Closed Bug 585438 Opened 14 years ago Closed 14 years ago

[SeaMonkey 2.1, mochitest-plain-4, Linux] test_invalidate_during_plugin_paint.html crashing

Categories

(SeaMonkey :: Testing Infrastructure, defect)

Product:
SeaMonkey
Component:
Testing Infrastructure
Platform:
x86
Linux
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

(Not tracked)

Status:
VERIFIED FIXED
Milestone:
seamonkey2.1b1

People

(Reporter: kairo, Unassigned)

References

(Blocks 1 open bug)

Details

(Keywords: intermittent-failure, Whiteboard: [fixed by bug 545716] [sm-perma][cc-orange]) 

test_invalidate_during_plugin_paint.html has been crashing (might be fatal assertion) for a while now on Linux (debug) boxes.

Example log:
http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1281270793.1281272251.1104.gz

Example output:
2715 INFO TEST-START | /tests/layout/generic/test/test_invalidate_during_plugin_paint.html
WARNING: NS_ENSURE_SUCCESS(rv, rv) failed with result 0x80004005: file /builds/slave/comm-central-trunk-linux-debug/build/mozilla/extensions/spellcheck/src/mozInlineSpellChecker.cpp, line 1476
++DOMWINDOW == 34 (0x99923d8) [serial = 1026] [outer = 0x9cce7d8]
For application/x-test found plugin libnptest.so
--DOMWINDOW == 33 (0xb6b9e58) [serial = 1004] [outer = 0x9cce7d8] [url = http://mochi.test:8888/tests/layout/generic/test/test_bug468167.html]
--DOMWINDOW == 32 (0xb0e2910) [serial = 1005] [outer = 0x9cce7d8] [url = http://mochi.test:8888/tests/layout/generic/test/test_bug469613.xul]
--DOMWINDOW == 31 (0xb12bc00) [serial = 1006] [outer = 0x9cce7d8] [url = http://mochi.test:8888/tests/layout/generic/test/test_bug469774.xul]
--DOMWINDOW == 30 (0xa6ebd18) [serial = 1002] [outer = 0x9cce7d8] [url = http://mochi.test:8888/tests/layout/generic/test/test_bug448860.html]
--DOMWINDOW == 29 (0xaacda78) [serial = 1024] [outer = 0x9cce7d8] [url = http://mochi.test:8888/tests/layout/generic/test/test_bug522632.html]
--DOMWINDOW == 28 (0xadad300) [serial = 1023] [outer = 0xb82e6a8] [url = about:blank]
--DOMWINDOW == 27 (0xb82e6e0) [serial = 1022] [outer = (nil)] [url = http://mochi.test:8888/tests/layout/generic/test/file_bug514732_1.html]
--DOCSHELL 0xac1e8b0 == 13
--DOMWINDOW == 26 (0xbdb8680) [serial = 1021] [outer = 0xaad31c0] [url = about:blank]
--DOMWINDOW == 25 (0xaad31f8) [serial = 1016] [outer = (nil)] [url = about:blank]
--DOMWINDOW == 24 (0xac292b8) [serial = 1020] [outer = 0xacda910] [url = about:blank]
--DOMWINDOW == 23 (0xacda948) [serial = 1015] [outer = (nil)] [url = about:blank]
--DOMWINDOW == 22 (0xab60dc8) [serial = 1019] [outer = 0xb26c348] [url = about:blank]
--DOMWINDOW == 21 (0xb26c380) [serial = 1014] [outer = (nil)] [url = about:blank]
--DOCSHELL 0xbafeee8 == 12
--DOCSHELL 0xaba7b30 == 11
--DOCSHELL 0xba85098 == 10
--DOMWINDOW == 20 (0xc714660) [serial = 1018] [outer = 0xb1e50e0] [url = about:blank]
--DOMWINDOW == 19 (0xb1e5118) [serial = 1017] [outer = (nil)] [url = http://mochi.test:8888/tests/layout/generic/test/file_bug514732_helper.html]
--DOCSHELL 0xb90be08 == 9
--DOCSHELL 0xc2e0460 == 8
--DOMWINDOW == 18 (0xc6d3020) [serial = 1011] [outer = 0x9cce7d8] [url = http://mochi.test:8888/tests/layout/generic/test/test_bug514732.html]
TEST-UNEXPECTED-FAIL | /tests/layout/generic/test/test_invalidate_during_plugin_paint.html | Exited with code -11 during test run
INFO | automation.py | Application ran for: 0:24:13.818668
INFO | automation.py | Reading PID log: /tmp/tmpmIzk6Opidlog
WARNING: nsExceptionService ignoring thread destruction after shutdown: file /builds/slave/comm-central-trunk-linux-debug/build/mozilla/xpcom/base/nsExceptionService.cpp, line 197


