Closed Bug 1540949 Opened 5 years ago Closed 5 years ago

Intermittent LeakSanitizer | leak at createTable, mozilla::detail::HashTable, reserve, reserve in web-platform-tests

Categories

(Core :: DOM: Service Workers, defect, P3)

defect

Tracking

()

RESOLVED WORKSFORME

People

(Reporter: intermittent-bug-filer, Unassigned)

References

Details

(Keywords: intermittent-failure, regression, Whiteboard: [stockwell disable-recommended])

#[markdown(off)]
Filed by: nbeleuzu [at] mozilla.com

https://treeherder.mozilla.org/logviewer.html#?job_id=237515122&repo=autoland

https://queue.taskcluster.net/v1/task/Xb-bEo9hTXi0UQP2BiL2aA/runs/0/artifacts/public/logs/live_backing.log

[task 2019-04-02T06:41:03.965Z] 06:41:03 INFO - TEST-OK | /service-workers/service-worker/xsl-base-url.https.html | took 1378ms
[task 2019-04-02T06:41:04.030Z] 06:41:04 INFO - PID 7988 | 1554187264022 Marionette INFO Stopped listening on port 2828
[task 2019-04-02T06:41:04.370Z] 06:41:04 INFO - PID 7988 | ###!!! [Parent][RunMessage] Error: Channel closing: too late to send/recv, messages will be lost
[task 2019-04-02T06:41:04.386Z] 06:41:04 INFO - PID 7988 | ###!!! [Child][MessageChannel] Error: (msgtype=0x3500FE,name=PContent::Msg_AccumulateChildHistograms) Closed channel: cannot send/recv
[task 2019-04-02T06:41:04.387Z] 06:41:04 INFO - PID 7988 | ###!!! [Child][MessageChannel] Error: (msgtype=0x3500FF,name=PContent::Msg_AccumulateChildKeyedHistograms) Closed channel: cannot send/recv
[task 2019-04-02T06:41:04.389Z] 06:41:04 INFO - PID 7988 | ###!!! [Child][MessageChannel] Error: (msgtype=0x350103,name=PContent::Msg_RecordDiscardedData) Closed channel: cannot send/recv
[task 2019-04-02T06:41:05.605Z] 06:41:05 INFO - PID 7988 | -----------------------------------------------------
[task 2019-04-02T06:41:05.605Z] 06:41:05 INFO - PID 7988 | Suppressions used:
[task 2019-04-02T06:41:05.605Z] 06:41:05 INFO - PID 7988 | count bytes template
[task 2019-04-02T06:41:05.605Z] 06:41:05 INFO - PID 7988 | 30 928 nsComponentManagerImpl
[task 2019-04-02T06:41:05.605Z] 06:41:05 INFO - PID 7988 | 611 17713 libfontconfig.so
[task 2019-04-02T06:41:05.605Z] 06:41:05 INFO - PID 7988 | 1 29 libglib-2.0.so
[task 2019-04-02T06:41:05.605Z] 06:41:05 INFO - PID 7988 | -----------------------------------------------------
[task 2019-04-02T06:41:05.687Z] 06:41:05 INFO - PID 7988 | -----------------------------------------------------
[task 2019-04-02T06:41:05.689Z] 06:41:05 INFO - PID 7988 | Suppressions used:
[task 2019-04-02T06:41:05.690Z] 06:41:05 INFO - PID 7988 | count bytes template
[task 2019-04-02T06:41:05.691Z] 06:41:05 INFO - PID 7988 | 30 928 nsComponentManagerImpl
[task 2019-04-02T06:41:05.693Z] 06:41:05 INFO - PID 7988 | 2 288 libfontconfig.so
[task 2019-04-02T06:41:05.694Z] 06:41:05 INFO - PID 7988 | -----------------------------------------------------
[task 2019-04-02T06:41:06.037Z] 06:41:06 INFO - PID 7988 | -----------------------------------------------------
[task 2019-04-02T06:41:06.038Z] 06:41:06 INFO - PID 7988 | Suppressions used:
[task 2019-04-02T06:41:06.039Z] 06:41:06 INFO - PID 7988 | count bytes template
[task 2019-04-02T06:41:06.039Z] 06:41:06 INFO - PID 7988 | 30 928 nsComponentManagerImpl
[task 2019-04-02T06:41:06.040Z] 06:41:06 INFO - PID 7988 | 611 17713 libfontconfig.so
[task 2019-04-02T06:41:06.040Z] 06:41:06 INFO - PID 7988 | 1 29 libglib-2.0.so
[task 2019-04-02T06:41:06.041Z] 06:41:06 INFO - PID 7988 | -----------------------------------------------------
[task 2019-04-02T06:41:06.047Z] 06:41:06 INFO - PID 7988 | -----------------------------------------------------
[task 2019-04-02T06:41:06.047Z] 06:41:06 INFO - PID 7988 | Suppressions used:
[task 2019-04-02T06:41:06.047Z] 06:41:06 INFO - PID 7988 | count bytes template
[task 2019-04-02T06:41:06.047Z] 06:41:06 INFO - PID 7988 | 30 928 nsComponentManagerImpl
[task 2019-04-02T06:41:06.047Z] 06:41:06 INFO - PID 7988 | 611 17713 libfontconfig.so
[task 2019-04-02T06:41:06.048Z] 06:41:06 INFO - PID 7988 | 1 29 libglib-2.0.so
[task 2019-04-02T06:41:06.048Z] 06:41:06 INFO - PID 7988 | -----------------------------------------------------
[task 2019-04-02T06:41:08.394Z] 06:41:08 INFO - PID 7988 | -----------------------------------------------------
[task 2019-04-02T06:41:08.394Z] 06:41:08 INFO - PID 7988 | Suppressions used:
[task 2019-04-02T06:41:08.395Z] 06:41:08 INFO - PID 7988 | count bytes template
[task 2019-04-02T06:41:08.395Z] 06:41:08 INFO - PID 7988 | 27 824 nsComponentManagerImpl
[task 2019-04-02T06:41:08.395Z] 06:41:08 INFO - PID 7988 | 4 832 mozJSComponentLoader::LoadModule
[task 2019-04-02T06:41:08.396Z] 06:41:08 INFO - PID 7988 | 611 17509 libfontconfig.so
[task 2019-04-02T06:41:08.396Z] 06:41:08 INFO - PID 7988 | 366 16104 _PR_Getfd
[task 2019-04-02T06:41:08.398Z] 06:41:08 INFO - PID 7988 | 1 29 libglib-2.0.so
[task 2019-04-02T06:41:08.398Z] 06:41:08 INFO - PID 7988 | -----------------------------------------------------
[task 2019-04-02T06:41:08.588Z] 06:41:08 INFO - Browser exited with return code 0
[task 2019-04-02T06:41:08.594Z] 06:41:08 INFO - PROCESS LEAKS None
[task 2019-04-02T06:41:08.595Z] 06:41:08 INFO - ERROR | LeakSanitizer | SUMMARY: AddressSanitizer: 8456 byte(s) leaked in 60 allocation(s).
[task 2019-04-02T06:41:08.595Z] 06:41:08 INFO - LeakSanitizer | To show the addresses of leaked objects add report_objects=1 to LSAN_OPTIONS
[task 2019-04-02T06:41:08.596Z] 06:41:08 INFO - This can be done in testing/mozbase/mozrunner/mozrunner/utils.py
[task 2019-04-02T06:41:08.596Z] 06:41:08 INFO - Allowed depth was 4
[task 2019-04-02T06:41:08.596Z] 06:41:08 INFO - TEST-FAIL | LeakSanitizer | leak at Alloc, nsTSubstring, nsTSubstring, nsTSubstring
[task 2019-04-02T06:41:08.596Z] 06:41:08 INFO - INFO | LeakSanitizer | Frame Alloc matched a expected leak
[task 2019-04-02T06:41:08.596Z] 06:41:08 INFO - TEST-FAIL | LeakSanitizer | leak at Alloc, nsTSubstring, Gecko_StartBulkWriteString, nsstring::nsAString::start_bulk_write_impl

