its presence. Wow.

/mjt