Perma [tier2][fission] layout/base/tests/test_bug851485.html | Scrolling after load works. | Reloading keeps scroll position - got 4904, expected 500
Categories
(Core :: Layout, defect, P5)
Tracking
()
People
(Reporter: intermittent-bug-filer, Assigned: hiro)
References
(Regression)
Details
(Keywords: intermittent-failure, regression, Whiteboard: [stockwell disabled] fission-soft-blocker)
Attachments
(3 files, 2 obsolete files)
Filed by: apavel [at] mozilla.com
Parsed log: https://treeherder.mozilla.org/logviewer?job_id=322203872&repo=mozilla-central
Full log: https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/EprbhutiS8S5JCFc4qJ-fg/runs/0/artifacts/public/logs/live_backing.log
[task 2020-11-18T18:09:37.967Z] 18:09:37 INFO - TEST-START | layout/base/tests/test_bug851485.html
[task 2020-11-18T18:09:38.682Z] 18:09:38 INFO - TEST-INFO | started process screentopng
[task 2020-11-18T18:09:39.041Z] 18:09:39 INFO - TEST-INFO | screentopng: exit 0
[task 2020-11-18T18:09:39.042Z] 18:09:39 INFO - Buffered messages logged at 18:09:38
[task 2020-11-18T18:09:39.042Z] 18:09:39 INFO - TEST-PASS | layout/base/tests/test_bug851485.html | Reloading keeps scroll position
[task 2020-11-18T18:09:39.042Z] 18:09:39 INFO - Buffered messages finished
[task 2020-11-18T18:09:39.043Z] 18:09:39 INFO - TEST-UNEXPECTED-FAIL | layout/base/tests/test_bug851485.html | Scrolling after load works.
[task 2020-11-18T18:09:39.043Z] 18:09:39 INFO - SimpleTest.ok@SimpleTest/SimpleTest.js:417:16
[task 2020-11-18T18:09:39.043Z] 18:09:39 INFO - verifyScroll@layout/base/tests/test_bug851485.html:19:5
[task 2020-11-18T18:09:39.043Z] 18:09:39 INFO - verifyAfterLoad/observer<@layout/base/tests/test_bug851485.html:39:21
[task 2020-11-18T18:09:39.044Z] 18:09:39 INFO - GECKO(13352) | MEMORY STAT | vsize 2690MB | residentFast 152MB | heapAllocated 21MB
[task 2020-11-18T18:09:39.045Z] 18:09:39 INFO - TEST-OK | layout/base/tests/test_bug851485.html | took 747ms
[task 2020-11-18T18:09:39.173Z] 18:09:39 INFO - Not taking screenshot here: see the one that was previously logged
[task 2020-11-18T18:09:39.174Z] 18:09:39 INFO - TEST-UNEXPECTED-FAIL | layout/base/tests/test_bug851485.html | Reloading keeps scroll position - got 4904, expected 500
[task 2020-11-18T18:09:39.175Z] 18:09:39 INFO - SimpleTest.is@SimpleTest/SimpleTest.js:500:14
[task 2020-11-18T18:09:39.176Z] 18:09:39 INFO - verifyAfterLoad@layout/base/tests/test_bug851485.html:34:7
[task 2020-11-18T18:09:39.176Z] 18:09:39 INFO - onload@layout/base/tests/test_bug851485.html:1:1
[task 2020-11-18T18:09:39.198Z] 18:09:39 INFO - TEST-PASS | layout/base/tests/test_bug851485.html | Scrolling after load works.
[task 2020-11-18T18:09:39.199Z] 18:09:39 ERROR - TEST-UNEXPECTED-FAIL | SimpleTest | this test already called finish!
[task 2020-11-18T18:09:39.202Z] 18:09:39 INFO - TEST-UNEXPECTED-ERROR | layout/base/tests/test_bug851485.html | called finish() multiple times
[task 2020-11-18T18:09:39.202Z] 18:09:39 INFO - TEST-INFO ```
|
||
Updated•4 years ago
|
|
||
Updated•4 years ago
|
|
||
Comment 1•4 years ago
|
||
Set release status flags based on info from the regressing bug 1677689
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
|
||
Comment 4•3 years ago
|
||
Should bug 1677689 be backed out, if it didn't actually fix the test on fission? Or the test skipped again?
