Having non-ascii characters in log is not helpful.
It is even worse for Jenkins jobs, as non-ascii characters appearing sufficiently soon make the "file" utility classify the test output as binary. Only text files are being compressed upon archival [0], so our result-processing job gets confused by a different file name.

The non-ascii string seems to be initialized at [1], and displayed at [2] via [3]. I suspect it is the usual confusion between null-terminated strings and u8 vectors.

[0] https://github.com/FDio/ci-management/blob/4ed95452d982886c61e28c11f8737fdde207de69/jjb/scripts/post_build_deploy_archives.sh#L118-L123
[1] https://github.com/FDio/vpp/blob/fa065f96d14da55117bbf52a1f85e38d835dffd0/src/plugins/rdma/device.c#L891-L894
[2] https://github.com/FDio/vpp/blob/fa065f96d14da55117bbf52a1f85e38d835dffd0/src/plugins/rdma/device.c#L185
[3] https://github.com/FDio/vpp/blob/fa065f96d14da55117bbf52a1f85e38d835dffd0/src/plugins/rdma/device.c#L48-L52