653994 - [Windows] crashtest: intermittent "99776-1.html | assertion count 1 is more than expected 0 assertions" (ASSERTION: Widgets that we paint must all be display roots: 'GetDisplayRootFor(aView) == aView')

Status: RESOLVED FIXED

(Core Graveyard :: Plug-ins, defect)

Description

[Serge Gautherie (:sgautherie)]

(First time I notice this failure.)

http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1304211780.1304214265.17429.gz&fulltext=1
WINNT 5.2 comm-central-trunk debug test crashtest on 2011/04/30 18:03:00
{
###!!! ASSERTION: Widgets that we paint must all be display roots: 'GetDisplayRootFor(aView) == aView', file e:/builds/slave/comm-cen-trunk-w32-dbg/build/mozilla/view/src/nsViewManager.cpp, line 445
...
REFTEST TEST-UNEXPECTED-FAIL | file:///e:/builds/slave/test/build/reftest/tests/layout/base/crashtests/99776-1.html | assertion count 1 is more than expected 0 assertions
}

Comment 1

[Serge Gautherie (:sgautherie)]

http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1332484394.1332486157.16634.gz
WINNT 5.2 comm-central-trunk debug test crashtest on 2012/03/22 23:33:14
s: cn-sea-qm-win2k3-01

Comment 2

[Serge Gautherie (:sgautherie)]

(In reply to Serge Gautherie (:sgautherie) from comment #1)
> http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1332484394.1332486157.
> 16634.gz
> WINNT 5.2 comm-central-trunk debug test crashtest on 2012/03/22 23:33:14
> s: cn-sea-qm-win2k3-01

I don't know if that's expected or unrelated,
but, fwiw, "For application/x-java-vm found plugin npjp2.dll" happened just before.

Comment 3

[Serge Gautherie (:sgautherie)]

http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1332447619.1332450077.20525.gz
WINNT 5.2 comm-central-trunk debug test crashtest on 2012/03/22 13:20:19
s: cn-sea-qm-win2k3-01

Same as previous comment.

Comment 4

[Serge Gautherie (:sgautherie)]

(In reply to Serge Gautherie (:sgautherie) from comment #2)
> I don't know if that's expected or unrelated,
> but, fwiw, "For application/x-java-vm found plugin npjp2.dll" happened just
> before.

Looks expected, per
{
7 <applet style="position:absolute;left:7;top:73;"></applet>
}

*****

http://tinderbox.mozilla.org/showlog.cgi?log=SeaMonkey/1333557762.1333559920.8703.gz&fulltext=1
WINNT 5.2 comm-central-trunk debug test crashtest on 2012/04/04 09:42:42
s: cn-sea-qm-win2k3-01

Comment 5

[Jesse Ruderman]

I reliably hit the same assertion on the same crashtest while trying to start the fuzzer. (Windows 7, Firefox trunk)

Based on when this bug was filed, it sounds like it's been happening (at least intermittently) ever since the assertion was added in bug 598854.

Comment 6

[Serge Gautherie (:sgautherie)]

[Mozilla/5.0 (Windows NT 5.0; rv:12.0a1) Gecko/20120126 Firefox/12.0a1 SeaMonkey/2.9a1] (custom debug build for Windows 2000)

I can reproduce by running this test alone locally and confirm the assertion is intermittent (yet frequent enough).

Ftr, I currently use "Java Plug-in 1.6.0_31".

Comment 7

[Robert O'Callahan (:roc) (email my personal email if necessary)]

I can reproduce this.

Comment 8

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Attachment: Avoid trying to paint plugin widgets in the case where a plugin fails to subclass our window

Comment 9

[Robert O'Callahan (:roc) (email my personal email if necessary)]

Comment on attachment 618551 [details] [diff] [review]
Avoid trying to paint plugin widgets in the case where a plugin fails to subclass our window

Apparently the Java plugin fails to subclass our HWND in some cases, letting NS_PAINT events go through to our WindowProc, which then tries to paint them, causing an assertion since we can't paint child widgets anymore. Instead, we just shouldn't try to paint plugin widgets in such a broken state.

Comment 10

[Robert O'Callahan (:roc) (email my personal email if necessary)]

http://hg.mozilla.org/integration/mozilla-inbound/rev/1863228248eb

Comment 11

[Ed Morley [:emorley]]

https://hg.mozilla.org/mozilla-central/rev/1863228248eb

Comment 12

[Serge Gautherie (:sgautherie)]

(In reply to Robert O'Callahan (:roc) (Mozilla Corporation) from comment #9)
> Apparently the Java plugin fails to subclass our HWND in some cases
> [...]
> we just shouldn't try to paint plugin widgets in such a broken state.

Is such a "misbehavior" case expected to (randomly) happen?
Or should a bug be filed at Oracle to investigate further and improve/fix their plugin?

Comment 13

[Robert O'Callahan (:roc) (email my personal email if necessary)]

I don't know. It doesn't really matter.

Comment 14

[Serge Gautherie (:sgautherie)]

Comment on attachment 618551 [details] [diff] [review]
Avoid trying to paint plugin widgets in the case where a plugin fails to subclass our window

[Approval Request Comment]
Detect and stop bad case early.
No risk.

Comment 15

[Alex Keybl [:akeybl]]

Comment on attachment 618551 [details] [diff] [review]
Avoid trying to paint plugin widgets in the case where a plugin fails to subclass our window

[Triage Comment]
No user impact provided, and a very intermittent orange. Approving for Aurora 14 but not Beta 13.

Comment 16

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/cfeda9529635