Open Bug 1641824 Opened 4 years ago Updated 2 years ago

Perma Tier 2 dt fission leakcheck | tab 209094 bytes leaked (APZEventState, ActiveElementManager, AsyncFreeSnowWhite, BackstagePass, BrowserChild, ...)

Categories

(DevTools :: General, defect, P5)

Product:
DevTools
Component:
General
Type:
defect

Tracking

(Fission Milestone:Future, firefox-esr68 unaffected, firefox77 unaffected, firefox78 affected, firefox79 affected)

Project Flags:
Fission Milestone Future
Tracking Flags:
Tracking Status
firefox-esr68 --- unaffected
firefox77 --- unaffected
firefox78 --- affected
firefox79 --- affected

People

(Reporter: intermittent-bug-filer, Unassigned)

References

(Blocks 1 open bug, Regression)

Details

(Keywords: intermittent-failure, regression, Whiteboard: [retriggered])

Filed by: btara [at] mozilla.com
Parsed log: https://treeherder.mozilla.org/logviewer.html#?job_id=304244448&repo=mozilla-central
Full log: https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/RxakMPy4QxmKfOOs_tbETQ/runs/0/artifacts/public/logs/live_backing.log

...
[task 2020-05-29T11:46:25.001Z] 11:46:24     INFO - TEST-INFO | leakcheck | tab leaked 1 nsXPConnect
[task 2020-05-29T11:46:25.001Z] 11:46:24     INFO - TEST-INFO | leakcheck | tab leaked 1 nsXULPopupManager
[task 2020-05-29T11:46:25.001Z] 11:46:24     INFO - TEST-INFO | leakcheck | tab leaked 1 nsXULPrototypeCache
[task 2020-05-29T11:46:25.001Z] 11:46:24     INFO - TEST-INFO | leakcheck | tab leaked 7 xpc::CompartmentPrivate
[task 2020-05-29T11:46:25.002Z] 11:46:24     INFO - TEST-UNEXPECTED-FAIL | leakcheck | tab 209094 bytes leaked (APZEventState, ActiveElementManager, AsyncFreeSnowWhite, BackstagePass, BrowserChild, ...)
[task 2020-05-29T11:46:25.002Z] 11:46:24     INFO - 
[task 2020-05-29T11:46:25.002Z] 11:46:24     INFO - leakcheck | Processing leak log file /tmp/tmpcXDx40.mozrunner/runtests_leaks_tab_pid4714.log
[task 2020-05-29T11:46:25.002Z] 11:46:24     INFO - 
[task 2020-05-29T11:46:25.003Z] 11:46:24     INFO - == BloatView: ALL (cumulative) LEAK AND BLOAT STATISTICS, tab process 4714
[task 2020-05-29T11:46:25.003Z] 11:46:24     INFO - 
[task 2020-05-29T11:46:25.003Z] 11:46:24     INFO -      |<----------------Class--------------->|<-----Bytes------>|<----Objects---->|
[task 2020-05-29T11:46:25.003Z] 11:46:24     INFO -      |                                      | Per-Inst   Leaked|   Total      Rem|
[task 2020-05-29T11:46:25.004Z] 11:46:24     INFO -    0 |TOTAL                                 |       34        0| 1163738        0|
[task 2020-05-29T11:46:25.004Z] 11:46:24     INFO - 
[task 2020-05-29T11:46:25.005Z] 11:46:24     INFO - nsTraceRefcnt::DumpStatistics: 1030 entries
...
Flags: needinfo?(poirot.alex)
Has Regression Range: --- → yes

Sorry, I don't know what this failure is about. My change was a metadata change, so I didn't touch the test or the execution environment.

:mccr8 - does this leak look like a real problem? If not, would it make sense to whitelist?

Flags: needinfo?(egao) → needinfo?(continuation)

Note that I also submitted a follow-up patch soon after the push in question. The follow-on patch is here and I've also triggered the same test there.

Failed on the follow-up patch too. I've also added a few retriggers on the first push (e31d66f10e6d) where it failed and a few for the previous one, where all of them were green.

Yeah, I see it failed on the follow-up push as well.

But again, I don't know why this is the case, I did not touch anything to do with the environment or the test itself.

I'll trigger them on my try pushes as well. Otherwise, I think waiting for mccr8 to chime in would be the best thing to do at the moment.

Sure, thank you for looking!

This leak is probably related to bug 1636192.
I fixed only leak that I reproduced locally, but there must have been others.

(In reply to Edwin Takahashi (:egao) from comment #2)

Sorry, I don't know what this failure is about. My change was a metadata change, so I didn't touch the test or the execution environment.

Could this have affected test chunking or something?

:mccr8 - does this leak look like a real problem?

It looks like a real problem. It looks similar to some other perma leaks we've had in devtools tests.

If not, would it make sense to whitelist?

We don't white list mochitests, only WPTs.

Flags: needinfo?(continuation)

(In reply to Andrew McCreight [:mccr8] from comment #9)

(In reply to Edwin Takahashi (:egao) from comment #2)

Sorry, I don't know what this failure is about. My change was a metadata change, so I didn't touch the test or the execution environment.

Could this have affected test chunking or something?

Yes, the changes I made may affecft test chunking as I changed the runtime dataset used in the chunking process. As a result it is possible chunks are slightly different than before the change.

If test chunking order is the reason for the leak, then I would imagine the path forward would be to fix the leak, not back out the chunking changes.

(In reply to Cristina Coroiu [:ccoroiu] from comment #1)

The leak is happening for devtools/client/webconsole/test/browser/browser_jsterm_syntax_highlight_output.js.

I'm not able to reproduce locally via
./mach mochitest --headless devtools/client/webconsole/test/browser/browser_jsterm_syntax_highlight_output.js
So it is probably a leak involving more than this one test.

Set release status flags based on info from the regressing bug 1637463

status-firefox77: --- → unaffected
status-firefox78: --- → affected
status-firefox79: --- → affected
status-firefox-esr68: --- → unaffected

Note that the reason why the retriggers point to 1637463 is because this patch modified the way the tests are chunked.
So the dt1 job before this push didn't contain the same tests. Now, either the new chunks are making the failure permanent because of a side effect between two suites. Or it only confused the retriggers, and we would find the failure on another dt* job.

Edit: According to https://treeherder.mozilla.org/#/jobs?repo=autoland&selectedTaskRun=SR_cztUlQNi-kiLDHAUWfA.0&searchStr=linux%2C18.04%2Cx64%2Cdebug%2Cmochitests%2Cwith%2Cfission%2Cenabled%2C&revision=2a44daf61d7f452aaf55f47c2962f8c2ed625704 , the failure was previously in dt4

This leak seems to have been fixed or is triaged via another bug?

Flags: needinfo?(poirot.alex)
Fission Milestone: --- → Future
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.