1705279 - Intermittent [tier2] gfx/layers/apz/test/mochitest/test_group_checkerboarding.html | Test timed out. -

Status: RESOLVED FIXED

(Core :: Panning and Zooming, defect, P5)

Description

[Treeherder Bug Filer]

Filed by: dluca [at] mozilla.com
Parsed log: https://treeherder.mozilla.org/logviewer?job_id=336581192&repo=autoland
Full log: https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/fiq9X56ZRsqiBAQls3zqXA/runs/0/artifacts/public/logs/live_backing.log

---

INFO - TEST-PASS | gfx/layers/apz/test/mochitest/test_group_checkerboarding.html | Starting subtest helper_checkerboard_zoomoverflowhidden.html 
[task 2021-04-15T02:37:28.812Z] 02:37:28     INFO - TEST-PASS | gfx/layers/apz/test/mochitest/test_group_checkerboarding.html | helper_checkerboard_zoomoverflowhidden.html | The initial_resolution is 1, which is some sane value 
[task 2021-04-15T02:37:28.812Z] 02:37:28     INFO - TEST-PASS | gfx/layers/apz/test/mochitest/test_group_checkerboarding.html | helper_checkerboard_zoomoverflowhidden.html | expected at least one nonempty paint 
[task 2021-04-15T02:37:28.812Z] 02:37:28     INFO - TEST-PASS | gfx/layers/apz/test/mochitest/test_group_checkerboarding.html | helper_checkerboard_zoomoverflowhidden.html | subframe should be layerized at this point 
[task 2021-04-15T02:37:28.812Z] 02:37:28     INFO - TEST-PASS | gfx/layers/apz/test/mochitest/test_group_checkerboarding.html | helper_checkerboard_zoomoverflowhidden.html | subframe should have a scroll id 
[task 2021-04-15T02:37:28.812Z] 02:37:28     INFO - TEST-PASS | gfx/layers/apz/test/mochitest/test_group_checkerboarding.html | helper_checkerboard_zoomoverflowhidden.html | The final resolution (3.0922749042510986) is greater than the initial resolution 
[task 2021-04-15T02:37:28.812Z] 02:37:28     INFO - Buffered messages finished
[task 2021-04-15T02:37:28.812Z] 02:37:28     INFO - TEST-UNEXPECTED-FAIL | gfx/layers/apz/test/mochitest/test_group_checkerboarding.html | Test timed out. - 
[task 2021-04-15T02:37:29.668Z] 02:37:29     INFO - GECKO(2153) | MEMORY STAT | vsize 6585MB | residentFast 83MB | heapAllocated 8MB
[task 2021-04-15T02:37:31.834Z] 02:37:31     INFO - TEST-OK | gfx/layers/apz/test/mochitest/test_group_checkerboarding.html | took 331258ms
[task 2021-04-15T02:37:35.143Z] 02:37:35     INFO - Error: Unable to restore focus, expect failures and timeouts.
[task 2021-04-15T02:37:35.143Z] 02:37:35     INFO - TEST-START | gfx/layers/apz/test/mochitest/test_group_double_tap_zoom.html```

Comment 1

[Intermittent Failures Robot]

31 failures in 870 pushes (0.036 failures/push) were associated with this bug yesterday.

Repository breakdown:

• autoland: 23
• mozilla-central: 7
• try: 1

Platform and build breakdown:

• macosx1015-64-qr: 21
  • opt: 9
  • debug: 12
• macosx1015-64-shippable-qr: 10
  • opt: 10

For more details, see:
https://treeherder.mozilla.org/intermittent-failures/bugdetails?bug=1705279&startday=2021-04-15&endday=2021-04-15&tree=all

Comment 2

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

Regression from bug 1704231?

Comment 3

[Botond Ballo [:botond]]

Confirmed via Try that this is a regression from bug 1704231.

Comment 4

[Timothy Nikkel (:tnikkel)]

This seems to be perma fail on mac right now? Meaning we are currently losing apz (and other area) and every test that comes after this test coverage on mac.

Comment 5

[Botond Ballo [:botond]]

In the Tier 2 WebRender suite, yes.

Comment 6

[Timothy Nikkel (:tnikkel)]

Which is the only suite of mochitests we run on mac. Anyways it seems it only almost perma fail on opt. We get some good runs on debug still.

Comment 7

[Botond Ballo [:botond]]

I tried reproducing the failure locally on Linux (with overscroll enabled manually), but wasn't able to.