Whiteboard: [stockwell disable-recommended] → [stockwell needswork:owner]

comments 3, 4 and 5: matches original, happening WPT. Mostly in service-workers/service-worker/.

Comment 6 (from 4/22): most of the failures match the original WPT leaks, but there are a handful of a second leak. This second leak is in opt-mochitest-devtools-chrome-e10s-4, not in web platform tests. Also, instead of only 2 unexpected-fails, there are either 70 or 95. This is a different issue. The leaks are happening in the directory devtools/client/debugger.

Comment 7 (from 4/24): Again, most of theses failures are the original WPT leaks. There's a second set of 8 leaks, again with many leaks, but this time either 82 or 83. They are happening in opt-mochitest-browser-chrome-e10s-11, so different than the set of secondary leaks from 4/22. These leaks are happening in the remote/test/browser/ directory.

Comment 8 (from 4/25): The original WPT leak has disappeared entirely. It looks like the automated WPT import in bug 1544415 finally managed to hit this leak, because there a change in the whitelisting for this directory, which I'm going to assume made this leak not show up any more. The remaining leaks are now all in various browser-chrome suites, but they all seem to be in remote/test/browser/ again.

The original intermittent failure here is no longer happening (because it has been covered up). I'll file a new bug for the remote/test/browser/ issue.

Status: NEW → RESOLVED
Closed: 5 years ago
Depends on: 1544415
Resolution: --- → WORKSFORME
Summary: Intermittent LeakSanitizer | leak at createTable, mozilla::detail::HashTable, reserve, reserve → Intermittent LeakSanitizer | leak at createTable, mozilla::detail::HashTable, reserve, reserve in web-platform-tests

These remote/test/browser failures are so large and frequent there's presumably a corresponding spike in leaks somewhere for debug builds.

See Also: → 1547303

I filed bug 1547303 for the other leak.

I've said this before, but we really need some kind of "large leak mode", where the test harness prints out a different message when you leak a ton of stuff, because otherwise you end up with things like this, where a frequent leak gets a signature of whatever object out of hundreds that are leaking happens to occur near the start of the alphabet. I guess we need it for LSan as well as leakcheck.

You need to log in before you can comment on or make changes to this bug.