Last good build:
rev:89de54f97780 / moz:4dd7bbd2ebeb

First bad build:
rev:8b0bbac0f237 / moz:b141a304ad08

Between those all tests have been broken due to some packaging issue, probably in connection to layout. When we came back from the all-tests-breakage, this one remained. Might be connected to the test plugin in some way, as that's loaded before this happens, see http://mxr.mozilla.org/comm-central/source/mozilla/modules/plugin/test/testplugin/ for its code.
Can't reproduce this failure locally.

Regression ranges:
http://hg.mozilla.org/comm-central/pushloghtml?fromchange=89de54f97780&tochange=8b0bbac0f237
http://hg.mozilla.org/comm-central/pushloghtml?fromchange=4dd7bbd2ebeb
&tochange=b141a304ad08

Still, may be an issue with the machine setup on our slaves, need to find out about that.
Erm, the second range is http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=4dd7bbd2ebeb&tochange=b141a304ad08 of course.


I somewhat suspect a connection with bug 571172, which introduced ANGLE but roc also landed large layers stuff in bug 564991 in that range.
Whiteboard: [orange]
Version: unspecified → Trunk
Whiteboard: [orange] → [sm-perma][orange]
OK, so I've been informed that ANGLE is only used for webgl right now, which is not connected in any way with this test, so the prime suspect is retained layers.

roc, any idea what's going on there and how we could iron this out? We need to run our SeaMonkey tests on CentOS 5 on VMs right now, just to give you a perspective of what systems this is happening on.
Do you know why there's no stack?
(In reply to comment #4)
> Do you know why there's no stack?

No, I wondered myself why there was none, as usually we do have stacks when things crash in tests.
Is it worth sticking a dump command in the checkEnd function to see what PaintCount it is getting and maybe how many times it is looping into that?
Also one in doTest function to see what the initial PaintCount is?
This also happens on Firefox (bug 580483), probably should be duped against that bug.
Blocks: 580483
(Though that times out rather than crashes)
For them, there's an intermittent failure, and they clearly get some log output on the plugin doing something, while we don't get any log output but just a crash, and it's permanent. So this seems to be something different.
Seems to be fixed by turning on OOPP for SeaMonkey! (bug 545716)

It only ran once since then so far, but as it was really perma-orange, I call it success, and let's reopen when it's still a problem.

And Ian, regarding comment #10, the Firefox case is something else and should be filed as its own bug (though it's possible that we'll hit that one intermittently as well in the future).
Status: NEW → RESOLVED
Closed: 14 years ago
Resolution: --- → FIXED
(In reply to comment #11)

Ftr,

First green:
http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1285315832.1285320698.13046.gz&fulltext=1
Linux comm-central-trunk debug test mochitests-4/5 on 2010/09/24 01:10:32
{
2784 INFO TEST-START | /tests/layout/generic/test/test_invalidate_during_plugin_paint.html
++DOMWINDOW == 18 (0xd4379a0) [serial = 1034] [outer = 0x894f920]
For application/x-test found plugin libnptest.so
2785 INFO TEST-PASS | /tests/layout/generic/test/test_invalidate_during_plugin_paint.html | Got 38 paints
2786 INFO SimpleTest finished /tests/layout/generic/test/test_invalidate_during_plugin_paint.html in 2082ms
}

Next build:
http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1285328443.1285331848.8616.gz&fulltext=1
Linux comm-central-trunk debug test mochitests-4/5 on 2010/09/24 04:40:43
is orange: passed this test, but timed out and crashed in 2 next tests, fwiw.

V.Fixed, at least ftb.
Status: RESOLVED → VERIFIED
Depends on: SM-OOPP
Flags: in-testsuite-
Whiteboard: [sm-perma][orange] → [fixed by bug 545716] [sm-perma][orange]
Target Milestone: --- → seamonkey2.1b1
Whiteboard: [fixed by bug 545716] [sm-perma][orange] → [fixed by bug 545716] [sm-perma]
Whiteboard: [fixed by bug 545716] [sm-perma] → [fixed by bug 545716] [sm-perma][cc-orange]
You need to log in before you can comment on or make changes to this bug.