Timothy or Hiro, could you check if you can reproduce the failure locally on Mac? If so, it may be more efficient to investigate it locally rather than via Try.

Comment 8

[Botond Ballo [:botond]]

Note, a local repro may need --enable-webrender as all failures so far are with WR.

Comment 9

[Timothy Nikkel (:tnikkel)]

Reproduces with and without wr.

Seems like we are waiting for this promise to resolve

https://searchfox.org/mozilla-central/rev/6a41d8d7ad42d57cf1adcc7067fd8f1689c42ab3/gfx/layers/apz/test/mochitest/helper_checkerboard_zoomoverflowhidden.html#86

We are trying to use those touch events to scroll up and to the left. The test seems to stop when we reach the top but haven't yet reach the left. So overscroll in the vert direction makes something go wrong?

Comment 10

[Timothy Nikkel (:tnikkel)]

So we zoom in, pan up and to the left to return to 0,0, then pan down and right.

During the pan up and left we are in state 3 for APZC 1 (we get many "got a touch-move in state 3"). Eventually those lead to:

APZC 2 dropping handoff; AllowImmediateHandoff=false
(1) APZC 2 taking overscroll during panning

Eventually the touch-move finish and we get "got a touch-move in state 3" and the handoff and overscroll messages stop too. APZC 1 changes state to NOTHING. APZC 2 changes state to OVERSCROLL_ANIMATION:

APZC 2 is overscrolled, starting snap-back
APZC 2 changing from state 0 to 9

We finish waiting for the promises falling the pan up and to the left call in the test and we send the pan down and to the right in the test, APZC 2 is still in state OVERSCROLL_ANIMATION. This seems to trigger APZC going into state NOTHING:

APZC 2 running CancelAnimation(0x3) in state 9

Now APZC 2 starts getting touch-moves it goes into state TOUCHING and then PANNING_LOCKED_X, it gets touch-end and goes into state NOTHING. We then get:

APZC 2 is overscrolled, starting snap-back
APZC 2 changing from state 0 to 9

again. And then we changes to state NOTHING and we are stuck waiting for a touch-end promise, even though we got the touch-end msg in the log. So perhaps apzc2 is the wrong place?

Comparing this to overscroll disabled we are different: at point (1) above we also get the "taking overscroll during panning" for APZC 1. We never go into state 9. and APZC 1 gets the touch moves and is in state 4, not APZC 2.

So maybe the touches for the pan down and to the right end up in the gutter and go to the wrong apzc?

Comment 11

[Timothy Nikkel (:tnikkel)]

APZC 2 is the rcd, APZC 1 is something else.

Comment 12

[Timothy Nikkel (:tnikkel)]

Attachment: Bug 1705279. Move test coords slightly in gfx/layers/apz/test/mochitest/helper_checkerboard_zoomoverflowhidden.html to avoid overscroll gutter. r?botond

Comment 13

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/1811dfb4d3f5
Move test coords slightly in gfx/layers/apz/test/mochitest/helper_checkerboard_zoomoverflowhidden.html to avoid overscroll gutter. r=botond

Comment 14

[Intermittent Failures Robot]

58 failures in 648 pushes (0.09 failures/push) were associated with this bug yesterday.

Repository breakdown:

• mozilla-central: 11
• autoland: 42
• try: 5

Platform and build breakdown:

• macosx1015-64-qr: 45
  • debug: 27
  • opt: 18
• macosx1015-64-shippable-qr: 13
  • opt: 13

For more details, see:
https://treeherder.mozilla.org/intermittent-failures/bugdetails?bug=1705279&startday=2021-04-16&endday=2021-04-16&tree=all

Comment 15

[Andreea Pavel [:apavel]]

https://hg.mozilla.org/mozilla-central/rev/1811dfb4d3f5

Comment 16

[Intermittent Failures Robot]

95 failures in 4255 pushes (0.022 failures/push) were associated with this bug in the last 7 days.

This is the #14 most frequent failure this week.

** This failure happened more than 75 times this week! Resolving this bug is a very high priority. **

** Try to resolve this bug as soon as possible. If unresolved for 1 week, the affected test(s) may be disabled. **

Repository breakdown:

• autoland: 65
• mozilla-central: 22
• try: 8

Platform and build breakdown:

• macosx1015-64-qr: 68
  • opt: 27
  • debug: 41
• macosx1015-64-shippable-qr: 27
  • opt: 27

For more details, see:
https://treeherder.mozilla.org/intermittent-failures/bugdetails?bug=1705279&startday=2021-04-12&endday=2021-04-18&tree=all