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
|
|
||
Updated•3 years ago
|
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
|
||
Comment 15•3 years ago
|
||
There are 31 total failures in the last 7 days on
- linux1804-64-shippable opt
- windows10-64-shippable opt
- windows10-64-shippable-qr opt
Recent failure log: https://treeherder.mozilla.org/logviewer?job_id=329926926&repo=mozilla-central&lineNumber=7169
[task 2021-02-14T12:02:53.464Z] 12:02:53 INFO - TEST-START | layout/base/tests/test_bug851485.html
[task 2021-02-14T12:02:54.017Z] 12:02:54 INFO - TEST-INFO | started process screenshot
[task 2021-02-14T12:02:54.082Z] 12:02:54 INFO - TEST-INFO | screenshot: exit 0
[task 2021-02-14T12:02:54.084Z] 12:02:54 INFO - Buffered messages logged at 12:02:53
[task 2021-02-14T12:02:54.084Z] 12:02:54 INFO - TEST-PASS | layout/base/tests/test_bug851485.html | We set the location's hash to 'anchor'
[task 2021-02-14T12:02:54.084Z] 12:02:54 INFO - TEST-PASS | layout/base/tests/test_bug851485.html | Reloading keeps scroll position
[task 2021-02-14T12:02:54.084Z] 12:02:54 INFO - Buffered messages finished
[task 2021-02-14T12:02:54.084Z] 12:02:54 INFO - TEST-UNEXPECTED-FAIL | layout/base/tests/test_bug851485.html | Scrolling after load works.
[task 2021-02-14T12:02:54.084Z] 12:02:54 INFO - SimpleTest.ok@SimpleTest/SimpleTest.js:417:16
[task 2021-02-14T12:02:54.084Z] 12:02:54 INFO - runTest@layout/base/tests/test_bug851485.html:58:5
[task 2021-02-14T12:02:54.084Z] 12:02:54 INFO - GECKO(4352) | MEMORY STAT | vsize 2104099MB | vsizeMaxContiguous 69085050MB | residentFast 99MB | heapAllocated 9MB
[task 2021-02-14T12:02:54.084Z] 12:02:54 INFO - TEST-OK | layout/base/tests/test_bug851485.html | took 585ms
Peter can you take a look?
|
|
||
Updated•3 years ago
|
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
|
||
Comment 19•3 years ago
|
||
There are 31 total failures in the last 7 days on:
- windows10-64-shippable-qr
- windows10-64-shippable
-linux1804-64-shippable-qr
-linux1804-64-shippable
"INFO - TEST-START | layout/base/tests/test_bug851485.html
[task 2021-02-28T23:34:17.506Z] 23:34:17 INFO - TEST-INFO | started process screenshot
[task 2021-02-28T23:34:17.576Z] 23:34:17 INFO - TEST-INFO | screenshot: exit 0
[task 2021-02-28T23:34:17.576Z] 23:34:17 INFO - Buffered messages logged at 23:34:17
[task 2021-02-28T23:34:17.576Z] 23:34:17 INFO - TEST-PASS | layout/base/tests/test_bug851485.html | We set the location's hash to 'anchor'
[task 2021-02-28T23:34:17.576Z] 23:34:17 INFO - TEST-PASS | layout/base/tests/test_bug851485.html | Reloading keeps scroll position
[task 2021-02-28T23:34:17.576Z] 23:34:17 INFO - Buffered messages finished
[task 2021-02-28T23:34:17.576Z] 23:34:17 INFO - TEST-UNEXPECTED-FAIL | layout/base/tests/test_bug851485.html | Scrolling after load works.
[task 2021-02-28T23:34:17.576Z] 23:34:17 INFO - SimpleTest.ok@SimpleTest/SimpleTest.js:417:16
[task 2021-02-28T23:34:17.576Z] 23:34:17 INFO - runTest@layout/base/tests/test_bug851485.html:58:5
[task 2021-02-28T23:34:17.577Z] 23:34:17 INFO - GECKO(4336) | MEMORY STAT | vsize 2103921MB | vsizeMaxContiguous 65007037MB | residentFast 98MB | heapAllocated 10MB
[task 2021-02-28T23:34:17.577Z] 23:34:17 INFO - TEST-OK | layout/base/tests/test_bug851485.html | took 583ms"
Peter can you please take a look?
