Issue metadata
Sign in to add a comment
|
chrome_public_test_vr_apk-pixelxl-ddview-nougat on Android failing on chromium.android on Android failing on chromium.android/Nougat Phone Tester |
||||||||||||||||||||||||||
Issue descriptionFiled by sheriff-o-matic@appspot.gserviceaccount.com on behalf of aberent@google.com chrome_public_test_vr_apk-pixelxl-cardboard-nougat on Android failing on chromium.android/Nougat Phone Tester Builders failed on: - Nougat Phone Tester: https://build.chromium.org/p/chromium.android/builders/Nougat%20Phone%20Tester There seem to be two different failure modes: - java.lang.AssertionError: Polling JavaScript boolean javascriptDone timed out. Did not obtain specific reason from testharness, apparently coming from enterFullscreenOrFail - a simple java.lang.AssertionError from enterPresentationOrFail Both are happening in multiple tests, see https://logs.chromium.org/v/?s=chromium%2Fbb%2Fchromium.android%2FNougat_Phone_Tester%2F2950%2F%2B%2Frecipes%2Fsteps%2Fchrome_public_test_vr_apk-pixelxl-ddview-nougat_on_Android%2F0%2Fstdout
,
Jan 23 2018
,
Jan 23 2018
Sheriff-O-Matic suggests the CL range 531109 - 531137 (https://test-results.appspot.com/revision_range?start=531109&end=531137). I suspect https://chromium-review.googlesource.com/c/chromium/src/+/879115 (since this is the only CL mentioning VR in this range), but I am not sure how to tell my device that it has a daydream viewer, so I am unable to reproduce the problem.
,
Jan 23 2018
Assigning to Brian Sheedy, who landed the suspect CL.
,
Jan 23 2018
Guessing, but chrome_public_test_vr_apk-pixelxl-cardboard-nougat step seem to have started failing at the same time, so probably has the same cause.
,
Jan 23 2018
Taking a look now.
,
Jan 23 2018
I'm unable to reproduce locally. I'll keep digging, but feel free to revert if it's causing issues. I'm not actually convinced that it was the test framework change, though. The same set of failures seemed to occur at https://ci.chromium.org/buildbot/chromium.android/Nougat%20Phone%20Tester/2944, with one green build between that and when my CL landed. The linked build has the 2D UI fallback path CL in it, which has the potential to break a lot of things. Not sure why there would be a green build in the middle, though.
,
Jan 23 2018
Talked with the author of the 2D UI fallback CL, and at least some of the failures are likely caused by a null check that somehow got omitted from a merge or something, causing segfaults. Fix incoming.
,
Jan 23 2018
Looking at the first build with the fix in place on our FYI bot (https://ci.chromium.org/buildbot/chromium.fyi/Android%20VR%20Tests/15427), it didn't fix everything, but it did bring the number of failed cardboard tests down to one and it's no longer due to a segfault. So, progress. Will continue to investigate.
,
Jan 23 2018
The remaining test failures appear to be due to an incompatibility between DOMUtils and the 2D UI fallback path. Re-assigning to Michael.
,
Jan 29 2018
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/90cb6716a63971dc8ec61f00415326c9606aa69d commit 90cb6716a63971dc8ec61f00415326c9606aa69d Author: Matthew Cary <mattcary@chromium.org> Date: Mon Jan 29 11:22:40 2018 vr_shell: disable tests causing bot failures VrShellTransitionTest#testExitPresentationWebVrToVrShell and VrShellNavigationTest#test2dFullscreenToWebVr continue to fail and keep the bot red. Bug: 804808 TBR: bsheedy@chromium.org,mthiesse@chromium.org Change-Id: I1721df42012b11662998f54627a8dd561c1a3253 Reviewed-on: https://chromium-review.googlesource.com/890182 Reviewed-by: Matthew Cary <mattcary@chromium.org> Commit-Queue: Matthew Cary <mattcary@chromium.org> Cr-Commit-Position: refs/heads/master@{#532392} [modify] https://crrev.com/90cb6716a63971dc8ec61f00415326c9606aa69d/chrome/android/javatests/src/org/chromium/chrome/browser/vr_shell/VrShellNavigationTest.java [modify] https://crrev.com/90cb6716a63971dc8ec61f00415326c9606aa69d/chrome/android/javatests/src/org/chromium/chrome/browser/vr_shell/VrShellTransitionTest.java
,
Jan 29 2018
VrShellTransitionTest#testExitPresentationWebVrToVrShell and VrShellNavigationTest#test2dFullscreenToWebVr are still failing (the former timing out the test, the latter with what appears to be the problem already referenced on the bug). I'm disabling those two tests in order to get the bot green as they seem to be the only things failing, and they've been failing consistently for days.
