Phil Ringnalda (:philor) Reporter
	Description • 15 years ago

+++ This bug was initially created as a clone of Bug #542928 +++

Cloned because the log looks exactly like bug 542928, other than starting after a different test, making me suspect that it's something broken in the way browser-test.js tries to restore focus - or in focus-manager, or in the way Mossop writes tests ;).

Running chrome://mochikit/content/browser/xpinstall/tests/browser_whitelist3.js...
Chrome file doesn't exist: /builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/mochitest/browser/xpinstall/tests/head.js
WARNING: Positioned frame that does not handle positioned kids; looking further up the parent chain: file /builds/moz2_slave/mozilla-central-linux-debug/build/layout/base/nsCSSFrameConstructor.cpp, line 5492
pldhash: for the table at address 0xb9e7340, the given entrySize of 48 probably favors chaining over double hashing.
++DOCSHELL 0xb9e72d8 == 16
++DOMWINDOW == 112 (0xb76bc48) [serial = 2536] [outer = (nil)]
WARNING: Positioned frame that does not handle positioned kids; looking further up the parent chain: file /builds/moz2_slave/mozilla-central-linux-debug/build/layout/base/nsCSSFrameConstructor.cpp, line 5492
WARNING: Positioned frame that does not handle positioned kids; looking further up the parent chain: file /builds/moz2_slave/mozilla-central-linux-debug/build/layout/base/nsCSSFrameConstructor.cpp, line 5492
++DOMWINDOW == 113 (0xbe23730) [serial = 2537] [outer = 0xb76bc18]
WARNING: NS_ENSURE_TRUE(aURI) failed: file /builds/moz2_slave/mozilla-central-linux-debug/build/docshell/base/nsDefaultURIFixup.cpp, line 76
pldhash: for the table at address 0xad9cd18, the given entrySize of 48 probably favors chaining over double hashing.
++DOCSHELL 0xad9ccb0 == 17
++DOMWINDOW == 114 (0xb9ef620) [serial = 2538] [outer = (nil)]
++DOMWINDOW == 115 (0xaeda038) [serial = 2539] [outer = 0xb9ef5f0]
WARNING: Positioned frame that does not handle positioned kids; looking further up the parent chain: file /builds/moz2_slave/mozilla-central-linux-debug/build/layout/base/nsCSSFrameConstructor.cpp, line 5492
WARNING: Positioned frame that does not handle positioned kids; looking further up the parent chain: file /builds/moz2_slave/mozilla-central-linux-debug/build/layout/base/nsCSSFrameConstructor.cpp, line 5492
###!!! ASSERTION: Uh, IsInModalState() called w/o a reachable top window?: 'Error', file /builds/moz2_slave/mozilla-central-linux-debug/build/dom/base/nsGlobalWindow.cpp, line 5904
gfxImageSurface::GetSize() const (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
gfxImageSurface::GetSize() const (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
gfxImageSurface::GetSize() const (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
gfxImageSurface::GetSize() const (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
NS_GetComponentManager_P (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
std::numeric_limits<long>::max() (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
std::vector<std::pair<int, int>, std::allocator<std::pair<int, int> > >::~vector() (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
MessageLoop::RunInternal() (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
MessageLoop::RunHandler() (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
MessageLoop::RunAllPending() (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
gfxPlatformGtk::GetPlatform() (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
gfxCornerSizes::gfxCornerSizes(gfxSize const&, gfxSize const&, gfxSize const&, gfxSize const&) (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
XRE_main (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/libxul.so)
UNKNOWN  (/builds/moz2_slave/mozilla-central-linux-debug-unittest-mochitest-other/build/firefox/firefox-bin)
__libc_start_main (/usr/src/debug/glibc-2.5-20061008T1257/csu/libc-start.c:262)
TEST-INFO | checking window state
TEST-INFO | (browser-test.js) | Waiting for window activation...
{domwindows and docshells trickle away}
NEXT ERROR TEST-UNEXPECTED-FAIL | automation.py | application timed out after 330 seconds with no output

	Phil Ringnalda (:philor) Reporter
	Comment 1 • 15 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1265696710.1265698661.8244.gz&fulltext=1#err2
Linux mozilla-central debug test mochitest-other on 2010/02/08 22:25:10
s: moz2-linux-slave20

	Phil Ringnalda (:philor) Reporter
	Comment 2 • 15 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1265737866.1265739956.16739.gz&fulltext=1
Linux mozilla-central debug test mochitest-other on 2010/02/09 09:51:06
s: moz2-linux-slave19

	Phil Ringnalda (:philor) Reporter
	Comment 3 • 15 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox3.6/1267276803.1267278738.10759.gz
Linux mozilla-1.9.2 opt test mochitest-other on 2010/02/27 05:20:03
s: moz2-linux-slave09

	Phil Ringnalda (:philor) Reporter
	Comment 4 • 15 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1267388139.1267389569.5114.gz
Linux mozilla-central opt test mochitest-other on 2010/02/28 12:15:39
s: moz2-linux-slave05