|
|
||
Updated•3 years ago
|
|
||
Comment 20•3 years ago
|
||
|
||
Updated•3 years ago
|
| Comment hidden (Intermittent Failures Robot) |
|
||
Comment 22•3 years ago
•
|
||
Atila, you don't need to land your patch to disable test_bug851485.html for Fission. ahal just disabled this same test for Fission in bug 1694833:
https://hg.mozilla.org/integration/autoland/rev/4b99cb914aad#l1.9
[test_bug851485.html]
skip-if =
toolkit == 'android' # Bug 1355821
fission && (os == 'win' || os == 'mac') # Bug 1694833 - initial triage
But let's keep this bug open for the Fission bug fix. I am tentatively reassigning this bug to Peter because this bug was filed as a regression from his test_bug851485.html changes in bug 1677689.
|
|
||
Updated•3 years ago
|
| Comment hidden (Intermittent Failures Robot) |
|
|
||
Comment 25•3 years ago
|
||
This is probably a test bug. Tracking for Fission M8 Release.
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
|
||
Comment 28•3 years ago
|
||
Update:
There have been 34 failures within the last 7 days:
- 19 failures on Linux 18.04 x64 WebRender Shippable opt
- 15 failures on Linux 18.04 x64 WebRender opt
Recent failure: https://treeherder.mozilla.org/logviewer?job_id=335092468&repo=autoland&lineNumber=11216
[task 2021-03-31T22:44:27.468Z] 22:44:27 INFO - TEST-PASS | layout/base/tests/test_bug851485.html | Reloading keeps scroll position
[task 2021-03-31T22:44:27.468Z] 22:44:27 INFO - Buffered messages finished
[task 2021-03-31T22:44:27.469Z] 22:44:27 INFO - TEST-UNEXPECTED-FAIL | layout/base/tests/test_bug851485.html | Scrolling after load works.
[task 2021-03-31T22:44:27.470Z] 22:44:27 INFO - SimpleTest.ok@SimpleTest/SimpleTest.js:417:16
[task 2021-03-31T22:44:27.470Z] 22:44:27 INFO - runTest@layout/base/tests/test_bug851485.html:58:5
[task 2021-03-31T22:44:27.470Z] 22:44:27 INFO - GECKO(1972) | MEMORY STAT | vsize 2675MB | residentFast 156MB | heapAllocated 18MB
[task 2021-03-31T22:44:27.470Z] 22:44:27 INFO - TEST-OK | layout/base/tests/test_bug851485.html | took 678ms
|
||
Updated•3 years ago
|
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
| Comment hidden (Intermittent Failures Robot) |
|
|
||
Comment 34•3 years ago
|
||
Hi Peter, are there any updates here?
There are 33 total failures in the last 7 days on
- linux1804-64-qr opt
- linux1804-64-shippable opt
- linux1804-64-shippable-qr opt
Recent failure log: https://treeherder.mozilla.org/logviewer?job_id=338776520&repo=autoland&lineNumber=7682
[task 2021-05-05T04:30:39.210Z] 04:30:39 INFO - TEST-START | layout/base/tests/test_bug851485.html
[task 2021-05-05T04:30:39.820Z] 04:30:39 INFO - TEST-INFO | started process screentopng
[task 2021-05-05T04:30:40.142Z] 04:30:40 INFO - TEST-INFO | screentopng: exit 0
[task 2021-05-05T04:30:40.142Z] 04:30:40 INFO - Buffered messages logged at 04:30:39
[task 2021-05-05T04:30:40.142Z] 04:30:40 INFO - TEST-PASS | layout/base/tests/test_bug851485.html | We set the location's hash to 'anchor'
[task 2021-05-05T04:30:40.143Z] 04:30:40 INFO - TEST-PASS | layout/base/tests/test_bug851485.html | Reloading keeps scroll position
[task 2021-05-05T04:30:40.143Z] 04:30:40 INFO - Buffered messages finished
[task 2021-05-05T04:30:40.144Z] 04:30:40 INFO - TEST-UNEXPECTED-FAIL | layout/base/tests/test_bug851485.html | Scrolling after load works.
