All TSAN-enabled builds fail in AWS-based GitLab CI jobs

Yesterday's mass-rebuild of Docker images caused some update to be pulled into tsan-fedora-39-amd64 that does not play nicely with AWS hosts because all TSAN-enabled builds now fail with an error message like:

FATAL: ThreadSanitizer: unexpected memory mapping 0x7d00e0772000-0x7d00e0c00000

While it is not clear what exactly happened, here are two jobs that were run in CI for the same commit:

• 2024-03-20 14:24, passed
• 2024-03-20 16:41, failed

The refreshed TSAN image was pushed to the container registry at 15:13.

The TSAN builds seemingly still work fine with the refreshed TSAN image on our bare metal runners, which use older kernels. This is consistent with similar reports found online:

https://stackoverflow.com/questions/77850769/fatal-threadsanitizer-unexpected-memory-mapping-when-running-on-linux-kernels

The simplest course of action is to apply the workaround mentioned in the StackOverflow post above (sysctl vm.mmap_rnd_bits=28) and remove it once the issue resolves itself as kernels and packages get updated over time.