I think it's the kernel itself that can't read the system clock, not systemd:
bruno@T400 ~ $ sudo journalctl -b -p err -- Logs begin at Wed 1969-12-31 19:00:06 EST, end at Fri 2016-01-29 09:00:20 EST. -- Dec 31 19:00:02 T400 kernel: rtc_cmos 00:03: hctosys: unable to read the hardware clock Dec 31 19:00:04 T400 systemd[1]: libcontainer-353-systemd-test-default-dependencies.scope: Scope has no PIDs. Refusing. This rabbit hole goes down pretty deep! I would have no idea how to tell the kernel to do a "directisa" read of the hardware clock. I'll keep investigating.
