Closed Bug 921173 Opened 11 years ago Closed 10 years ago

Intermittent browser_thumbnails_background_crash.js | application timed out after 330 seconds with no output

Categories

(Toolkit :: General, defect)

x86_64
macOS
defect
Not set
normal

Tracking

()

RESOLVED WORKSFORME

People

(Reporter: RyanVM, Unassigned)

Details

(Keywords: intermittent-failure)

https://tbpl.mozilla.org/php/getParsedLog.php?id=28415191&tree=Mozilla-Inbound

Rev4 MacOSX Snow Leopard 10.6 mozilla-inbound debug test mochitest-browser-chrome on 2013-09-26 08:54:10 PDT for push 3055b4a78245
slave: talos-r4-snow-079

10:29:18     INFO -  [Child 972] WARNING: NS_ENSURE_TRUE(compMgr) failed: file ../../../xpcom/glue/nsComponentManagerUtils.cpp, line 58
10:29:18     INFO -  nsStringStats
10:29:18     INFO -   => mAllocCount:          14268
10:29:18     INFO -   => mReallocCount:          738
10:29:18     INFO -   => mFreeCount:           14268
10:29:18     INFO -   => mShareCount:          22557
10:29:18     INFO -   => mAdoptCount:           1313
10:29:18     INFO -   => mAdoptFreeCount:       1313
10:29:27     INFO -  [Parent 882] WARNING: parent WaitForMessage() failed: 0x10004003 (ipc/rcv) timed out: file ../../../ipc/glue/GeckoChildProcessHost.cpp, line 670
10:29:27     INFO -  [Parent 882] WARNING: waitpid failed pid:906 errno:10: file ../../../ipc/chromium/src/base/process_util_posix.cc, line 251
10:29:35     INFO -  System JS : ERROR (null):0
10:29:35     INFO -                       uncaught exception: 2147746065
10:34:59  WARNING -  TEST-UNEXPECTED-FAIL | chrome://mochitests/content/browser/toolkit/components/thumbnails/test/browser_thumbnails_background_crash.js | application timed out after 330 seconds with no output
10:35:07     INFO -  SCREENSHOT: <see log>
10:35:07     INFO -  Can't trigger Breakpad, just killing process
10:35:07     INFO -  Failed to kill process 882: [Errno 3] No such process
10:35:07     INFO -  INFO | runtests.py | exit 9
10:35:07     INFO -  INFO | runtests.py | Application ran for: 1:38:06.448981
10:35:07     INFO -  INFO | zombiecheck | Reading PID log: /var/folders/H0/H0uJAwizEwqe43TH00w9xU+++-k/-Tmp-/tmpRIIqhtpidlog
10:35:08     INFO -  == BloatView: ALL (cumulative) LEAK AND BLOAT STATISTICS, tab process 972
10:35:08     INFO -       |<----------------Class--------------->|<-----Bytes------>|<----------------Objects---------------->|<--------------References-------------->|
10:35:08     INFO -                                                Per-Inst   Leaked    Total      Rem      Mean       StdDev     Total      Rem      Mean       StdDev
10:35:08     INFO -     0 TOTAL                                          24        0   269592        0 ( 2276.52 +/-  2214.06)   277986        0 ( 1222.10 +/-  2733.66)
10:35:08     INFO -  nsTraceRefcntImpl::DumpStatistics: 556 entries
10:35:08     INFO -  TEST-PASS | leakcheck | no leaks detected!
10:35:08     INFO -  ==> process 893 will purposefully crash
10:35:08     INFO -  TEST-INFO | leakcheck | plugin process: deliberate crash and thus no leak log
10:35:08     INFO -  ==> process 897 will purposefully crash
10:35:08     INFO -  TEST-INFO | leakcheck | plugin process: deliberate crash and thus no leak log
10:35:08     INFO -  ==> process 898 will purposefully crash
10:35:08     INFO -  TEST-INFO | leakcheck | plugin process: deliberate crash and thus no leak log
10:35:08     INFO -  runtests.py | Running tests: end.
10:37:06    ERROR - Return code: 2304
So...

* A "normal" (!abnormal) ipc:content-shutdown notification was broadcasted
  as soon as the test started.
* A bunch of DOM windows are closed immediately afterward.  Their URLs are
  of pages opened by previous tests.
* Immediately after that, it looks like the plugin-container process quits.
* The test doesn't get past the first capture.  The capture doesn't even
  time out.

Did the parent process shut down?
(In reply to Drew Willcoxon :adw from comment #1)
> So...
> 
> * A "normal" (!abnormal) ipc:content-shutdown notification was broadcasted
>   as soon as the test started.

Yeah, this is somewhat expected - the process used by the previous test can cause this.

> * A bunch of DOM windows are closed immediately afterward.  Their URLs are
>   of pages opened by previous tests.
> * Immediately after that, it looks like the plugin-container process quits.
> * The test doesn't get past the first capture.  The capture doesn't even
>   time out.
> 
> Did the parent process shut down?

Yes, it looks like it did.  TBH I don't think this will turn out to be directly related to this test but a more general e10s desktop problem :(
I also noticed in the #e10s channel that people are reporting the "System JS : ERROR (null):0...uncaught exception: 2147746065" - which we see in this report - are coming up in other contexts too, so I suspect some other issue is biting us here too.
Yes, that appears in all debug logs. Haven't had a chance to file it, though.
Whiteboard: [triage]
Whiteboard: [triage]
Whiteboard: p=0
Haven't seen this in a while!
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → WORKSFORME
No longer blocks: fxdesktopbacklog
Flags: firefox-backlog-
Whiteboard: p=0
You need to log in before you can comment on or make changes to this bug.