Handling clocksource switch more gracefully
It looks like we are way more fragile over Dovetail when Linux decides to switch away from the TSC clocksource, due to real hardware issues (yes, even in 2021) or due to improperly configures hypervisor setups. The system may even lock up, seen with latency being started after echo acpi_pm > /sys/devices/system/clocksource/clocksource0/current_clocksource
.
While switching away from the TSC is clearly a sign that something is fundamentally wrong with the system, that sign needs to be identifiable. From a user perspective, it may just look like "Xenomai locked up".