1059175 - B2G-Inbound closed due to ICS Emulator Debug Mochitests Leaks in most of the tests

Status: RESOLVED FIXED

(Core :: General, defect)

Description

[Carsten Book [:Tomcat]]

Starting with https://tbpl.mozilla.org/?tree=B2g-Inbound&rev=c7b35bd70fc3 we see memory leaks in most tests in ICS Emulator Debug  like https://tbpl.mozilla.org/php/getParsedLog.php?id=46835326&tree=B2g-Inbound

23:44:47     INFO -  nsTraceRefcnt::DumpStatistics: 1051 entries
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 1 Composer2D (12 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 3 CondVar (72 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 1 ExpirationTrackerObserver (16 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 10 Mutex (200 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 1 PProcLoaderParent (416 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 1 PSharedBufferManagerChild (416 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 1 PSharedBufferManagerParent (416 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 2 PerThreadData (24 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 4 ReentrantMonitor (96 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 2 RefCountedMonitor (96 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 6 RefCountedTask (48 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 5 SocketConsumerBase (100 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 3 WeakReference<MessageListener> (24 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 3 ipc::MessageChannel (864 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 1 nsAStreamCopier (88 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 1 nsPipe (184 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 1 nsSocketTransport (352 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 1 nsSocketTransportService (168 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 139 nsStringBuffer (1112 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 29 nsTArray_base (116 bytes)
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 2 nsThread (280 bytes)
23:44:47  WARNING -  TEST-UNEXPECTED-FAIL | leakcheck | 5100 bytes leaked (Composer2D, CondVar, ExpirationTrackerObserver, Mutex, PProcLoaderParent, ...)

i reverted the inital csets where this leak happened (Bug 1047143) but this didn't helped, also talked with alive on irc and he thinks that more some gecko might be responsible for this problem. 

Can someone look into this leak?

Comment 1

[Ryan VanderMeulen [:RyanVM]]

We went from 137 to 139 leaking nsStringBuffer.

Before:
21:44:04     INFO -  TEST-INFO | leakcheck | leaked 137 nsStringBuffer (1096 bytes)

After:
23:44:47     INFO -  TEST-INFO | leakcheck | leaked 139 nsStringBuffer (1112 bytes)

Comment 2

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Yeah, it's hard to know what caused that.  r=me to bump the threshold and reopen.

Comment 3

[Ryan VanderMeulen [:RyanVM]]

Threshold bumped to 5116 bytes. I'll reopen the trees in a bit.
https://hg.mozilla.org/integration/b2g-inbound/rev/5413606a5a7e

Comment 4

[Ryan VanderMeulen [:RyanVM]]

Just waiting on some B2G desktop test results from a 12-cset Bumper Bot push that landed when I set b-i to Approval Only. They must have landed on Gaia either before it was closed or by people with push rights during the closure.

Comment 5

[Sam Foster [:sfoster] (he/him)]

> i reverted the inital csets where this leak happened (Bug 1047143) but this
> didn't helped, also talked with alive on irc and he thinks that more some
> gecko might be responsible for this problem. 

Do I understand from this that bug 1047143 may not be responsible for the leak? I'm happy to help figure this out and I can walk through that patch with someone to see if we can find the culprit, but the leak report itself is gobbledy-gook to me, I'll need an interpreter!

Comment 6

[Ryan VanderMeulen [:RyanVM]]

(In reply to Sam Foster [:sfoster] from comment #5)
> Do I understand from this that bug 1047143 may not be responsible for the
> leak?

Correct. Gaia's reopened, feel free to re-land whenever.

Comment 7

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/5413606a5a7e