,
Jan 29 2018
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/f719bcbaa6f13ba9e997b8570a59b8a354dc1a26 commit f719bcbaa6f13ba9e997b8570a59b8a354dc1a26 Author: bsheedy <bsheedy@chromium.org> Date: Mon Jan 29 23:20:18 2018 Disable all failing VR tests Disables more failing VR tests that were missed in the previous CL due to crbug.com/804808 . Bug: 804808 Change-Id: I64692f52891ee333c807b2e12be08d096de625b7 Reviewed-on: https://chromium-review.googlesource.com/892079 Reviewed-by: Michael Thiessen <mthiesse@chromium.org> Commit-Queue: Brian Sheedy <bsheedy@chromium.org> Cr-Commit-Position: refs/heads/master@{#532651} [modify] https://crrev.com/f719bcbaa6f13ba9e997b8570a59b8a354dc1a26/chrome/android/javatests/src/org/chromium/chrome/browser/vr_shell/VrShellControllerInputTest.java [modify] https://crrev.com/f719bcbaa6f13ba9e997b8570a59b8a354dc1a26/chrome/android/javatests/src/org/chromium/chrome/browser/vr_shell/VrShellNavigationTest.java [modify] https://crrev.com/f719bcbaa6f13ba9e997b8570a59b8a354dc1a26/chrome/android/javatests/src/org/chromium/chrome/browser/vr_shell/VrShellTransitionTest.java
,
Jan 30 2018
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/c7b73e44c8fb778fd329ddfd2c48c24ddb51367f commit c7b73e44c8fb778fd329ddfd2c48c24ddb51367f Author: Michael Thiessen <mthiesse@chromium.org> Date: Tue Jan 30 17:58:10 2018 Allow injecting input directly into r.id.content for DOMUtils. This works around an issue where DOMUtils injects its input through the rootView (view.getRootView()), which causes the input to be swallowed by the GvrLayout in VR (this is because in VR, screen touches are used for headset alignment rather than user input). DOMUtils clicks will optionally get injected directly into r.id.content rather than the root of the view hierarchy. We need to inject into r.id.content, rather that cvc.getContainerView() because the CompositorViewHolder needs to be able to intercept the events to correctly set the EventForwarder offsets. Gross. Bug: 804808 Change-Id: I6b246262d1e7b64437382b67db0a0d10518ee839 Reviewed-on: https://chromium-review.googlesource.com/884010 Commit-Queue: Michael Thiessen <mthiesse@chromium.org> Reviewed-by: David Trainor <dtrainor@chromium.org> Cr-Commit-Position: refs/heads/master@{#532940} [modify] https://crrev.com/c7b73e44c8fb778fd329ddfd2c48c24ddb51367f/chrome/android/javatests/src/org/chromium/chrome/browser/vr_shell/VrShellControllerInputTest.java [modify] https://crrev.com/c7b73e44c8fb778fd329ddfd2c48c24ddb51367f/chrome/android/javatests/src/org/chromium/chrome/browser/vr_shell/VrShellNavigationTest.java [modify] https://crrev.com/c7b73e44c8fb778fd329ddfd2c48c24ddb51367f/chrome/android/javatests/src/org/chromium/chrome/browser/vr_shell/VrShellTransitionTest.java [modify] https://crrev.com/c7b73e44c8fb778fd329ddfd2c48c24ddb51367f/chrome/android/javatests/src/org/chromium/chrome/browser/vr_shell/util/TransitionUtils.java [modify] https://crrev.com/c7b73e44c8fb778fd329ddfd2c48c24ddb51367f/content/public/test/android/javatests/src/org/chromium/content/browser/test/util/DOMUtils.java [modify] https://crrev.com/c7b73e44c8fb778fd329ddfd2c48c24ddb51367f/content/public/test/android/javatests/src/org/chromium/content/browser/test/util/TouchCommon.java
,
Jan 30 2018
Bots seem to be happy with the DOMUtils fix and the tests re-enabled.
,
Feb 8 2018
,
Mar 1 2018
,
Aug 29
|
|||||||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||||||
Comment 1 by aber...@chromium.org
, Jan 23 2018