Closed Bug 410330 Opened 17 years ago Closed 12 years ago

Very high memory and CPU usage. Browser becomes unresponsive with flash plugin images

Categories

(Core Graveyard :: Plug-ins, defect)

x86
Windows XP
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED WORKSFORME

People

(Reporter: securifirm, Unassigned)

References

()

Details

(Keywords: regression)

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1.11) Gecko/20071127 Firefox/2.0.0.11
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8.1.11) Gecko/20071127 Firefox/2.0.0.11

Just visit the URL and your memory and CPU usage will increase dramatically.  This happened at home and work using the same browser version on XP.

Reproducible: Always

Steps to Reproduce:
1. 
2.
3.
I see the same in today's latest trunk build, seems to be caused the the ads from s3.amazonaws.com , think we've seen this one before maybe. 

CPU 100%, and memory just keeps creeping up. 

Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9b3pre) Gecko/2007123105 Minefield/3.0b3pre Firefox/3.0 ID:2007123105
Keywords: qawanted
It looks like the Porsche ad (flash) at the top of the page is the cause.
IE6 died aswell when I looked at the page
Status: UNCONFIRMED → NEW
Ever confirmed: true
I did a search for s3.amazonaws.com on bugzilla, but couldn't find a relevant bug. Regression window is http://bonsai.mozilla.org/cvsquery.cgi?module=PhoenixTinderbox&date=explicit&mindate=1186088640&maxdate=1186089659
so caused by Bug 382378 or Bug 390318.
When I disable the flash plugin there is no problem.
Blocks: 382378
Component: General → Plug-ins
Keywords: regression
Product: Firefox → Core
QA Contact: general → plugins
Version: unspecified → Trunk
It doesn't look like the top banner is the one doing this because sometimes it's an image file and the CPU/Memory still creep up majorly.

I took a look into changing the HTML some to see what might be causing the issue and it looks like the Video player on the right hand side is the one thats causing the issue.  After removing this from the source, everything sped up again and didn't see the memory usage creeping up.
I am also encountering this bug on the Sun-Sentinel website in Windows Vista using Mozilla Firefox 3 Beta 2.  I was clicking the link to go to view the comments, but somewhere in all of the flash content on the website it really bogged things down and appeared to grow rapidly towards infinity until there was a Windows error.
not sure if this is related to the same bug, but browser locks up on this site (warning lots of flash), disabling the flash will load the site fine.
http://jewelrytelevision.typepad.com/jewelry_television/2007/04/index.html

Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9b3pre) Gecko/2008010314 Minefield/3.0b3pre
Flash 9.0 r115
I am getting suiterunner freezing
 (Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9b3pre) Gecko/2008011609 SeaMonkey/2.0a1pre ID:2008011609  Adblock Plus 0.7.5.3
ChatZilla 0.9.80
Console² 0.3.9
DOM Inspector 1.9b3pre
JavaScript Debugger 0.9.87.2
Nightly Tester Tools 1.3b4
SeaMonkey Debug and QA UI 1.0pre)
freezes when I try and go here
http://www.fantasybookspot.com/forum/search.php?search_id=newposts
and hit search

I haven't tried a new profile yet

It was working up until sometime last weekend/early this week.

I updated did a make clean - yes I build my own - as I thought stuff just might have got out of sync.

Just tried latest Minefield and it is not happy either - have added request to test the site to the daily build thread

Saw the same type of thing, extremely high CPU usage, on the following site:

www.wcax.com (news site-lot's of flash, slideshows but those are not the problem)

I confirmed that the scrolling info near the top of the page (yellow text in a thin read bar) was causing the problem, CPU @98%+. Hover over the scrolling the text (links in this case) stops the scrollin and CPU usage returns to 0%. I then found several other sites with similar scrolling text and saw the same amount of CPU usage. Other sites without scrolling text but with flash type ads (not all of them though), sometimes triggered the same type of usage.
I had similar problems when I had Flash Player 9.0 r115 installed.  My issues were resolved by reverting back to 9.0 r47.  Archived versions of Flash can be found at 
http://www.adobe.com/go/tn_14266 
9.0 r47 is in a 74 MB file on that site.
In light of kau's previous comment I realized that I failed to mention that this behavior is not exhibited when viewing the same sites in Internet Explorer 7.0. My apologizes for that oversight (I'm old[er]). Considering this, is reverting back to an older version of Flash Player the actual answer to the problem seen or is there a new factor/problem/feature in Flash player 9.0 r115 that may be triggering this in the nightly builds and not in IE? Since this behavior of high memory/CPU usage is not exhibited in IE my question would be why not? Unfortunately I'm not what you might call a "developer" since my skills are basic at best so I would have a very hard time in tracking the answer down.
I've had a chance to test the same website (www.wcax.com) in Firefox 2.0.0.12 (scroll bar appears only for weather alert hence the delay in mentioning this). No problem with high CPU usage seen. High CPU usage with same website still seen in latest nightly build:

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9b4pre) Gecko/2008020904 Minefield/3.0b4pre

Just for info's sake.
The page at the url field mentioned, doesn't seem to exist anymore.
Is there another site that shows this high memory and CPU usage?
Does http://jewelrytelevision.typepad.com/jewelry_television/2007/04/index.html show the bug (the high cpu usage _and_ the memory usage)?
Does http://www.fantasybookspot.com/forum/search.php?search_id=newposts show the bug?

Kirk, for the issue you mentioned in comment 8, I filed bug 420715, as I think that's a different issue than this one.
Yes this jewelry site does crank up my CPU here, but there are tons of embedded flash/ Youtube vids on that page, so folks with high end systems may not see as much impact.  
http://jewelrytelevision.typepad.com/jewelry_television/2007/04/index.html

Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9b4pre) Gecko/2008030305 Minefield/3.0b4pre Firefox/3.0 ID:2008030305 <-Latest nightly / Vista HP

system:
AMD 1.4ghz CPU Thunderbird 
1gig PC133 SDRAM (very old system, but Vista seems to run ok) 
ATI 9500Pro 128meg ram video card
(In reply to comment #12)
> The page at the url field mentioned, doesn't seem to exist anymore.
> Is there another site that shows this high memory and CPU usage?
> Does http://jewelrytelevision.typepad.com/jewelry_television/2007/04/index.html
> show the bug (the high cpu usage _and_ the memory usage)?
> Does http://www.fantasybookspot.com/forum/search.php?search_id=newposts show
> the bug?
> 
> Kirk, for the issue you mentioned in comment 8, I filed bug 420715, as I think
> that's a different issue than this one.
> 
Martijn, Thank you sir, I'm new to Bugzilla and I need time to learn the ropes. It's definitely script related and has been verified by several of my cohorts that it definitely looks like a regression.
hi

in ie6, and firefox2 works good, but
firefox3 beta 5 causes high CPU usage!

at http://www.harisnyadiszkont.hu
http://jewelrytelevision.typepad.com/jewelry_television/2007/04/index.html shows problem when flash images are allowed to load on modest core duo laptop


(In reply to comment #15)
> at http://www.harisnyadiszkont.hu

suspect problem is unrelated to this bug.  mouse over the scrolling adds on the right middle and you will see the scrolling stops and cpu drops - probably javascript loop (they are not flash images)

Summary: Very high memory and CPU usage. Browser becomes unresponsive → Very high memory and CPU usage. Browser becomes unresponsive with flash plugin images
Is the following this bug, or something else?

"High CPU usage with major menu lagging when playing embedded video on digg.com in Windows"

Steps to reproduce:
1.) Go to http://digg.com/political_opinion/Glenn_Beck_Federal_Reserve_Free_Money_and_AngryRenter_com
2.) Click on the image that says "play" to bring up the video on top of current page (make sure you aren't blocking any relevant scripts).
3.) If video doesn't start automatically, click play.
4.) Look at CPU usage in Task Manager and try browsing Firefox menus while video is playing
To mmortal03 and comment #17, that is not the same.  Don't confuse video playback with regular flash content.
Not sure if this is _just_ a Firefox problem, could be the Flash Plugin itself _but_it_is_worse_with_Firefox_than it is with IE7.
See below for rational.

Tests done with:
FireFox 3.0: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9) Gecko/2008052906 Firefox/3.0
Internet Explorer: 7.0.5730.11
Process Explorer: v11.20 (newest July 2008)

Test URL 1: http://www.ggmania.com
This game site page always has a _lot_ of flash embedded videos.  Note: no videos are playing, just the flash "cover" is displayed.
At the time of testing 7 flash objects were embedded.

Test URL 2: http://www.mozilla.org/
Fairly tame page. ;-)

Test URL 1 (with flash)
FireFox 3.0: Looking at this page causes my (CPU 2.8Hz P4) to run at 18-28%  (average %24) after loading the page, none of the flash videos have been activated.  Note: that usually this site has my CPU running at 60%, so this was a minimal CPU value for this bug.
IExplorer 7.0: The same machine has the IE process running at ~6-12% (average 8%).  Note: again this is a minimal value, the same page usually has IE7 running at 20%.

