Closed Bug 542608 Opened 13 years ago Closed 13 years ago

Audio but no video on flash sites with Flashblock installed


(Core Graveyard :: Plug-ins, defect)

Not set


(blocking2.0 alpha1+, blocking1.9.2 -, status1.9.2 .2-fixed, status1.9.1 unaffected)

Tracking Status
blocking2.0 --- alpha1+
blocking1.9.2 --- -
status1.9.2 --- .2-fixed
status1.9.1 --- unaffected


(Reporter: jmjjeffery, Assigned: MatsPalmgren_bugz)



(Keywords: regression, verified1.9.2)


(1 file)

Visit YouTube or MSNBC and view a video.  There is audio, but no video. 

Reported here:

I saw at first the same shift in the video, or rather non-movement with scroll, then after turning off OOPP, and back on during testing I lost the video component.

regression shows window:

Works in cset:
01/25/10 09:19build
Broken in cset:

Leaving as the possible problem of losing the video.
Keywords: regression
bug 535295 maybe related?
Is this OOPP-only, or in-process plugins as well?
I can't reproduce this using Flash
(In reply to comment #2)
> Is this OOPP-only, or in-process plugins as well?

Does matter if OOPP is on or off, still no video, just audio.
Well, further testing reveals that the Flashblock addon is likely the problem.  Tested with new profile and was working OK.  Went back to my normal profile and video was not showing, only sound.  Even though the site was allowed through Flashblock, there was no video until I 'disabled' the addon. 

So, don't know if something broke the in-line function of Flashblock, or what.. not a programmer. 

If something thinks this is not a Minefield issue, but rather an update is required in Flashblock, feel free to close as INVALID.
This is looks like bug 542447, which got duped to bug 542447.

I'm not sure the duping was correct, so I'm waiting for bug 540604 to verify.
Gaah!!!  Gotta proofread.

This looks like bug 542447, which got duped to bug 542263.

I'm not sure the duping was correct, so I'm waiting for bug 540604 to verify.
Or maybe not.  Never mind.

Only one changeset:
Null-check the root pres context result. Unregister the plugin frame earlier, at Stop instead of Destroy time. b=521426 r=roc

Bug 521426 (Crash [@ PL_DHashTableEnumerate] called from nsRootPresContext::GetPluginGeometryUpdates because nsPresContext::RootPresContext() can return non-nsRootPresContext object)

With Flashblock actively blocking youtube, clicking on the flashblock placeholder allows the video to play normally. With youtube whitelisted the audio is heard but the video is not seen. The flashblock content XBL uses a DOM UIEvent to send a whitelist signal to chrome. The flashblock chrome code then checks the whitelist:

	checkLoadFlash : function (e) {
		if(!gFlashblockEnabled || Flashblock.checkWhitelist(

So somehow the time taken to check the whitelist interferes with the timing and the video does not paint.

If I use Stylish (or nsIStylesheetService) to change the style of the youtube embed
@-moz-document domain( {
  #movie_player {
    display: block;
The video reappears. It doesn't matter what I change the display: to I just need to toggle it and the flashplayer appears, the video  becomes visible and restarts from the beginning.  I would guess that toggling the display style causes a reflow and reinits the flashplayer.
Blocks: 521426
This bug also causes youtube videos to scroll as the page is scrolled, instead of remaining fixed in place.
...but only if the site is not whitelisted in flashblock and a user presses the flashblock icon to start playback.
Which doesn't make any sense, since once the flash object is re-inserted into the DOM flashblock is out of the picture.
I think that flashblock is important enough this needs to block 1.9.3alpha1.
blocking2.0: --- → alpha1
Attached patch Patch rev. 1Splinter Review
We need to always register the frame when there's a widget (balancing the
condition for when we unregister), not just the first time when creating
the widget.  (Flashblock caused a re-instantiation of a stopped plugin using
the existing frame/widget.)
Attachment #424024 - Flags: review?(roc)
OS: Windows 7 → All
Hardware: x86 → All
(In reply to comment #15)
> Test builds:

All seems good here..just tested on several sites, with and without flashblock, and all appears normal.
Why is this marked 1.9.2-unaffected? As far as I can tell it should affect 1.9.2.

Thanks Mats! I guess we should be able to have a test here.
blocking1.9.2: --- → ?
> Why is this marked 1.9.2-unaffected? 

Probably because bug 521426 hasn't landed on 1.9.2 yet.  This should be marked the same way that bug is, imo.
Yes, my plan was to merge this fix with the 1.9.2 patch on bug 521426
and asking for approval on that, but this works too.
Closed: 13 years ago
Flags: in-testsuite?
Resolution: --- → FIXED
Target Milestone: --- → mozilla1.9.3a1
Will wait another day or two to allow the regression fix to bake.
I can verify that this is fixed on m-c with the latest hourly:
Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.3a1pre) Gecko/20100128 Minefield/3.7a1pre ID:20100128145729


Users report issue with Flashblock is fixed in the latest hourly builds. Thanks guys for the quick response.
blocking1.9.2: ? → -
Comment on attachment 424024 [details] [diff] [review]
Patch rev. 1

Attachment #424024 - Flags: approval1.9.2.2? → approval1.9.2.2+
I can VERIFY FIXED with latest nightly on Win32 Vista.
Running here on XP, I cannot reproduce this issue using Flashblock If I whitelist youtube and watch videos there, they will play fine (just as they do if I don't whitelist but select the flashblock button on the video when it shows up). 

Have we actually seen this problem in Firefox 3.6 at all?
Al, this bug was a regression fix for bug 521426 which never landed on
1.9.2 by itself.  Both patches together on 1.9.2 (comment 25).
Thanks, Mats. Does this mean that there is no scenario to repro the bug on 1.9.2 since the original regression by itself?
Not with a nightly build, no.
Summary: Audio No Video on flash sites → Audio but no video on flash sites with Flashblock installed
I was able to reproduce this bug with an older Minefield build. A recent build on trunk doesn't show this problem. Even the candidate build 3 of Firefox 3.6.2 looks good. Marking as verified on 1.9.2 with Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv: Gecko/20100316 Firefox/3.6.2
Keywords: verified1.9.2
Flags: in-testsuite?
Product: Core → Core Graveyard
You need to log in before you can comment on or make changes to this bug.