561288 - [OOPP] Mouse events in most of Firefox UI hang with CPU intensive Flash

Status: RESOLVED INCOMPLETE

(Core Graveyard :: Plug-ins, defect)

Description

[Hugh Nougher [:Hughman]]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2.4) Gecko/20100413 Firefox/3.6.4 (.NET CLR 3.5.30729)
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.2.4) Gecko/20100413 Firefox/3.6.4 (.NET CLR 3.5.30729)

When I load this page with the flash game which uses the whole CPU I cannot get menus, tab changes or buttons to work in any of the Firefox UI. Sometimes it does work but takes a few seconds to happen.

On top of that the mouse events in the address bar for text editing/highlighting is working absolutely fine. The star does not work. The only real way I can get away from the page is to use Ctrl+Tab (the only key events that seem to work).

Using Latest Beta version of 3.6.4 with OOPP. I have tried brand new profile with no change.

Reproducible: Always

Steps to Reproduce:
1. Load the page that contains CPU intensive game at https://www.paypal-offers.com.au/birthday/.
2. Enter to main game menu.
3. Cant use menus or change tabs.


Expected Results:  
Using the rest of Firefox normally would be the expected result.

SYSTEMINFO.exe Copy/Paste

OS Name:                   Microsoft Windows XP Professional
OS Version:                5.1.2600 Service Pack 2 Build 2600
OS Manufacturer:           Microsoft Corporation
OS Configuration:          Member Workstation
OS Build Type:             Multiprocessor Free
System Manufacturer:       Dell Inc.
System Model:              OptiPlex GX280
System type:               X86-based PC
Processor(s):              2 Processor(s) Installed.
                           [01]: x86 Family 15 Model 4 Stepping 1 GenuineIntel ~3391 Mhz
                           [02]: x86 Family 15 Model 4 Stepping 1 GenuineIntel ~3391 Mhz
BIOS Version:              DELL   - 7
System Locale:             en-us;English (United States)
Input Locale:              en-us;English (United States)
Total Physical Memory:     1,014 MB
Available Physical Memory: 246 MB


I have also tried changing the plugin_container.exe in task manager to LOW priority like Bug #558555 with almost no change to problem.

Comment 1

[Hugh Nougher [:Hughman]]

I have found another more accessable example at
http://www.adidas.com/de/campaigns/outdoor/outdoor_relaunch/content/?strCountry_adidascom=en

On this page it happens where you are in the highlights and panning.
eg: Move mouse to far bottom right of flash display area, wait for the pan to finish in that direction then quickly move mouse over top-left cornor on way out to click on menu/tab. The tab will not change until the flash animation has at least stopped moving.

I have also tried a few other flash pages that use 3D and high CPU usage. Some worked fine and others dont. It seems to be more related to flash pages that are trying to refresh at the fastest possible rate since flash pages that have (guessing) regulated frame rates do not have a problem.

Comment 2

[(mostly gone) XtC4UaLL [:xtc4uall]]

Bug 560727, what is about Facebook Flash Games, may be related.
please also name the Flash version you're using.

Comment 3

[Hugh Nougher [:Hughman]]

I would have to agree that the facebook bug seems to be the same.
The flash version for that computer was 10.0.32.18 (I did try updating but Adobe Download Manager is useless).


I am now at home on a much more powerful computer (2.5GHz Core 2 Quad, 6GB DDR3, ATI HD5850 1GB, Windows 7) and it is completely repeatable except that its only about 1 second lag.
All following tests are done in a new profile in safe mode with the adidas ad as soon as it loaded up.

Mozilla/5.0 (Windows; U; Windows NT 6.1; en-GB; rv:1.9.2.3) Gecko/20100401 Firefox/3.6.3
RUN 1:
	Flash 10.0.42.34
	Very bad response.
RUN 2:
	Flash 10.0.45.2
	No Change

Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.2.4) Gecko/20100413 Firefox/3.6.4
RUN 1:
	Flash 10.0.42.34
	Very bad response.
RUN 2:
	Flash 10.0.45.2
	No Change

Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.3a5pre) Gecko/20100418 Minefield/3.7a5pre
RUN 1:
	Flash 10.0.42.34
	Very response but better than 3.6.x.
RUN 2:
	Flash 10.0.42.34
	gfx.font_rendering.directwrite.enabled=true
	mozilla.widget.render-mode=6
	(ie: Direct2D Enabled, although im not certain)
	This seems to have had the worst effect of any test I have done so far. Even the flash was slow. It was the most similar to the original result.
RUN 3:
	Flash 10.0.45.2
	No Change to RUN 1.
RUN 4:
	Flash 10.0.45.2
	gfx.font_rendering.directwrite.enabled=true
	mozilla.widget.render-mode=6
	No Change to RUN 2.