	Phil Ringnalda (:philor) Reporter
	Comment 5 • 15 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1267800425.1267803153.12001.gz&fulltext=1
Linux mozilla-central debug test mochitest-other on 2010/03/05 06:47:05
s: moz2-linux-slave19

	Daniel Holbert [:dholbert]
	Comment 6 • 15 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1270247707.1270249588.17306.gz
Linux mozilla-central debug test mochitest-other on 2010/04/02 15:35:07
s: moz2-linux-slave27

	Phil Ringnalda (:philor) Reporter
	Comment 7 • 14 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1272151634.1272153821.16014.gz
Linux mozilla-central debug test mochitest-other on 2010/04/24 16:27:14
s: moz2-linux-slave09

	Phil Ringnalda (:philor) Reporter
	Comment 8 • 14 years ago

http://tinderbox.mozilla.org/Firefox/1272315154.1272316851.30703.gz
Linux mozilla-central debug test mochitest-other on 2010/04/26 13:52:34
s: moz2-linux-slave18

	Karl Tomlinson (:karlt)
	Comment 9 • 14 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1272493151.1272495512.15178.gz
Linux mozilla-central debug test mochitest-other
s: moz2-linux-slave08

	Daniel Holbert [:dholbert]
	Comment 10 • 14 years ago

See also bug 565458, which looks like the same issue in a different test.

	(no longer active)
	Comment 11 • 14 years ago

Also similar: bug 565437

	(no longer active)
	Comment 12 • 14 years ago

This could fix the problem, because waitForFocus does a crazy number of checks to make sure that the window is focused correctly before running its callback.

This works fine locally.  I've pushed this to try as well.  I think we can take this patch because at least it shouldn't make anything worse.

	(no longer active)
	Comment 13 • 14 years ago

I landed the patch as http://hg.mozilla.org/mozilla-central/rev/0e11e454db6c.

I'll wait a few days to see if this happens again or not.  If it doesn't, I'll go ahead and mark all the related bugs as FIXED, and will ask people to reopen them if they see this problem again.

	(no longer active)
	Comment 14 • 14 years ago

This patch turned browser-chrome tests orange across the board.  I backed it out: 

http://hg.mozilla.org/mozilla-central/rev/859847fe8713

I don't have any idea what happened.  I tired to reproduce this on my own machine to no avail, but clearly there's something very wrong here, since the orange consistent in every test run on every platform.

Neil, could you please take a look to see if you see anything suspicious?  Here are the failure logs:

Linux:
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1273708610.1273709647.16121.gz
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1273708611.1273709753.16677.gz
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1273708027.1273710042.18308.gz
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1273708028.1273710670.21266.gz
Linux64:
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1273709393.1273711110.22947.gz
OS X:
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1273709753.1273710995.22540.gz
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1273708287.1273710261.19044.gz
Windows:
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1273709780.1273710817.21645.gz

	Neil Deakin
	Comment 15 • 14 years ago

Is it fixed by the patch in bug 554873?

	(no longer active)
	Comment 16 • 14 years ago

(In reply to comment #15)
> Is it fixed by the patch in bug 554873?

Dunno, we need to wait for that bug to land, and reland this one to see if it fixes that issue.  If that bug is not close to landing, I could just push a try server build with both patches.

	Marco Bonardo [:mak]
	Comment 17 • 14 years ago

IIRC, I initially used waitForFocus there, as you did, but Dao asked to use "activate" event since he noticed it was more reliable than waitForFocus.
Since a bunch of tests don't even change the initial tab, I guess that the latest patch in bug 554873 could help setting skipLoadCheck = true here.

	Marco Bonardo [:mak]
	Comment 18 • 14 years ago

PS: if you want a nice platform where to check focus issues, you should try latest Ubuntu 64, i was able to reproduce a bunch of failures there.

	Phil Ringnalda (:philor) Reporter
	Comment 19 • 14 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1273867335.1273869355.27474.gz
Rev3 Fedora 12 mozilla-central debug test mochitest-other on 2010/05/14 13:02:15
s: talos-r3-fed-033

	(no longer active)
	Comment 20 • 14 years ago

I pushed this patch on top of the one in bug 554873 to the try server to see if it works.  It works locally, but so did my previous attempt.

	(no longer active)
	Comment 21 • 14 years ago

(In reply to comment #20)
> I pushed this patch on top of the one in bug 554873 to the try server to see if
> it works.  It works locally, but so did my previous attempt.

So, with that patch applied, both attachment 445002 [details] [diff] [review] and attachment 445643 [details] [diff] [review] worked successfully on the try server.

	Marco Bonardo [:mak]
	Comment 22 • 14 years ago

why passing false as the third param? I guess we shold skip the check here, we just care that the window is focused.

