Closed Bug 1161150 Opened 10 years ago Closed 10 years ago

Intermittent Android 4.3 PROCESS-CRASH | Automation Error: Missing end of test marker (process crashed?) during testFilePicker

Categories

(Firefox for Android Graveyard :: General, defect)

ARM
Android
defect
Not set
normal

Tracking

(firefox38 unaffected, firefox38.0.5 unaffected, firefox39 unaffected, firefox40 fixed, firefox-esr31 unaffected, firefox-esr38 unaffected)

RESOLVED FIXED
Firefox 40
Tracking Status
firefox38 --- unaffected
firefox38.0.5 --- unaffected
firefox39 --- unaffected
firefox40 --- fixed
firefox-esr31 --- unaffected
firefox-esr38 --- unaffected

People

(Reporter: RyanVM, Assigned: mcomella)

References

Details

(Keywords: crash, intermittent-failure)

07:51:42 INFO - SimpleTest START 07:51:42 INFO - TEST-START | testFilePicker 07:51:42 INFO - INFO | automation.py | Application ran for: 0:05:45.341078 07:51:42 INFO - INFO | zombiecheck | Reading PID log: /tmp/tmpck5edepidlog 07:51:42 INFO - Contents of /data/anr/traces.txt: 07:51:42 INFO - 07:51:42 INFO - 07:51:42 INFO - Stopping web server 07:51:42 INFO - Stopping web socket server 07:51:42 INFO - Stopping ssltunnel 07:51:42 INFO - WARNING | leakcheck | refcount logging is off, so leaks can't be detected! 07:51:42 INFO - runtests.py | Running tests: end. 07:51:42 WARNING - PROCESS-CRASH | Automation Error: Missing end of test marker (process crashed?)
Unlike comments 1 and 2, comments 3 - 9 are crashes in testInputUrlBar, on Android 2.3.
I suspect a regression from bug 1137483.
Blocks: 1137483
wrt testInputUrlBar, it seems we always fail after [1]: assertUrlBarText(mStringHelper.ABOUT_HOME_URL); It seems just after this we drag the home pager to lose focus from the url bar, click the url bar, and enter some text; we seem to crash because the application does not respond to our input events. If we're always failing on 2.3, this is an issue around slowness - either race condition or doing too much at the same time. Bug 1137483 added some additional layout functionality - perhaps that is making the emulator just slow enough to cause these issues. Notably, Gecko is probably still loading in the background. Some proposed fixes: * Wait a bit longer before performing the actions after goBack * Wait for Gecko:Ready [1]: https://mxr.mozilla.org/mozilla-central/source/mobile/android/tests/browser/robocop/testInputUrlBar.java?rev=3dae1f0d8b0b#100
(In reply to Michael Comella (:mcomella) from comment #13) > * Wait for Gecko:Ready Nevermind - we already do this.
Before the failure, we also see: 10:27:11 INFO - 05-07 09:24:40.246 I/GeckoConsole( 488): 1431015880250 Toolkit.GMP WARN GMPInstallManager.parseResponseXML got node name: html, expected: updates 10:27:11 INFO - 05-07 09:24:40.306 E/GeckoConsole( 488): [JavaScript Error: "1431015880274 Toolkit.GMP ERROR GMPWrapper(gmp-gmpopenh264) findUpdates() - updateTask for gmp-gmpopenh264 threw: {"target":{},"message":"got node name: html, expected: updates"}" {file: "resource://gre/modules/Log.jsm" line: 749}] 10:27:11 INFO - 05-07 09:25:05.115 W/WindowManager( 64): Input event injection timed out. 10:27:11 INFO - 05-07 09:25:10.135 I/InputDispatcher( 64): Application is not responding: AppWindowToken{4068b9b8 token=HistoryRecord{405f1230 Going to see if I can see this JS error locally.
Before the failure, we also see: 09:04:08 INFO - 05-07 08:04:18.358 D/Telemetry( 490): StartUISession: frecency.1 ^ This means the swipe event happens as expected, and we manage to click the url bar again. We type some content and the browser search screen opens. That means we're either failing inside sendKeys or getUrlBarText [1]. 09:04:08 INFO - 05-07 08:04:19.808 D/GeckoLayerClient( 490): Window-size changed to (1024,572) ^ Not sure why this is happening though notably liuche mentioned the new search bar hides under the keyboard on 2.3, even though we fixed it on other versions such that this does not happen. This probably means the window is getting resized (when it shouldn't) when BrowserSearch is shown. We also add a window background only on the BrowserSearch screen - perhaps this overdraw is adding too much extra overhead to 2.3? [1]: https://mxr.mozilla.org/mozilla-central/source/mobile/android/tests/browser/robocop/testInputUrlBar.java?rev=3dae1f0d8b0b#112
I thought fixing bug 1159360 might fix the issues here. After tracking that down (in bug 1159360), I made some conclusions in bug 1159360 comment 6. The try push in comment 21 makes the same backout as in the other comment. Note that this backout breaks things and is not a final solution.
Assignee: nobody → michael.l.comella
Fixed by the backout of bug 1137483 comment 44.
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → FIXED
Product: Firefox for Android → Firefox for Android Graveyard
You need to log in before you can comment on or make changes to this bug.