repowerd

Bug #1602933
Comment #0

Comment 0 for bug 1602933

Revision history for this message

Selene ToyKeeper (toykeeper) wrote on 2016-07-14:

As of yesterday's phone images (2016-07-12, krillin rc-p 383, arale rc-p 376, when repowerd landed), I'm getting unusual behavior during automated tests. This is easiest to see, manually, with video playback.

Steps to trigger this issue:
* Flash the affected image.
* Put a video on the device, at least 2 minutes long.
* Start video playback. Symptoms seem most severe when played via "adb shell" with "ubuntu-app-launch mediaplayer-app file:///media/phablet/3562-3231/Videos/long-video.avi".
* Wait at least 60 seconds.
* Unplug USB.

Expected result: Phone stays active until video playback completes.

Actual result: Phone suspends after a few seconds. Depending on how the video was started, sound may or may not continue playing.

Playing a video from inside the video scope (without launching mediaplayer-app) results in sound but no visuals. Playing it with mediaplayer-app at a command line (like the automated tests do) results in a full suspend with no sound.

I'm getting similar behavior during other automated tests too. What happens is, roughly:
* USB connect.
* Reboot.
* Adt runs to set up the device. This takes a few minutes.
* USB disconnect.

At this point, what should happen is:
* If the current app has no "keep screen on" state, the screen should dim after 50 seconds then shut off after another 10 seconds. Otherwise, the screen should stay on until the "keep screen on" state is cleared. (adt sets this state sometimes to allow testing power with the display on, or it should also stay on until the video finishes)
* Measurements start after a while, and continue for a pre-determined length of time.

What actually happens instead:
* A few seconds after USB gets unplugged, the screen turns off. If music isn't playing, the phone also goes into suspend mode. The "keep screen on" state appears to be ignored, and the user's screen-off delay setting appears to be ignored too. No 50-seconds-then-dim-then-10-seconds-then-off happens.
* Measurements happen at the expected times, but the phone isn't in the expected state so the results are wrong.

Will attach comparison measurements below.

As of yesterday's phone images (2016-07-12, krillin rc-p 383, arale rc-p 376, when repowerd landed), I'm getting unusual behavior during automated tests.  This is easiest to see, manually, with video playback.

Steps to trigger this issue:
* Flash the affected image.
* Put a video on the device, at least 2 minutes long.
* Start video playback.  Symptoms seem most severe when played via "adb shell" with "ubuntu-app-launch mediaplayer-app file:///media/phablet/3562-3231/Videos/long-video.avi".
* Wait at least 60 seconds.
* Unplug USB.

Expected result: Phone stays active until video playback completes.

Actual result: Phone suspends after a few seconds.  Depending on how the video was started, sound may or may not continue playing.

Playing a video from inside the video scope (without launching mediaplayer-app) results in sound but no visuals.  Playing it with mediaplayer-app at a command line (like the automated tests do) results in a full suspend with no sound.

I'm getting similar behavior during other automated tests too.  What happens is, roughly:
* USB connect.
* Reboot.
* Adt runs to set up the device.  This takes a few minutes.
* USB disconnect.

At this point, what should happen is:
* If the current app has no "keep screen on" state, the screen should dim after 50 seconds then shut off after another 10 seconds.  Otherwise, the screen should stay on until the "keep screen on" state is cleared.  (adt sets this state sometimes to allow testing power with the display on, or it should also stay on until the video finishes)
* Measurements start after a while, and continue for a pre-determined length of time.

What actually happens instead:
* A few seconds after USB gets unplugged, the screen turns off.  If music isn't playing, the phone also goes into suspend mode.  The "keep screen on" state appears to be ignored, and the user's screen-off delay setting appears to be ignored too.  No 50-seconds-then-dim-then-10-seconds-then-off happens.
* Measurements happen at the expected times, but the phone isn't in the expected state so the results are wrong.

Will attach comparison measurements below.