At the moment, the PAPR RTC implementation (actually a paravirt firmware interface, rather than a normal device) works directly off host time, and so doesn't respect the options such as clock=vm which can be specified in the -rtc command line option.
This series is a first cut at addressing this. Caveats: * I haven't yet investigated how clock=vm will interact with migration. I suspect badly. * We also use a time stamp when delivering RTAS events in spapr_events.c. For now I haven't changed the code here which uses host time (via qemu_get_timedate()). It's not entirely clear which clock these timestamps should be linked to.