[task 2021-05-05T04:30:40.144Z] 04:30:40 INFO - SimpleTest.ok@SimpleTest/SimpleTest.js:417:16
[task 2021-05-05T04:30:40.145Z] 04:30:40 INFO - runTest@layout/base/tests/test_bug851485.html:58:5
[task 2021-05-05T04:30:40.145Z] 04:30:40 INFO - GECKO(12403) | MEMORY STAT | vsize 2691MB | residentFast 144MB | heapAllocated 18MB
[task 2021-05-05T04:30:40.146Z] 04:30:40 INFO - TEST-OK | layout/base/tests/test_bug851485.html | took 632ms
|
|
||
Updated•3 years ago
|
| Comment hidden (Intermittent Failures Robot) |
|
|
||
Comment 36•3 years ago
|
||
|
||
Comment 37•3 years ago
|
||
Pushed by abutkovits@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/fd8143f262ec disable test_bug851485.html on Linux_fission for frequent failures. r=intermittent-reviewers,ahal DONTBUILD
|
|
||
Updated•3 years ago
|
|
|
||
Comment 38•3 years ago
|
||
| bugherder | ||
|
||
Comment 39•3 years ago
|
||
peterv, just want to be sure this is still on your radar; it seems like we think this is just a test bug, but having said that, it's a test that was intended/expected to work with fission, per your changes in bug 1677689.
We've got this flagged as an M8 Fission blocker (per comment 25) at the moment, which means this is coming up in Fission nag/burndown charts -- so I want to be sure it's doesn't slip through the cracks. (Having said that: worst-case, since we're pretty confident about it being a test bug, I could also imagine this being morphed into "real-soon" post-fission-release followup work, if needed.)
|
|
||
Comment 40•3 years ago
|
||
Notably, this test does use sendMouseEvent just before the test failure; and that API is known to be potentially-broken with fission, per bug 1528935.
If that API is failing here (i.e. if we failed to send the click to the link-that-is-expected-to-trigger-scrolling), that would easily explain this test failure.
|
|
||
Comment 41•3 years ago
|
||
(having said that: I think the sendMouseEvent call here is for a same-origin iframe. So it's presumably not crossing any process boundaries. I don't know enough about our event-handling code before/after fission to know whether that changes things or not; but it seems like it might mean that this usage should be supported and unrelated to bug 1528935...)
|
|
||
Comment 42•3 years ago
|
||
|
|
||
Comment 43•3 years ago
|
||
Pushed by dholbert@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/3051f9d50710 diagnostic: Add some logging to mochitest test_bug851485.html, to provide more info when it fails. (no review)
|
|
||
Comment 44•3 years ago
|
||
On a whim, I wrote a diagnostic patch (just landed) to log some more info when the test fails, and I tried running the test locally with this patch applied and with fission enabled last week, on a computer that I wasn't actively using.
I encountered this bug's failure only once, and unfortunately rr had a hiccup during that recording and didn't successfully capture the issue. I ran the test for several compute-days after that (with rr) and didn't hit the failure again, so I wasn't able to diagnose further.
Hopefully we can reproduce it on infra with this diagnostic patch (maybe in a try run, since it's disabled on trunk right now), and gain a bit more info; or maybe someone can capture it in rr and gain a bit more information about what's going on.
|
||
Comment 45•3 years ago
|
||
FWIW, I tried and failed to reproduce the issue on linux debug and opt with --verify.
|
|
||
Comment 46•3 years ago
|
||
|
|
||
Comment 47•3 years ago
|
||
|
|
||
Comment 48•3 years ago
|
||
Still happens on Windows and Mac.
|
|
||
Comment 49•3 years ago
|
||
The try run from comment 47 didn't yet have my diagnostic patch from comment 43 (because it seems autoland hasn't merged to central in a little while).
I did a Try run based off of a recent autoland commit, to get some test failures with the additional logging. Here's the try run:
https://treeherder.mozilla.org/jobs?repo=try&revision=b09326a3937762ccf60970248db42c9a85add324
And here's a failure in this test on Mac (the only one I see so far):
https://treeherder.mozilla.org/logviewer?job_id=343525491&repo=try&lineNumber=6684
The logging says:
INFO - TEST-PASS | layout/base/tests/test_bug851485.html | We set the location's hash to 'anchor'
INFO - TEST-PASS | layout/base/tests/test_bug851485.html | Reloading keeps scroll position
INFO - Sending click
INFO - IntersectionObserver: entering callback
INFO - IntersectionObserver: some entry isIntersecting; disconnecting
INFO - Promise resolved (for IntersectionObserver). win.scrollY is 0
INFO - Buffered messages finished
INFO - TEST-UNEXPECTED-FAIL | layout/base/tests/test_bug851485.html | Scrolling after load works.
