phone clock goes years into the future during power tests

[Selene ToyKeeper]

This may be related to Bug 1475542, but instead of going to the past to a hardcoded date before a test, the phone goes years into the future during or after a test.

Again I'm not quite sure where to file this bug, but I've noticed that the phone's time/date sometimes jump ahead to a fairly consistent range of dates in the future while power tests are running. This only happens during/after a test, and even happens later in the same boot after I force ntpdate to set the correct time.

On arale devices, the clock jumps to 2021-01-2X. On krillin devices, it jumps to 2017-01-2X. I've seen it happen 136 times, and only during/after a test, never before. So, the test forces the correct date via ntpdate, disconnects to collect power measurements, then reconnects to collect logs. During this post-test log collection is when I see the severely skewed date.

I see these date symptoms only during two tests... the ones where wifi is turned off.
     89 power_usage_bt_on_wifi_disconnected
     47 power_usage_flight_mode_on

Power test logs are kept here: http://people.canonical.com/~platform-qa/power-results/?C=M;O=D
Each test set and test run there includes quite a bit of data and logs to help with debugging.

The specific recent test runs with this symptom are:
Arale tests:
2015-07-14_07:19:59-arale-56-power_usage_flight_mode_on/6
2015-07-14_08:39:52-arale-56-power_usage_bt_on_wifi_disconnected/3
2015-07-14_20:33:23-arale-58-power_usage_flight_mode_on/6
2015-07-15_03:33:10-arale-60-power_usage_bt_on_wifi_disconnected/2
2015-07-15_07:09:48-arale-59-power_usage_bt_on_wifi_disconnected/3
2015-07-15_12:12:39-arale-60-power_usage_flight_mode_on/3
2015-07-16_15:26:03-arale-60-power_usage_flight_mode_on/2
2015-07-16_15:26:03-arale-60-power_usage_flight_mode_on/3
2015-07-16_16:55:11-arale-60-power_usage_bt_on_wifi_disconnected/2
2015-07-17_00:28:56-arale-61-power_usage_bt_on_wifi_disconnected/2
2015-07-17_00:28:56-arale-61-power_usage_bt_on_wifi_disconnected/3
2015-07-17_00:28:56-arale-61-power_usage_bt_on_wifi_disconnected/4
2015-07-17_00:28:56-arale-61-power_usage_bt_on_wifi_disconnected/5
2015-07-17_00:28:56-arale-61-power_usage_bt_on_wifi_disconnected/6
2015-07-17_00:28:56-arale-61-power_usage_bt_on_wifi_disconnected/7

... and krillin tests:
2015-07-13_17:52:04-krillin-65-power_usage_flight_mode_on/2
2015-07-13_17:52:04-krillin-65-power_usage_flight_mode_on/4
2015-07-13_17:52:04-krillin-65-power_usage_flight_mode_on/5
2015-07-13_17:52:04-krillin-65-power_usage_flight_mode_on/6
2015-07-13_19:15:15-krillin-65-power_usage_bt_on_wifi_disconnected/1
2015-07-13_19:15:15-krillin-65-power_usage_bt_on_wifi_disconnected/2
2015-07-13_19:15:15-krillin-65-power_usage_bt_on_wifi_disconnected/3
2015-07-13_19:15:15-krillin-65-power_usage_bt_on_wifi_disconnected/4
2015-07-13_19:15:15-krillin-65-power_usage_bt_on_wifi_disconnected/5
2015-07-13_19:15:15-krillin-65-power_usage_bt_on_wifi_disconnected/6
2015-07-13_19:15:15-krillin-65-power_usage_bt_on_wifi_disconnected/7
2015-07-13_23:05:48-krillin-132-power_usage_flight_mode_on/2
2015-07-13_23:05:48-krillin-132-power_usage_flight_mode_on/3
2015-07-13_23:05:48-krillin-132-power_usage_flight_mode_on/6
2015-07-13_23:05:48-krillin-132-power_usage_flight_mode_on/7
2015-07-14_00:38:10-krillin-132-power_usage_bt_on_wifi_disconnected/1
2015-07-14_00:38:10-krillin-132-power_usage_bt_on_wifi_disconnected/2
2015-07-14_00:38:10-krillin-132-power_usage_bt_on_wifi_disconnected/3
2015-07-14_00:38:10-krillin-132-power_usage_bt_on_wifi_disconnected/4
2015-07-14_00:38:10-krillin-132-power_usage_bt_on_wifi_disconnected/6
2015-07-14_00:38:10-krillin-132-power_usage_bt_on_wifi_disconnected/7
2015-07-14_00:38:10-krillin-132-power_usage_bt_on_wifi_disconnected/8

Is there enough data in the logs there to identify the cause of this issue? I can collect more logs or command output if necessary, to include with future test runs.