Test URL 2: (no flash)
FireFox 3.0: Looking at this page causes my (CPU 2.8Hz P4) to run at 0-2% after loading the page.
IExplorer 7.0: The same machine has the IE process running at 0%.

The same goes for other similar sites.

So obviously the flash plug-in is using some CPU even if the flash objects are "dormant" (or at least not doing anything visible).

However the CPU usage under Firefox 3 is significantly higher and fluctuates more than under IE7.



Some notes on the thread behavior:
Firefox threads that have the most context switches shown in Process Explorer are the main exe, mozcrt19.dll!endthread+0xa0 and interestingly enough WinMMdll!timeGetSystemTime+0x44.  Makes me wonder why Firefox needs to get the time continuously...
The flash.dll thread in both programs (FF3 & IE7) shows no context switches.

Handles, IO, INet, etc. show no change.  So it is just local, memory resident activity.
Flags: wanted1.9.1?
This really sounds like a flash problem to me, whether it's worse in Firefox than it is in IE or not. The flash plugins are not identical for the two browsers, so it's definitely possible that the performance characterstics of the two versions are somewhat different.

Has anyone reported this problem to Adobe?

Wanted1.9.1- unless this really is a problem on our end...
Flags: wanted1.9.1? → wanted1.9.1-
I have just reported this to Adobe.  I could not find anything in their bug database on this so I created a new report.

You could help by voting or adding information for the bug on their site: http://bugs.adobe.com/jira/browse/FP-417


Useful links:
Flash Version test: http://www.adobe.com/go/tn_15507
Flash Player Bug site: https://bugs.adobe.com/flashplayer/
Report for this bug: http://bugs.adobe.com/jira/browse/FP-417
I have the same kind of problem with Seamonkey 1.1.10, it does not happened on the same machine with Seamonkey 1.1.9.

Hi, I have a test case.
 The same problem occurs with the following setup:
Mozilla/5.0 (X11; U; Linux i686; es-ES; rv:1.9.0.1) Gecko/2008070206 Firefox/3.0.1
Shockwave Flash 9.0 r124
 I have tested the page:
http://www.br-online.de/br/jsp/seitentyp/liveStreamFenster.jsp?welle=bayern4klassik
 Press pause and then play to begin listening.
 Without touching the Firefox window, this is the CPU and memory resources consumed along the time (CPU measured with ps, mem with pmap)
Time (HH:MM)  %CPU    Memory
  ----        ----    -------
14:35          6%       195M
14:40         10%
14:53                   325M
14:58                   350M          
15:05         30%       380M
15:16         33%       420M
15:20                   500M
15:27         48%       500M
15:36                   560M
15:48                   560M
16:04         65%       790M
16:10         78%       790M
After reloading the page:
16:22          5%       195M

 Hope that it helps.
do you see this with current beta of 3.1 and flash v10?
 http://www.mozilla.com/en-US/firefox/all-beta.html
similar machine and results to #19
(In reply to comment #25)
> similar machine and results to #19

but what versions?
Old bug, the most of the links are no longer valid. Anyway, I cannot reproduce on Nightly 21.0a1 (2013-01-28) Win 7. Does anyone else still encounter this?
Keywords: qawanted
Tested this with my old link, which still works: http://gameguru.box.sk/

With my i7 ivy-bridge it is possible that the machine is so fast it is no longer noticeable.  Perhaps somebody with a slower machine can test it?

But it seems the bug is no longer present in Firefox 18.0.01 Windows 64bit with Flashplayer 11.5.502.  I do notice a high amount of RAM usage on sites, like the above, with a lot of flash video, but the CPU problem seems to have been fixed.
Seems Adobe threw my login out of their bug tracker, I don't really want to give them my data yet again.  Can somebody check the status of: http://bugs.adobe.com/jira/browse/FP-417   ?
(In reply to Duncan from comment #29)
> Can somebody check the status of:
> http://bugs.adobe.com/jira/browse/FP-417   ?

Resolved: 05/17/10 04:03 AM
Thanks XtC4UaLL

Well if somebody could test this link http://gameguru.box.sk/ with the flash movies on a lower end machine (e.g. AtSubnotebook, older P4 machine or similar) and give us results.  Then we can close it, if all is well.
I tested http://gameguru.box.sk/ on a AMD Athlon XP 2600+ 2GB RAM Win XP x86. After the page load is complete, the CPU usage looks normal ( < 5%).
Based on this, I'm closing the bug.
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → WORKSFORME
Forgot to mention, I tested on Firefox 18.0.1.
Product: Core → Core Graveyard
You need to log in before you can comment on or make changes to this bug.