There is an intermittent test failure in gnome-remote-desktop, in
practice mainly affecting autopkgtest. It could also cause a build-time
test failure, but for whatever reason, in practice that has only been
observed on s390x.

The failure occurs in the "gnome-remote-desktop/rdp" test, and the
distinctive error message appears to be "RDP test client exited
incorrectly: -11", which I believe indicates that a test client
segfaulted. If I'm understanding correctly, this is not the actual
gnome-remote-desktop RDP server crashing, but instead it's a crash in a
RDP client that was run against it for testing purposes.

If a different test failure is observed, please report it separately
rather than entangling it with this one.

Sample output (from a s390x buildd log, back in March):

https://buildd.debian.org/status/fetch.php?pkg=gnome-remote-desktop&arch=s390x&ver=48%7Ealpha-2%2Bb1&stamp=1742129611&raw=0

Also observed in for example
<https://ci.debian.net/packages/g/gnome-remote-desktop/testing/amd64/61429293/>;
I would estimate that 5% to 10% of autopkgtest runs fail like this.

I do not have any specific knowledge about the internals of
gnome-remote-desktop or its test infrastructure. I am aware that some
project members consider me to be personally responsible for every test
failure in GNOME, and I am sorry that I cannot accept that
responsibility.

I suspect that a good first thing to try, for anyone interested in
investigating this, would be to reproduce it on a developer system by
running the tests in a loop, and get a stack trace from systemd-coredump
or similar. If it can be reproduced during build-time tests (by doing a
single build, then running the equivalent of dh_auto_test repeatedly)
then that is probably an easier and more efficient environment for
interactive debugging than the autopkgtest would be.

    smcv