	(no longer active)
	Comment 23 • 14 years ago

(In reply to comment #22)
> why passing false as the third param? I guess we shold skip the check here, we
> just care that the window is focused.

Hrm, OK, I was under the impression that passing false to that parameter would skip the load event check.  But I was wrong!  I think that it makes sense to pass true there, as a way of not doing stuff that we don't need, but anyways, this shows that the patch in that bug has fixed some problems for consumers of waitForFocus, even those which don't pass true as the third parameter.

	Neil Deakin
	Comment 24 • 14 years ago

The third argument means 'I expect the page loaded in the window to be about:blank', not 'skip loading the page'. The next iteration of the patch in bug 554873 will fix up issues with this and clarify the intent.

	(no longer active)
	Comment 25 • 14 years ago

OK, then, thanks for making this clear, Neil.  So, I think we're ready to land here once bug 554873 lands.

	Marco Bonardo [:mak]
	Comment 26 • 14 years ago

", window, false" are defaults, thus you can omit them (just using the first patch should be good).

	(no longer active)
	Comment 27 • 14 years ago

Comment on attachment 445643 [details] [diff] [review]
Possible fix on top of bug 554873

(In reply to comment #26)
> ", window, false" are defaults, thus you can omit them (just using the first
> patch should be good).

Yes.  Marking this patch obsolete to make things clear.

	(no longer active)
	Comment 28 • 14 years ago

Relanded: <http://hg.mozilla.org/mozilla-central/rev/45f4470a306e>.  Hopefully it sticks this time.

	(no longer active)
	Comment 29 • 14 years ago

I'll go ahead and mark the rest of the bugs like this one as FIXED, and will watch them in case someone reopens them, and we'll track any possible remaining problems from there.

	Robert O'Callahan (:roc) (email my personal email if necessary)
	Comment 30 • 14 years ago

Rev3 Fedora 12 mozilla-central opt test mochitest-other
http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1274417154.1274418227.25489.gz

	Phil Ringnalda (:philor) Reporter
	Comment 31 • 14 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1274467303.1274469721.14975.gz
Rev3 Fedora 12x64 mozilla-central debug test mochitest-other on 2010/05/21 11:41:43
s: talos-r3-fed64-030

	(no longer active)
	Comment 32 • 14 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1274472619.1274476596.12886.gz

	(no longer active)
	Comment 33 • 14 years ago

I'm out of ideas here.  We probably need someone who knows about the internals of the focus code and how it works on Linux to help here.

	Timothy Nikkel (:tnikkel)
	Comment 34 • 14 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1274558019.1274560322.11367.gz
Rev3 Fedora 12x64 mozilla-central debug test mochitest-other on 2010/05/22 12:53:39
s: talos-r3-fed64-038

	Phil Ringnalda (:philor) Reporter
	Comment 35 • 14 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1274577982.1274579946.22494.gz
Rev3 Fedora 12x64 mozilla-central debug test mochitest-other on 2010/05/22 18:26:22
s: talos-r3-fed64-012

	Dão Gottwald [:dao]
	Comment 36 • 14 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1274709130.1274710847.14137.gz

	Phil Ringnalda (:philor) Reporter
	Comment 37 • 14 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1275086413.1275090224.5797.gz
Rev3 Fedora 12x64 mozilla-central debug test mochitest-other on 2010/05/28 15:40:13
s: talos-r3-fed64-050

	Phil Ringnalda (:philor) Reporter
	Comment 38 • 14 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox/1275357988.1275360541.21747.gz
Rev3 Fedora 12 mozilla-central debug test mochitest-other on 2010/05/31 19:06:28
s: talos-r3-fed-019

	Karl Tomlinson (:karlt)
	Comment 40 • 14 years ago

Looks like this Linux bug was fixed in bug 567704.
Comment 39 is on Mac, and so seems different.
I would mark this as a dupe, but a patch landed here.

	(no longer active)
	Comment 41 • 14 years ago

(In reply to comment #40)
> Looks like this Linux bug was fixed in bug 567704.
> Comment 39 is on Mac, and so seems different.
> I would mark this as a dupe, but a patch landed here.

No, that bug is not relevant to this issue.

	Dave Townsend [:mossop]
	Comment 42 • 14 years ago

Seems to have gone away so there is no reason to keep this open.

	Phil Ringnalda (:philor) Reporter
	Comment 43 • 14 years ago

http://tinderbox.mozilla.org/showlog.cgi?log=Firefox3.6/1296506544.1296507640.12389.gz
Linux mozilla-1.9.2 test mochitest-other on 2011/01/31 12:42:24
s: moz2-linux-slave18

TEST-PASS | chrome://mochikit/content/browser/xpinstall/tests/browser_whitelist2.js | Install should have been triggered by the right uri
Running chrome://mochikit/content/browser/xpinstall/tests/browser_whitelist3.js...
Waiting for window activation...
TEST-UNEXPECTED-FAIL | automation.py | application timed out after 330 seconds with no output