So we have one more piece of information here, which is win.scrollY is 0 at the time of the test-failure. We're expecting to have scrolled (by at least 3000px) but we're not scrolling at all, it seems.
|
|
||
Comment 50•3 years ago
|
||
Also, one side note: given the use of IntersectionObserver in the test, it's conceivable that there's a connection between this bug and bug 1671539 (another apparently-fission-specific IntersectionObserver-related test failure).
|
||
Comment 51•3 years ago
|
||
| bugherder | ||
|
|
||
Comment 52•3 years ago
|
||
This bug is a soft blocker for Fission M8. We'd like to fix it before our M8 Release experiment, but we won't delay the experiment waiting for it.
|
|
||
Comment 53•3 years ago
|
||
I managed to catch this in rr on Linux (by running the test continuously on 4 machines overnight). I'll see if I can get to the bottom of what's going on...
|
|
||
Comment 54•3 years ago
|
||
Pernosco trace: https://pernos.co/debug/-IllENi-iynvkylXutktsQ/index.html
I'm partway through stepping through it. My high-level takeaway right now is: the test seems to be expecting the document to stay the same (i.e. no navigation) from the point where the IntersectionObserver is constructed. But in fact the document does navigate (i.e. we get OnPageHide/OnPageShow, and a new Document object, with an initially-null root frame). So, we get a win.scrollY of 0 (per comment 49) because we haven't yet constructed frames in the new document at the point where it gets queried, which means we take this return CSSIntPoint(0, 0); statement:
https://searchfox.org/mozilla-central/rev/8f08c21f093be1c1c42438697f8bca67af94fc77/dom/base/nsGlobalWindowOuter.cpp#4108-4110
I haven't yet quite figured out what all this means, but at this point I'm suspecting/hoping it's a bug in the test's expectations, rather than an actual fission bug.
|
Assignee | |
Comment 55•3 years ago
|
||
(In reply to Daniel Holbert [:dholbert] from comment #54)
Pernosco trace: https://pernos.co/debug/-IllENi-iynvkylXutktsQ/index.html
I'm partway through stepping through it. My high-level takeaway right now is: the test seems to be expecting the document to stay the same (i.e. no navigation) from the point where the IntersectionObserver is constructed. But in fact the document does navigate (i.e. we get OnPageHide/OnPageShow, and a new Document object, with an initially-null root frame).
This comment made me think that this load event is not the one for the reload. Triggering the runTest function after the load event for the iframe's window received fixed this failure.
Stealing.
|
|
Assignee | |
Comment 56•3 years ago
|
||
Though the test funtion is called twice since the parent document reloads the
iframe document, but the test will not run for the second call since it's
prevented by testRunning flag in test_bug842853.html.
|
|
||
Comment 57•3 years ago
•
|
||
(In reply to Hiroyuki Ikezoe (:hiro) from comment #55)
(In reply to Daniel Holbert [:dholbert] from comment #54)
Pernosco trace: https://pernos.co/debug/-IllENi-iynvkylXutktsQ/index.html
I'm partway through stepping through it. My high-level takeaway right now is: the test seems to be expecting the document to stay the same (i.e. no navigation) from the point where the IntersectionObserver is constructed. But in fact the document does navigate (i.e. we get OnPageHide/OnPageShow, and a new Document object, with an initially-null root frame).
This comment made me think that this load event is not the one for the reload.
Yup, I had the same realization! I was going to post a patch with a slightly different approach (adjusting the test file to ensure we get a load event before we call reload on the frame and wait for a second one), but this makes sense too.
(I was imagining that maybe the test had a reason for firing runTest() before load -- the fact that it has a slowly-loading .sjs CSS file seems like maybe it is intentionally delaying load to happen later... But I can't find any mention of that or reason why the scrolling bug might involves fiddly load-timing on the original bug page that added this file_bug842853.html helper. So I think this is fine and probably the original test could have used the load event like you're changing it to do here.)
Thanks for the patch!
|
|
||
Comment 58•3 years ago
|
||
Pushed by hikezoe.birchill@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/8397357bef1f Trigger the test function after the child document received the load event. r=dholbert
|
||
Comment 59•3 years ago
|
||
| bugherder | ||
|
|
Assignee | |
Updated•3 years ago
|
|
|
||
Updated•3 years ago
|
|
|
||
Updated•3 years ago
|
|
|
||
Updated•3 years ago
|
Description
•