964918 - test_kill / test_kill_multiple fails with JavascriptException: TypeError: this._activeApp is undefined / app://system.gaiamobile.org/js/app_window_manager.js, line 516

Status: VERIFIED FIXED

(Firefox OS Graveyard :: General, defect)

Description

[Stephen Donner [:stephend] Not actively reading bugmail]

Attachment: logcat.txt

In our Jenkins CI, running test_kill_multiple a few times can eventually fail, with the following stack:

Traceback (most recent call last):
File "/var/jenkins/workspace/b2g.hamachi.mozilla-central.unittests.test_kill_multiple/.env/local/lib/python2.7/site-packages/marionette_client-0.7.2-py2.7.egg/marionette/marionette_test.py", line 143, in run
testMethod()
File "/var/jenkins/workspace/b2g.hamachi.mozilla-central.unittests.test_kill_multiple/tests/python/gaia-ui-tests/gaiatest/tests/unit/test_kill.py", line 25, in test_kill_multiple
self.apps.launch(app.name)
File "/var/jenkins/workspace/b2g.hamachi.mozilla-central.unittests.test_kill_multiple/tests/python/gaia-ui-tests/gaiatest/gaia_test.py", line 81, in launch
result = self.marionette.execute_async_script("GaiaApps.launchWithName('%s')" % name, script_timeout=launch_timeout)
File "/var/jenkins/workspace/b2g.hamachi.mozilla-central.unittests.test_kill_multiple/.env/local/lib/python2.7/site-packages/marionette_client-0.7.2-py2.7.egg/marionette/marionette.py", line 1080, in execute_async_script
filename=os.path.basename(frame[0]))
File "/var/jenkins/workspace/b2g.hamachi.mozilla-central.unittests.test_kill_multiple/.env/local/lib/python2.7/site-packages/marionette_client-0.7.2-py2.7.egg/marionette/marionette.py", line 584, in _send_message
self._handle_error(response)
File "/var/jenkins/workspace/b2g.hamachi.mozilla-central.unittests.test_kill_multiple/.env/local/lib/python2.7/site-packages/marionette_client-0.7.2-py2.7.egg/marionette/marionette.py", line 619, in _handle_error
raise JavascriptException(message=message, status=status, stacktrace=stacktrace)
JavascriptException: TypeError: this._activeApp is undefined
stacktrace:
@app://system.gaiamobile.org/js/app_window_manager.js, line 516

Comment 1

[Stephen Donner [:stephend] Not actively reading bugmail]

See https://developer.mozilla.org/en-US/Firefox_OS/Platform/Automated_testing/gaia-ui-tests for how to get an env. set up with Gaia UI Tests.

Comment 2

[Milan Sreckovic [:milan] (needinfo for best results)]

There are some interesting messages in there:

E/msm7627a.hwcomposer(  853): drawLayerUsingCopybit: copybit stretch failed
E/msm7627a.hwcomposer(  853): hwc_set: Copybit layer draw failed!

Comment 3

[Diego Wilson [:diego]]

(In reply to Milan Sreckovic [:milan] from comment #2)
> There are some interesting messages in there:
> 
> E/msm7627a.hwcomposer(  853): drawLayerUsingCopybit: copybit stretch failed
> E/msm7627a.hwcomposer(  853): hwc_set: Copybit layer draw failed!

These messages are not fatal. They're more like warnings. The only effect should be that it falls back to GPU composition of that frame.

Comment 4

[Diego Wilson [:diego]]

Are you seeing graphic artifacts?

Comment 5

[Stephen Donner [:stephend] Not actively reading bugmail]

(In reply to Diego Wilson [:diego] from comment #4)
> Are you seeing graphic artifacts?

If you can VPN in, there are screenshots here: http://qa-selenium.mv.mozilla.com:8080/job/b2g.hamachi.mozilla-central.unittests.test_kill_multiple/15/HTML_Report/?

Comment 6

[Stephen Donner [:stephend] Not actively reading bugmail]

Attachment: 1st-screenshot.png

Comment 7

[Stephen Donner [:stephend] Not actively reading bugmail]

Attachment: 2nd-screenshot.png

Comment 8

[Diego Wilson [:diego]]

Interesting. Is the issue reproducible if you disable hardware composer?

Settings -> Device Information -> More Information -> Developer -> Enable hardware composer -> uncheck

If not, can you make sure you have these changes?

https://www.codeaurora.org/cgit/quic/la/platform/hardware/qcom/display/commit/?id=4716987848deaa573b14219b0322a16b29038d29
https://www.codeaurora.org/cgit/quic/la/platform/hardware/qcom/display/commit/?id=0c12df86df998019b35ade29e5f24e144866fc4d

Comment 9

[Sotaro Ikeda [:sotaro]]

> 
> If not, can you make sure you have these changes?

I do not know which device is used to test this. About Hamachi/buri, the change is already applied. Hamachi's default repository is caf.

>  <default revision="b2g/ics_strawberry"
>           remote="caf"
>           sync-j="4" />

>  <project path="hardware/qcom/display" name="platform/hardware/qcom/display" revision="b2g_ics_1.2" />

https://github.com/mozilla-b2g/b2g-manifest/blob/master/hamachi.xml

Comment 10

[Robert Wood [:rwood]]

Hi :diego, is the 'enable hardware composer' setting available via the settings API? i.e.:

https://developer.mozilla.org/en-US/Firefox_OS/Platform/Settings_list

If it can be set programmatically by the automation framework (post-build), will the setting be persistent between device reboots? Thanks :)

Comment 11

[Diego Wilson [:diego]]

(In reply to Robert Wood [:rwood] from comment #10)
> Hi :diego, is the 'enable hardware composer' setting available via the
> settings API?

Yes it can. The setting name is "layers.composer2d.enabled".

> If it can be set programmatically by the automation framework (post-build),
> will the setting be persistent between device reboots? Thanks :)

Yes.

Comment 12

[Stephen Donner [:stephend] Not actively reading bugmail]

Bob, Bebe, Zac: can someone please try to manipulate this within the job, remotely (or, locally, if you can reproduce), and post the results here?  Thanks!

Comment 13

[Zac C (:zac)]

I don't think Graphics is the problem.

Looking at the actual test code and the exception it is throwing, this seems like a race condition between launching and killing apps quickly using testing javascript atoms.

Although Alive might be interested in the race condition as an edge case for when AppWindowManager can get tripped up, I think this is most likely a gaiatest problem and one for us to address.

Comment 14

[Zac C (:zac)]

Fixed:
https://github.com/mozilla-b2g/gaia/commit/5535f688a7997c3b5c96b49b596779c847c1e00c

Comment 15

[Stephen Donner [:stephend] Not actively reading bugmail]

(In reply to Zac C (:zac) from comment #14)
> Fixed:
> https://github.com/mozilla-b2g/gaia/commit/
> 5535f688a7997c3b5c96b49b596779c847c1e00c

Thx Zac!