Also once the tab with the flash is in the background the UI is completely responsive as normal.

Comment 4

[Boris]

I am with Mozilla/5.0 (Windows; U; Windows NT 5.1; es-ES; rv:1.9.2.4) Gecko/20100413 Firefox/3.6.4 - Build ID: 20100413172113, and the latest flash(10.0.45.2) and the same happen.
If yout lost the focus of the browser, changing to (by example)explorer and go back to firefox, are yo get back a little of response of the firefox ui or flash buttons response?

Comment 5

[Hugh Nougher [:Hughman]]

(In reply to comment #4)
> If yout lost the focus of the browser, changing to (by example)explorer and go
> back to firefox, are yo get back a little of response of the firefox ui or
> flash buttons response?

I just did a test along these lines and I have to say there was a little more response from the UI after changing to another window and then changing back.

Another thing I noticed was when another window (even a firefox popup) was covering the flash element the UI of the firefox containing the flash was completely normal.

Also on top of that the CSS mouseover buttons at the top of that adidas page respond fine while the firefox UI does not. However if you bring up the menu (which darkens the whole page) and hide it again, it wont hide until the rest of the firefox UI updates (the other CSS menus are still responsive at this time though).

Comment 6

[Scoobidiver (away)]

Build : Mozilla/5.0 (Windows NT 6.1; WOW64; rv:2.0b7pre) Gecko/20100924 Firefox/4.0b7pre
FP : 10.1.82.76

I can reproduce the unresponsive browser on the URL of comment 1 that uses 80% of my Pentium T4300 processor CPU.
If I set dom.ipc.plugins.enabled to false and restart FF, the browser is slow but still responsive.

Comment 8

[Johnny Stenback (:jst)]

Jim, I'm guessing this is a dupe as well, but I'll leave it to you to dup appropriately. Either way, I don't think we want to ship with this bug.

Comment 10

[Jim Mathies [:jimm]]

I'm seeing about 30-40% on the adidas site for the container. That doesn't seem out of the ordinary for heavy animation.

I did notice the flash app doesn't load with ipc disabled. Probably need to file that.

Comment 11

[Jim Mathies [:jimm]]

(In reply to comment #3)
> I am now at home on a much more powerful computer (2.5GHz Core 2 Quad, 6GB
> DDR3, ATI HD5850 1GB, Windows 7) and it is completely repeatable except that
> its only about 1 second lag.
> All following tests are done in a new profile in safe mode with the adidas ad
> as soon as it loaded up.
> 
> Firefox/3.6.3
> RUN 1:
>     Flash 10.0.42.34
>     Very bad response.
> RUN 2:
>     Flash 10.0.45.2
>     No Change
> 
> Firefox/3.6.4
> RUN 1:
>     Flash 10.0.42.34
>     Very bad response.
> RUN 2:
>     Flash 10.0.45.2
>     No Change


So this would indicate this isn't related to OOPP, Flash is just dragging the browser down.


> 
> Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.3a5pre) Gecko/20100418
> Minefield/3.7a5pre
> RUN 1:
>     Flash 10.0.42.34
>     Very response but better than 3.6.x.
> RUN 2:
>     Flash 10.0.42.34
>     gfx.font_rendering.directwrite.enabled=true
>     mozilla.widget.render-mode=6
>     (ie: Direct2D Enabled, although im not certain)
>     This seems to have had the worst effect of any test I have done so far.
> Even the flash was slow. It was the most similar to the original result.
> RUN 3:
>     Flash 10.0.45.2
>     No Change to RUN 1.
> RUN 4:
>     Flash 10.0.45.2
>     gfx.font_rendering.directwrite.enabled=true
>     mozilla.widget.render-mode=6
>     No Change to RUN 2.

D2D issues are covered in other bugs, and we've recently starting blacklisting drivers that have issues. So potentially this might be fixed. Regardless we need to track down an appropriate D2D bug and keep this one about 3.6.4.

At this point I'm thinking we can close this out, unless someone has an OOPP specific flash slow down issue in 3.6.4 to report. 

Also, bug 583475 covers 4.0 issues revolving around retained layers.

Comment 12

[Jim Mathies [:jimm]]

bug 558339 is flash slowdown/D2D related.

Comment 13

[Jim Mathies [:jimm]]

The paypal site originally reported is no longer live. According to comment 3 this might be general slowness with flash running, which might be tied to any number of bugs. There's also some cross posting about 2D accel which is covered in other bugs. Unless we can hammer this out into a 3.6.3 -> 3.6.4 regression, this bug is invalid. Resolving incomplete.