508767 - random failure (And crash) in test_384370.js

Status: RESOLVED FIXED

(Toolkit :: Places, defect)

Description

[Marco Bonardo [:mak]]

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1249526259.1249532124.12153.gz
WINNT 5.2 mozilla-central unit test on 2009/08/05 19:37:39

Comment 1

[Marco Bonardo [:mak]]

*** PLACES TESTS: validated import
TEST-INFO | (xpcshell/head.js) | test 1 finished
TEST-INFO | (xpcshell/head.js) | exiting test
Leaked URLs:
  file:///e:/builds/moz2_slave/mozilla-central-win32-unittest/build/objdir/dist/bin/
  file:///e:/builds/moz2_slave/mozilla-central-win32-unittest/build/objdir/dist/bin/
  chrome://global/content/viewPartialSource.xul
  chrome://browser/content/viewSourceOverlay.xul

... long list of urls...

  http://example.tld/
  http://example.tld/
  http://slint.us/
  http://slint.us/
  http://en.wikipedia.org/wiki/Diplodocus
  http://en.wikipedia.org/wiki/Diplodocus

  <<<<<<<
TEST-UNEXPECTED-FAIL | e:\builds\moz2_slave\mozilla-central-win32-unittest\build\objdir\_tests\xpcshell\test_browser_places\unit\test_384370.js | application crashed (minidump found)

Comment 2

[Marco Bonardo [:mak]]

I can't be sure but this sounds like a crash of the leaked urls dumper.
I'm not even sure if it should still be enabled on unit boxes, but if it is crashing on windows, could potentially explain some of the crashes we saw recently?

David, is it possible that the leaked urls debug code can crash?
can we try disable it on unit boxes and see if anything changes regarding number of windows boxes crashes?

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

The point of landing it in the first place was to get more data about random leaks on the unit test boxes.

It's certainly possible you're right, though; I'm not sure why it would have started crashing recently, though.

Comment 4

[Marco Bonardo [:mak]]

for the sake of tracking information, David tried disabling uri leaking stuff, and nothing changed, so that's not involved.

Comment 5

[Marco Bonardo [:mak]]

i think this was just another crash due to bug 507199. so i'm solving accordingly, if we see it again, we can reopen with a stacktrace.