s390x autopkgtest hangs

[Benjamin Drung]

The s390x autopkgtest hangs [1]. The issue happens when executing the test test_get_interface [2] and the execution freezes on sock.recv(8192) in line 82 [3] until the test times out.

[1] https://autopkgtest.ubuntu.com/results/autopkgtest-mantic/mantic/s390x/n/nvme-stas/20230711_031309_d317c@/log.gz
[2] https://salsa.debian.org/debian/nvme-stas/-/blob/debian/sid/test/test-iputil.py#L31
[3] https://salsa.debian.org/debian/nvme-stas/-/blob/debian/sid/staslib/iputil.py#L82

[Mateus Rodrigues de Morais (mateus-morais)]

This issue was caused by a hard-coded byteorder='little' on a big-endian architecture (s390x), which resulted in garbage messages being sent to the Netlink socket. As these are ignored, nothing was then being received on recv(), causing the subsequent hang and, ultimately, the autopkgtest timeout.

A fix has been proposed at https://code.launchpad.net/~mateus-morais/ubuntu/+source/nvme-stas/+git/nvme-stas/+merge/450606.

[Benjamin Drung (bdrung)]

Fixed in 2.2.2-2.

[Launchpad Janitor (janitor)]

This bug was fixed in the package nvme-stas - 2.2.2-2

---------------
nvme-stas (2.2.2-2) unstable; urgency=medium

  [ Mateus Rodrigues de Morais ]
  * d/p/Fix-byte-ordering-in-iputil-operations.patch: make byteorder
    architecture-aware in order to fix s390x autopkgtest failures.
    (LP: #2026878)

 -- Benjamin Drung <email address hidden> Tue, 05 Sep 2023 13:45:22 +0200

[Olivier Gayot (ogayot)]

Sadly, the fix for this bug was dropped when Debian moved to nvme-stas 2.3-1.

I filed it again as bug 2045000 (the code had slightly changed in the meantime) and the fix is now merged upstream.