1189709 - Silverlight fails to run on 20150730030208 Nightly

Status: VERIFIED FIXED

(Core :: IPC, defect)

Description

[Ekanan Ketunuti]

Url: 
http://www.dcs-digital.com/setweb/ondemandnew.php?onid=3082&&size=b
http://demos.telerik.com/silverlight/
http://bubblemark.com/silverlight2.html

The first bad revision is 

changeset:   255050:f290fbb44af6
user:        Aaron Klotz <[e-mail]>
date:        Tue Jul 21 01:21:51 2015 -0600
files:       ipc/glue/MessageChannel.cpp ipc/glue/MessageChannel.h ipc/glue/Neutering.h ipc/glue/WindowsMessageLoop.cpp ipc/glue/moz.build
description:
Bug 1185639 - Allow deferred message processing to happen between consecutive IPC message dispatches. r=jimm

Comment 1

[Jim Mathies [:jimm]]

confirming, I can reproduce.

Comment 2

[David [:Ceremony]]

Dupe of https://bugzilla.mozilla.org/show_bug.cgi?id=1187005

Regression introduced by https://bugzilla.mozilla.org/show_bug.cgi?id=1165981

Comment 3

[Jim Mathies [:jimm]]

(In reply to David H. from comment #2)
> Dupe of https://bugzilla.mozilla.org/show_bug.cgi?id=1187005
> 
> Regression introduced by https://bugzilla.mozilla.org/show_bug.cgi?id=1165981

That's a bug specific to 64-bit edition.

Comment 4

[Ekanan Ketunuti]

(In reply to David H. from comment #2)
> Dupe of https://bugzilla.mozilla.org/show_bug.cgi?id=1187005
>
> Regression introduced by https://bugzilla.mozilla.org/show_bug.cgi?id=1165981

Nothing related. This bug is NOT about "Silverlight is not recognized as installed".

Bug 1165981 landed on m-c on 2015-07-22 (Nightly 20150722030205) which is prior to Bug 1185639. Silverlight works fine on that build.

Comment 5

[Ekanan Ketunuti]

Merge day is coming up.

Comment 6

[Ekanan Ketunuti]

I can't reproduce it in Nightly 8/9/2015 and I found a fix range.

Last bad: mozilla-inbound revision 3d180b8dcf42 
http://ftp.mozilla.org/pub/mozilla.org/firefox/tinderbox-builds/mozilla-inbound-win32/1438932581/

First good: mozilla-inbound revision ce43b991306f 
http://ftp.mozilla.org/pub/mozilla.org/firefox/tinderbox-builds/mozilla-inbound-win32/1438933961/

Pushlog: https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=3d180b8dcf42&tochange=ce43b991306f

Fixed by Bob Owen, Bug 1185529.

Comment 7

[Ekanan Ketunuti]

(In reply to Ekanan Ketunuti from comment #6)
> I can't reproduce it in Nightly 8/9/2015 and I found a fix range.

Sorry, I meant Nightly 8/8/2015 (20150808030207) :(

Comment 8

[Bob Owen (:bobowen) - PTO until 3rd April]

I'm still getting frequent problems with SilverLight in Nightly on Amazon where I get a white tab, sometimes with sound in the background.
I don't get the issue on any other sites using SilverLight that I've tried.

I can't reproduce this if I back out the changeset for bug 1185639.

Comment 9

[(No longer employed by Mozilla) Aaron Klotz]

Looks like Silverlight is doing some fun stuff with window subclassing and the like during initialization. This is not currently compatible with window neutering which now happens sooner in IPC calls.

Comment 10

[Ekanan Ketunuti]

[Tracking Requested - why for this release]: regression


(In reply to Bob Owen (:bobowen) from comment #8)
> I'm still getting frequent problems with SilverLight in Nightly on Amazon
> where I get a white tab, sometimes with sound in the background.
> I don't get the issue on any other sites using SilverLight that I've tried.
> 

I've been playing around with this for a while and seeing this too.

STR:
1. Open any link in comment 0 and activate Silverlight.
2. Switch to the other tab.
3. Wait for a while (5 or 10 seconds maybe, I can't really tell.)
4. Switch back to tab opened in step 1.

AR:
Silverlight became white.

Comment 11

[(No longer employed by Mozilla) Aaron Klotz]

Attachment: Patch

The problem here was that we were neutering too aggressively in IPC Call()s. When the plugin tried to subclass the plugin window, it was inadvertently subclassing the NeuteredWindowProc instead.

Intercepting SetWindowLong* and playing around with that was too fragile, so I opted to reduce the scope of the NeuteredWindowRegion and add DeferredMessagePlaybackSuppressor to hold back deferred message playback until the end the call.

This fixes the problem with all three URLs described in the but report.

Comment 12

[(No longer employed by Mozilla) Aaron Klotz]

Comment on attachment 8653192 [details] [diff] [review]
Patch

I'm going to update this patch, so cancelling the review request for now.

Comment 13

[(No longer employed by Mozilla) Aaron Klotz]

Attachment: Patch (r2)

Comment 14

[Pulsebot]

https://hg.mozilla.org/integration/mozilla-inbound/rev/993384a9dc2e

Comment 15

[(No longer employed by Mozilla) Aaron Klotz]

Comment on attachment 8653593 [details] [diff] [review]
Patch (r2)

Approval Request Comment
[Feature/regressing bug #]: bug 1185639
[User impact if declined]: Silverlight won't start
[Describe test coverage new/current, TreeHerder]: Plugin test suite, manually tested with silverlight.
[Risks and why]: Low, this patch reduces the scope of the changes in the regressing bug.
[String/UUID change made/needed]: None

Comment 16

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/993384a9dc2e

Comment 17

[Sylvestre Ledru [:Sylvestre]]

Comment on attachment 8653593 [details] [diff] [review]
Patch (r2)

Sure, taking it for 42.

Comment 18

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-aurora/rev/97c700862ca4

Comment 19

[Ekanan Ketunuti]

verified fixed in nightly 8/31

Comment 20

[Ekanan Ketunuti]

verified fixed in aurora 9/01.

Comment 21

[(No longer employed by Mozilla) Aaron Klotz]

Comment on attachment 8653593 [details] [diff] [review]
Patch (r2)

Approval Request Comment
[Feature/regressing bug #]: bug 1185639
[User impact if declined]: Silverlight busted. Must be landed as a follow-up to bug 1185639.
[Describe test coverage new/current, TreeHerder]: Plugin test suite
[Risks and why]: Low, has been working well on aurora and beta
[String/UUID change made/needed]: None

Comment 22

[Ritu Kothari (:ritu) (Inactive, please n-i to RyanVM, jcristau, or pascal)]

Comment on attachment 8653593 [details] [diff] [review]
Patch (r2)

Approving for uplift to moz-release to be included in 41.0.1 build2. This patch is needed to fix the recent flash hangs (bug 1207766).

Comment 23

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/releases/mozilla-release/rev/989787953aca

Comment 24

[Bogdan Maris, Desktop QA]

During our Firefox 41.0.1 build 2 smoke testing we ran a number of plugin tests that included Silverlight across platforms (Windows 10 32bit, Windows 7 64bit, Mac OS X 10.10.5) and we can say that this issue is verified fixed.