1771288 - A11y regression: UIAutomator unable to locate widget content elements in all app clients (UI Tests on mobile apps failing)

Status: VERIFIED FIXED

(Core :: Disability Access APIs, defect, P1)

Description

[Oana Horvath [:ohorvath]]

+++ This bug was initially created as a clone of Bug #1769510 +++
Previous cases of this issue: https://bugzilla.mozilla.org/show_bug.cgi?id=1758540 affected a lot more tests, but after the latest fix, it still left a few tests broken.
Assuming this is GeckoView related as there is nothing else on A-C upgrade that is causing issues.
Started with AC versions: 102.0.20220524015637 and 102.0.20220524143152

UiAutomator, which we use for locating elements in Gecko on Android, is not working for validating page content in some UI tests, on the test pages from this domain:
https://itisatrap.org/firefox/harmful.html
http://itisatrap.org/firefox/its-an-attack.html
http://itisatrap.org/firefox/its-a-trap.html
http://itisatrap.org/firefox/unwanted.html
the "Address not found" page of a bad URL

Failing tests reports logged:
from Focus:
https://github.com/mozilla-mobile/focus-android/issues/7113
https://github.com/mozilla-mobile/focus-android/issues/6733

from Fenix:
https://github.com/mozilla-mobile/fenix/issues/25356
https://github.com/mozilla-mobile/fenix/issues/25394
https://github.com/mozilla-mobile/fenix/issues/25380

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1758540

Comment 2

[BugBot [:suhaib / :marco/ :calixte]]

:eeejay, since you are the author of the regressor, bug 1758540, could you take a look?
For more information, please visit auto_nag documentation.

Comment 3

[Oana Horvath [:ohorvath]]

Also found this in the failed test logs:
E/QueryController(13723): Got null root node from accessibility - Retrying... 05-26 15:33:19.553: W/audio_hw_generic(1862): Hardware backing HAL too slow, could only write 0 of 661 frames 05-26 15:33:19.643: D/CompatibilityChangeReporter(478): Compat change id reported: 135634846; UID 10152; state: DISABLED 05-26 15:33:19.646: D/CompatibilityChangeReporter(478): Compat change id reported: 143937733; UID 10152; state: ENABLED 05-26 15:33:19.642: I/chatty(1862): uid=1041(audioserver) writer identical 6 lines 05-26 15:33:19.657: W/audio_hw_generic(1862): Hardware backing HAL too slow, could only write 0 of 661 frames 05-26 15:33:19.669: D/Zygote(289): Forked child process 15855 05-26 15:33:19.672: W/audio_hw_generic(1862): Hardware backing HAL too slow, could only write 0 of 661 frames 05-26 15:33:19.681: I/sh.CrashHandle(15855): Late-enabling -Xcheck:jni 05-26 15:33:19.682: I/ActivityManager(478): Start proc 15855:org.mozilla.focus.debug:mozilla.components.lib.crash.CrashHandler/u0a152 for service {org.mozilla.focus.debug/mozilla.components.lib.crash.handler.CrashHandlerService}
Test reports here:
https://github.com/mozilla-mobile/focus-android/issues?q=is%3Aissue+is%3Aopen++Got+null+root+node+from+accessibility+-+Retrying

Comment 4

[Eitan Isaacson [:eeejay]]

I think I fixed this issue in bug 1772019. Can you track it when it lands in a-c and see if it does the trick?

Fixed the Fenix errors for me locally, but I coudn't figure out how to load a local geckoview into Focus to test.

Comment 5

[Oana Horvath [:ohorvath]]

Hi! We re-ran the affected tests and we can confirm that the latest fix is working.
For this: https://bugzilla.mozilla.org/show_bug.cgi?id=1771288#c3 I couldn't yet find a proof that it's related to the same issue, so I'll close this as fixed. Thank you!

Comment 6

[Ryan VanderMeulen [:RyanVM]]

Note that eeejay's fix landed in 103. But we're preffing the feature off for 102 in bug 1772165, so it shouldn't be affected there eventually.